Sync ollama docs from 12e04379 on 2026-06-12

Sync ollama docs from 421faa02 on 2026-05-12
Sync ollama docs from 9330bb91 on 2026-04-12
2026-06-12 04:00:06 +00:00 · 2026-05-12 04:00:09 +00:00 · 2026-04-12 04:00:07 +00:00
53 changed files with 1107 additions and 373 deletions
@@ -4,8 +4,8 @@ This is a mirror of the Ollama repository.

 **Synced from:** https://github.com/ollama/ollama.git
 **Branch:** main
-**Commit:** 8f45236d09332949aa91774dc9eb46caf2abbbc1
-**Sync Date:** 2026-03-12
+**Commit:** 12e04379cd5a9a2f150104a02c5761213f4be777
+**Sync Date:** 2026-06-12
 **Content:** Paths: docs

 ---
@@ -398,6 +398,7 @@ curl http://localhost:11434/api/generate -d '{
    "num_keep": 5,
    "seed": 42,
    "num_predict": 100,
+    "draft_num_predict": 4,
    "top_k": 20,
    "top_p": 0.9,
    "min_p": 0.0,
@@ -28,9 +28,9 @@ Errors are returned in the `application/json` format with the following structur
 If an error occurs mid-stream, the error will be returned as an object in the `application/x-ndjson` format with an `error` property. Since the response has already started, the status code of the response will not be changed.

 ```json
-{"model":"gemma3","created_at":"2025-10-26T17:21:21.196249Z","response":" Yes","done":false}
-{"model":"gemma3","created_at":"2025-10-26T17:21:21.207235Z","response":".","done":false}
-{"model":"gemma3","created_at":"2025-10-26T17:21:21.219166Z","response":"I","done":false}
-{"model":"gemma3","created_at":"2025-10-26T17:21:21.231094Z","response":"can","done":false}
+{"model":"gemma4","created_at":"2025-10-26T17:21:21.196249Z","response":" Yes","done":false}
+{"model":"gemma4","created_at":"2025-10-26T17:21:21.207235Z","response":".","done":false}
+{"model":"gemma4","created_at":"2025-10-26T17:21:21.219166Z","response":"I","done":false}
+{"model":"gemma4","created_at":"2025-10-26T17:21:21.231094Z","response":"can","done":false}
 {"error":"an error was encountered while running the model"}
 ```
@@ -28,7 +28,7 @@ Once Ollama is running, its API is automatically available and can be accessed v

 ```shell
 curl http://localhost:11434/api/generate -d '{
-  "model": "gemma3",
+  "model": "gemma4",
  "prompt": "Why is the sky blue?"
 }'
 ```
@@ -184,6 +184,7 @@ curl -X POST http://localhost:11434/v1/chat/completions \
 - [x] Reproducible outputs
 - [x] Vision
 - [x] Tools
+- [x] Reasoning/thinking control (for thinking models)
 - [ ] Logprobs

 #### Supported request fields
@@ -207,6 +208,9 @@ curl -X POST http://localhost:11434/v1/chat/completions \
 - [x] `top_p`
 - [x] `max_tokens`
 - [x] `tools`
+- [x] `reasoning_effort` (`"high"`, `"medium"`, `"low"`, `"none"`)
+- [x] `reasoning`
+  - [x] `effort` (`"high"`, `"medium"`, `"low"`, `"none"`)
 - [ ] `tool_choice`
 - [ ] `logit_bias`
 - [ ] `user`
@@ -5,13 +5,13 @@ title: Streaming
 Certain API endpoints stream responses by default, such as `/api/generate`. These responses are provided in the newline-delimited JSON format (i.e. the `application/x-ndjson` content type). For example:

 ```json
-{"model":"gemma3","created_at":"2025-10-26T17:15:24.097767Z","response":"That","done":false}
-{"model":"gemma3","created_at":"2025-10-26T17:15:24.109172Z","response":"'","done":false}
-{"model":"gemma3","created_at":"2025-10-26T17:15:24.121485Z","response":"s","done":false}
-{"model":"gemma3","created_at":"2025-10-26T17:15:24.132802Z","response":" a","done":false}
-{"model":"gemma3","created_at":"2025-10-26T17:15:24.143931Z","response":" fantastic","done":false}
-{"model":"gemma3","created_at":"2025-10-26T17:15:24.155176Z","response":" question","done":false}
-{"model":"gemma3","created_at":"2025-10-26T17:15:24.166576Z","response":"!","done":true, "done_reason": "stop"}
+{"model":"gemma4","created_at":"2025-10-26T17:15:24.097767Z","response":"That","done":false}
+{"model":"gemma4","created_at":"2025-10-26T17:15:24.109172Z","response":"'","done":false}
+{"model":"gemma4","created_at":"2025-10-26T17:15:24.121485Z","response":"s","done":false}
+{"model":"gemma4","created_at":"2025-10-26T17:15:24.132802Z","response":" a","done":false}
+{"model":"gemma4","created_at":"2025-10-26T17:15:24.143931Z","response":" fantastic","done":false}
+{"model":"gemma4","created_at":"2025-10-26T17:15:24.155176Z","response":" question","done":false}
+{"model":"gemma4","created_at":"2025-10-26T17:15:24.166576Z","response":"!","done":true, "done_reason": "stop"}
 ```

 ## Disabling streaming
@@ -19,7 +19,7 @@ Certain API endpoints stream responses by default, such as `/api/generate`. Thes
 Streaming can be disabled by providing `{"stream": false}` in the request body for any endpoint that support streaming. This will cause responses to be returned in the `application/json` format instead:

 ```json
-{"model":"gemma3","created_at":"2025-10-26T17:15:24.166576Z","response":"That's a fantastic question!","done":true}
+{"model":"gemma4","created_at":"2025-10-26T17:15:24.166576Z","response":"That's a fantastic question!","done":true}
 ```

 ## When to use streaming vs non-streaming
@@ -19,7 +19,7 @@ For endpoints that return usage metrics, the response body will include the usag

 ```json
 {
-  "model": "gemma3",
+  "model": "gemma4",
  "created_at": "2025-10-17T23:14:07.414671Z",
  "response": "Hello! How can I help you today?",
  "done": true,
@@ -2,6 +2,10 @@
 title: Structured Outputs
 ---

+<Note>
+  Ollama's Cloud currently does not support structured outputs.
+</Note>
+
 Structured outputs let you enforce a JSON schema on model responses so you can reliably extract structured data, describe images, or keep every reply consistent.

 ## Generating structured JSON
@@ -96,12 +100,11 @@ Provide a JSON schema to the `format` field.
    ```
  </Tab>
  <Tab title="JavaScript">
-    Serialize a Zod schema with `zodToJsonSchema()` and parse the structured response:
+    Serialize a Zod schema with `z.toJSONSchema()` and parse the structured response:

    ```javascript
    import ollama from 'ollama'
-    import { z } from 'zod'
-    import { zodToJsonSchema } from 'zod-to-json-schema'
+    import * as z from 'zod'

    const Country = z.object({
      name: z.string(),
@@ -112,7 +115,7 @@ Provide a JSON schema to the `format` field.
    const response = await ollama.chat({
      model: 'gpt-oss',
      messages: [{ role: 'user', content: 'Tell me about Canada.' }],
-      format: zodToJsonSchema(Country),
+      format: z.toJSONSchema(Country),
    })

    const country = Country.parse(JSON.parse(response.message.content))
@@ -173,7 +176,7 @@ class ImageDescription(BaseModel):
  text_content: Optional[str] = None

 response = chat(
-  model='gemma3',
+  model='gemma4',
  messages=[{
    'role': 'user',
    'content': 'Describe this photo and list the objects you detect.',
@@ -7,7 +7,7 @@ Vision models accept images alongside text so the model can describe, classify,
 ## Quick start

 ```shell
-ollama run gemma3 ./image.png whats in this image?
+ollama run gemma4 ./image.png whats in this image?
 ```


@@ -28,7 +28,7 @@ Provide an `images` array. SDKs accept file paths, URLs or raw bytes while the R
    curl -X POST http://localhost:11434/api/chat \
    -H "Content-Type: application/json" \
    -d '{
-        "model": "gemma3",
+        "model": "gemma4",
        "messages": [{
        "role": "user",
        "content": "What is in this image?",
@@ -52,7 +52,7 @@ Provide an `images` array. SDKs accept file paths, URLs or raw bytes while the R
    # img = Path(path).read_bytes()

    response = chat(
-      model='gemma3',
+      model='gemma4',
      messages=[
        {
          'role': 'user',
@@ -71,7 +71,7 @@ Provide an `images` array. SDKs accept file paths, URLs or raw bytes while the R

    const imagePath = '/absolute/path/to/image.jpg'
    const response = await ollama.chat({
-      model: 'gemma3',
+      model: 'gemma4',
      messages: [
        { role: 'user', content: 'What is in this image?', images: [imagePath] }
      ],
@@ -5,7 +5,7 @@ title: CLI Reference
 ### Run a model

 ```
-ollama run gemma3
+ollama run gemma4
 ```

 ### Launch integrations
@@ -21,6 +21,7 @@ Configure and launch external applications to use Ollama models. This provides a
 - **OpenCode** - Open-source coding assistant
 - **Claude Code** - Anthropic's agentic coding tool
 - **Codex** - OpenAI's coding assistant
+- **VS Code** - Microsoft's IDE with built-in AI chat
 - **Droid** - Factory's AI coding agent

 #### Examples
@@ -63,7 +64,7 @@ I'm a basic program that prints the famous "Hello, world!" message to the consol
 #### Multimodal models

 ```
-ollama run gemma3 "What's in this image? /Users/jmorgan/Desktop/smile.png"
+ollama run gemma4 "What's in this image? /Users/jmorgan/Desktop/smile.png"
 ```

 ### Generate embeddings
@@ -81,13 +82,13 @@ echo "Hello world" | ollama run nomic-embed-text
 ### Download a model

 ```
-ollama pull gemma3
+ollama pull gemma4
 ```

 ### Remove a model

 ```
-ollama rm gemma3
+ollama rm gemma4
 ```

 ### List models
@@ -113,7 +114,7 @@ ollama signout
 First, create a `Modelfile`

 ```
-FROM gemma3
+FROM gemma4
 SYSTEM """You are a happy cat."""
 ```

@@ -132,7 +133,7 @@ ollama ps
 ### Stop a running model

 ```
-ollama stop gemma3
+ollama stop gemma4
 ```

 ### Start Ollama
@@ -229,4 +229,26 @@ curl https://ollama.com/api/chat \

 ## Local only

-Ollama can run in local-only mode by [disabling Ollama's cloud](./faq#how-do-i-disable-ollama-cloud) features.
+Ollama can run in local-only mode by [disabling Ollama's cloud](./faq#how-do-i-disable-ollama-cloud) features.
+
+## Deprecations
+
+Ollama will occasionally deprecate and retire older cloud models as newer and better open-source models are released.
+Tools and applications relying on Ollama Cloud models may need to be updated to keep working. Impacted users will be
+notified in advance of model deprecation and retirement. Deprecations will be communicated through email and on the
+Ollama website.
+
+Ollama Cloud model retirement does not affect local models.
+
+### Upcoming deprecations
+
+| Retirement date | Model | Recommended alternative |
+| --- | --- | --- |
+| June 16, 2026 | `kimi-k2-thinking` | `kimi-k2.6` |
+| June 16, 2026 | `kimi-k2:1t` | `kimi-k2.6` |
+| June 16, 2026 | `minimax-m2`  | `minimax-m3` |
+| June 16, 2026 | `glm-4.6` | `glm-5.1` |
+| June 16, 2026 | `qwen3-next:80b` | `qwen3.5` |
+| June 16, 2026 | `qwen3-vl:235b` | `qwen3.5` |
+| June 16, 2026 | `qwen3-vl:235b-instruct` | `qwen3.5` |
+| June 16, 2026 | `cogito-2.1:671b` | `deepseek-v4-flash` |
@@ -37,5 +37,5 @@ ollama ps
 ```
 ```
 NAME             ID              SIZE      PROCESSOR    CONTEXT    UNTIL
-gemma3:latest    a2af6cc3eb7f    6.6 GB    100% GPU     65536      2 minutes from now
+gemma4:latest    c6eb396dbd59    9.6 GB    100% GPU     131072     2 minutes from now
 ```
@@ -3,9 +3,11 @@
 Install prerequisites:

 - [Go](https://go.dev/doc/install)
- C/C++ Compiler e.g. Clang on macOS, [TDM-GCC](https://github.com/jmeubank/tdm-gcc/releases/latest) (Windows amd64) or [llvm-mingw](https://github.com/mstorsjo/llvm-mingw) (Windows arm64), GCC/Clang on Linux.
+- [CMake](https://cmake.org/download/) 3.24 or newer
+- C/C++ compiler: Clang on macOS, Visual Studio 2022 C++ tools on Windows, or GCC/Clang on Linux
+- [Ninja](https://github.com/ninja-build/ninja/releases) in `PATH` is recommended, especially on Windows

-Then build and run Ollama from the root directory of the repository:
+For pure Go iteration against an existing native payload, run Ollama from the repository root:

 ```shell
 go run . serve
@@ -14,53 +16,73 @@ go run . serve
 > [!NOTE]
 > Ollama includes native code compiled with CGO.  From time to time these data structures can change and CGO can get out of sync resulting in unexpected crashes.  You can force a full build of the native code by running `go clean -cache` first. 

+## Native build model
+
+For a fresh checkout, or after changing native code, build from the repository root. On macOS arm64, this builds Metal inference. On all other platforms this builds CPU-only inference. It builds the Go binary at the repository root and installs the native runtime payload under `build/lib/ollama`.
+
+```shell
+cmake -B build .
+cmake --build build --parallel 8
+./ollama serve
+```
+
+To install into a standard prefix layout:
+
+```shell
+cmake --install build --prefix /path/to/install
+```
+
+On all platforms except macOS arm64, to build GPU backends select the backends explicitly:
+
+```shell
+cmake -B build . -DOLLAMA_LLAMA_BACKENDS="cuda_v13;vulkan"
+cmake --build build --parallel 8
+```
+
+Supported backend values are `cuda_v12`, `cuda_v13`, `rocm_v7_1`, `rocm_v7_2`, `vulkan`, `cuda_jetpack5`, and `cuda_jetpack6`.
+
+Use standard CMake architecture overrides to narrow GPU builds for local hardware:
+
+```shell
+# CUDA
+cmake -B build . -DOLLAMA_LLAMA_BACKENDS=cuda_v13 -DCMAKE_CUDA_ARCHITECTURES=native
+
+# ROCm / HIP
+cmake -B build . -DOLLAMA_LLAMA_BACKENDS=rocm_v7_2 -DCMAKE_HIP_ARCHITECTURES=gfx1100
+```
+
+You can tune GGML build options by setting `GGML_*` values during configure. For example, to build CUDA v12 for Pascal without flash attention kernels:
+
+```shell
+cmake -B build . -DOLLAMA_LLAMA_BACKENDS=cuda_v12 -DCMAKE_CUDA_ARCHITECTURES=61 -DGGML_CUDA_FA=OFF
+```

 ## macOS (Apple Silicon)

-macOS Apple Silicon supports Metal which is built-in to the Ollama binary. No additional steps are required.
+Additional prerequisites:

-## macOS (Intel)
-
-Install prerequisites:
-
- [CMake](https://cmake.org/download/) or `brew install cmake`
-
-Then, configure and build the project:
+MLX Metal requires the Metal toolchain. Install [Xcode](https://developer.apple.com/xcode/) first, then:

 ```shell
-cmake -B build
-cmake --build build
-```
-
-Lastly, run Ollama:
-
-```shell
-go run . serve
+xcodebuild -downloadComponent MetalToolchain
 ```

 ## Windows

-Install prerequisites:
+Additional prerequisites:

- [CMake](https://cmake.org/download/)
 - [Visual Studio 2022](https://visualstudio.microsoft.com/downloads/) including the Native Desktop Workload
 - (Optional) AMD GPU support
    - [ROCm](https://rocm.docs.amd.com/en/latest/)
-    - [Ninja](https://github.com/ninja-build/ninja/releases)
 - (Optional) NVIDIA GPU support
-    - [CUDA SDK](https://developer.nvidia.com/cuda-downloads?target_os=Windows&target_arch=x86_64&target_version=11&target_type=exe_network)
- (Optional) VULKAN GPU support
-    - [VULKAN SDK](https://vulkan.lunarg.com/sdk/home) - useful for AMD/Intel GPUs
+    - [CUDA SDK](https://developer.nvidia.com/cuda-downloads?target_os=Windows&target_arch=x86_64&target_type=exe_network)
+- (Optional) Vulkan GPU support
+    - [Vulkan SDK](https://vulkan.lunarg.com/sdk/home) - useful for AMD/Intel GPUs
 - (Optional) MLX engine support
    - [CUDA 13+ SDK](https://developer.nvidia.com/cuda-downloads)
    - [cuDNN 9+](https://developer.nvidia.com/cudnn)

-Then, configure and build the project:
-
-```shell
-cmake -B build
-cmake --build build --config Release
-```
+For Ninja builds, run CMake from a Developer PowerShell/Command Prompt or another shell where the Visual Studio compiler is available.

 > Building for Vulkan requires VULKAN_SDK environment variable:
 > 
@@ -73,36 +95,20 @@ cmake --build build --config Release
 > set VULKAN_SDK=C:\VulkanSDK\<version>
 > ```

-> [!IMPORTANT]
-> Building for ROCm requires additional flags:
-> ```
-> cmake -B build -G Ninja -DCMAKE_C_COMPILER=clang -DCMAKE_CXX_COMPILER=clang++
-> cmake --build build --config Release
-> ```
-
-
-
-Lastly, run Ollama:
-
-```shell
-go run . serve
-```
-
 ## Windows (ARM)

-Windows ARM does not support additional acceleration libraries at this time.  Do not use cmake, simply `go run` or `go build`.
+Windows ARM does not support additional acceleration libraries at this time.

 ## Linux

-Install prerequisites:
+Additional prerequisites:

- [CMake](https://cmake.org/download/) or `sudo apt install cmake` or `sudo dnf install cmake`
 - (Optional) AMD GPU support
    - [ROCm](https://rocm.docs.amd.com/projects/install-on-linux/en/latest/install/quick-start.html)
 - (Optional) NVIDIA GPU support
    - [CUDA SDK](https://developer.nvidia.com/cuda-downloads)
- (Optional) VULKAN GPU support
-    - [VULKAN SDK](https://vulkan.lunarg.com/sdk/home) - useful for AMD/Intel GPUs
+- (Optional) Vulkan GPU support
+    - [Vulkan SDK](https://vulkan.lunarg.com/sdk/home) - useful for AMD/Intel GPUs
    - Or install via package manager: `sudo apt install vulkan-sdk` (Ubuntu/Debian) or `sudo dnf install vulkan-sdk` (Fedora/CentOS)
 - (Optional) MLX engine support
    - [CUDA 13+ SDK](https://developer.nvidia.com/cuda-downloads)
@@ -111,57 +117,17 @@ Install prerequisites:
 > [!IMPORTANT]
 > Ensure prerequisites are in `PATH` before running CMake.

-
-Then, configure and build the project:
-
-```shell
-cmake -B build
-cmake --build build
-```
-
-Lastly, run Ollama:
-
-```shell
-go run . serve
-```
-
 ## MLX Engine (Optional)

-The MLX engine enables running safetensor based models. It requires building the [MLX](https://github.com/ml-explore/mlx) and [MLX-C](https://github.com/ml-explore/mlx-c) shared libraries separately via CMake.  On MacOS, MLX leverages the Metal library to run on the GPU, and on Windows and Linux, runs on NVIDIA GPUs via CUDA v13.
+The MLX engine enables running safetensor based models. On macOS arm64, MLX is enabled by default. On other platforms, MLX backends are selected with `OLLAMA_MLX_BACKENDS`.

-### macOS (Apple Silicon)
-
-Requires the Metal toolchain. Install [Xcode](https://developer.apple.com/xcode/) first, then:
-
-```shell
-xcodebuild -downloadComponent MetalToolchain
-```
-
-Verify it's installed correctly (should print "no input files"):
-
-```shell
-xcrun metal
-```
-
-Then build:
-
-```shell
-cmake -B build --preset MLX
-cmake --build build --preset MLX --parallel
-cmake --install build --component MLX
-```
-
-> [!NOTE]
-> Without the Metal toolchain, cmake will silently complete with Metal disabled. Check the cmake output for `Setting MLX_BUILD_METAL=OFF` which indicates the toolchain is missing.
-
-### Windows / Linux (CUDA)
+### CUDA

 Requires CUDA 13+ and [cuDNN](https://developer.nvidia.com/cudnn) 9+.

 ```shell
-cmake -B build --preset "MLX CUDA 13"
-cmake --build build --target mlx --target mlxc --config Release --parallel
-cmake --install build --component MLX --strip
+cmake -B build . -DOLLAMA_MLX_BACKENDS=cuda_v13
+cmake --build build --parallel 8
 ```

 ### Local MLX source overrides
@@ -173,17 +139,20 @@ export OLLAMA_MLX_SOURCE=/path/to/mlx
 export OLLAMA_MLX_C_SOURCE=/path/to/mlx-c
 ```

-For example, using the helper scripts with local mlx and mlx-c repos:
-```shell
-OLLAMA_MLX_SOURCE=../mlx OLLAMA_MLX_C_SOURCE=../mlx-c ./scripts/build_linux.sh
+On macOS arm64:

-OLLAMA_MLX_SOURCE=../mlx OLLAMA_MLX_C_SOURCE=../mlx-c ./scripts/build_darwin.sh
+```shell
+OLLAMA_MLX_SOURCE=../mlx OLLAMA_MLX_C_SOURCE=../mlx-c cmake -B build .
+cmake --build build --parallel 8
 ```

+For CUDA:
+
 ```powershell
 $env:OLLAMA_MLX_SOURCE="../mlx"
 $env:OLLAMA_MLX_C_SOURCE="../mlx-c"
-./scripts/build_darwin.ps1
+cmake -B build . -DOLLAMA_MLX_BACKENDS=cuda_v13
+cmake --build build --parallel 8
 ```

 ## Docker
@@ -206,42 +175,13 @@ To run tests, use `go test`:
 go test ./...
 ```

-> NOTE: In rare circumstances, you may need to change a package using the new
-> "synctest" package in go1.24.
->
-> If you do not have the "synctest" package enabled, you will not see build or
-> test failures resulting from your change(s), if any, locally, but CI will
-> break.
->
-> If you see failures in CI, you can either keep pushing changes to see if the
-> CI build passes, or you can enable the "synctest" package locally to see the
-> failures before pushing.
->
-> To enable the "synctest" package for testing, run the following command:
->
-> ```shell
-> GOEXPERIMENT=synctest go test ./...
-> ```
->
-> If you wish to enable synctest for all go commands, you can set the
-> `GOEXPERIMENT` environment variable in your shell profile or by using:
->
-> ```shell
-> go env -w GOEXPERIMENT=synctest
-> ```
->
-> Which will enable the "synctest" package for all go commands without needing
-> to set it for all shell sessions.
->
-> The synctest package is not required for production builds.
-
 ## Library detection

-Ollama looks for acceleration libraries in the following paths relative to the `ollama` executable:
+Ollama looks for native helper binaries and acceleration libraries in installed and local development layouts:

-* `./lib/ollama` (Windows)
-* `../lib/ollama` (Linux)
-* `.` (macOS)
-* `build/lib/ollama` (for development)
+* `../lib/ollama` for standard installs where `ollama` is under `bin/`
+* `./lib/ollama` for Windows release-style payloads and local dist output
+* `.` for macOS release artifacts that colocate helpers with `ollama`
+* `build/lib/ollama` and `dist/<platform>/lib/ollama` for local development builds

 If the libraries are not found, Ollama will not run with any acceleration libraries.
@@ -70,12 +70,16 @@ docker run -d --device /dev/kfd --device /dev/dri -v ollama:/root/.ollama -p 114

 ## Vulkan Support

-Vulkan is bundled into the `ollama/ollama` image.  
+Vulkan is bundled into the `ollama/ollama` image and is enabled by default when
+the container can access the GPU devices.

 ```shell
-docker run -d --device /dev/kfd --device /dev/dri -v ollama:/root/.ollama -p 11434:11434 -e OLLAMA_VULKAN=1 --name ollama ollama/ollama
+docker run -d --device /dev/kfd --device /dev/dri -v ollama:/root/.ollama -p 11434:11434 --name ollama ollama/ollama
 ```

+Use `OLLAMA_VULKAN=0` to disable Vulkan, or `GGML_VK_VISIBLE_DEVICES=<ids>` to
+select specific Vulkan devices.
+

 ## Run model locally

@@ -88,4 +92,3 @@ docker exec -it ollama ollama run llama3.2
 ## Try different models

 More models can be found on the [Ollama library](https://ollama.com/library).
-
@@ -75,6 +75,10 @@
    {
      "source": "/integrations/clawdbot",
      "destination": "/integrations/openclaw"
+    },
+    {
+      "source": "/integrations/poolside",
+      "destination": "/integrations/pool"
    }
  ],
  "navigation": {
@@ -110,7 +114,9 @@
                "group": "Assistants",
                "expanded": true,
                "pages": [
-                  "/integrations/openclaw"
+                  "/integrations/openclaw",
+                  "/integrations/hermes",
+                  "/integrations/hermes-desktop"
                ]
              },
              {
@@ -118,15 +124,21 @@
                "expanded": true,
                "pages": [
                  "/integrations/claude-code",
+                  "/integrations/codex-app",
                  "/integrations/codex",
+                  "/integrations/copilot-cli",
+                  "/integrations/cline-cli",
                  "/integrations/opencode",
                  "/integrations/droid",
                  "/integrations/goose",
-                  "/integrations/pi"
+                  "/integrations/oh-my-pi",
+                  "/integrations/pi",
+                  "/integrations/pool"
                ]
              },
              {
                "group": "IDEs & Editors",
+                "expanded": true,
                "pages": [
                  "/integrations/cline",
                  "/integrations/jetbrains",
@@ -160,6 +172,12 @@
            "group": "More information",
            "pages": [
              "/cli",
+              {
+                "group": "Assistant Sandboxing",
+                "pages": [
+                  "/integrations/nemoclaw"
+                ]
+              },
              "/modelfile",
              "/context-length",
              "/linux",
@@ -4,6 +4,7 @@ title: Hardware support

 ## Nvidia
 Ollama supports Nvidia GPUs with compute capability 5.0+ and driver version 531 and newer.
+Nvidia GPUs with compute capability 5.0 through 6.2 require driver version 570 or newer.

 Check your compute compatibility to see if your card is supported:
 [https://developer.nvidia.com/cuda-gpus](https://developer.nvidia.com/cuda-gpus)
@@ -61,6 +62,10 @@ Ollama supports the following AMD GPUs via the ROCm library:

 ### Linux Support

+Ollama requires the AMD ROCm v7 driver on Linux. You can install or upgrade
+using the `amdgpu-install` utility from
+[AMD's ROCm documentation](https://rocm.docs.amd.com/projects/install-on-linux/en/latest/).
+
 | Family         | Cards and accelerators                                                                                                                                         |
 | -------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------- |
 | AMD Radeon RX  | `9070 XT` `9070 GRE` `9070` `9060 XT` `9060 XT LP` `9060` `7900 XTX` `7900 XT` `7900 GRE` `7800 XT` `7700 XT` `7700` `7600 XT` `7600` `6950 XT` `6900 XTX` `6900XT` `6800 XT` `6800` `5700 XT` `5700` `5600 XT` `5500 XT` |
@@ -71,7 +76,7 @@ Ollama supports the following AMD GPUs via the ROCm library:

 ### Windows Support

-With ROCm v6.1, the following GPUs are supported on Windows.
+Ollama requires an AMD ROCm v7 / HIP7-capable driver stack on Windows.

 | Family         | Cards and accelerators                                                                                               |
 | -------------- | -------------------------------------------------------------------------------------------------------------------- |
@@ -138,12 +143,9 @@ Ollama supports GPU acceleration on Apple devices via the Metal API.

 ## Vulkan GPU Support

-> **NOTE:**
-> Vulkan is currently an Experimental feature.  To enable, you must set OLLAMA_VULKAN=1 for the Ollama server as
-described in the [FAQ](faq#how-do-i-configure-ollama-server)
-
 Additional GPU support on Windows and Linux is provided via
-[Vulkan](https://www.vulkan.org/). On Windows most GPU vendors drivers come
+[Vulkan](https://www.vulkan.org/). Vulkan is enabled by default when the
+backend is installed. On Windows most GPU vendors drivers come
 bundled with Vulkan support and require no additional setup steps. Most Linux
 distributions require installing additional components, and you may have
 multiple options for Vulkan drivers between Mesa and GPU Vendor specific packages
@@ -169,4 +171,9 @@ To select specific Vulkan GPU(s), you can set the environment variable
 `GGML_VK_VISIBLE_DEVICES` to one or more numeric IDs on the Ollama server as
 described in the [FAQ](faq#how-do-i-configure-ollama-server). If you
 encounter any problems with Vulkan based GPUs, you can disable all Vulkan GPUs
-by setting `GGML_VK_VISIBLE_DEVICES=-1` 
+by setting `OLLAMA_VULKAN=0` or `GGML_VK_VISIBLE_DEVICES=-1`.
+
+On mixed iGPU/dGPU systems where the Vulkan iGPU is unstable, keep Vulkan
+enabled and set `GGML_VK_VISIBLE_DEVICES` to the discrete GPU index. For
+example, use `GGML_VK_VISIBLE_DEVICES=1` when `Vulkan1` is the discrete
+GPU.
@@ -5,7 +5,7 @@ sidebarTitle: Welcome

 <img src="/images/welcome.png" noZoom className="rounded-3xl" />

-[Ollama](https://ollama.com) is the easiest way to get up and running with large language models such as gpt-oss, Gemma 3, DeepSeek-R1, Qwen3 and more.
+[Ollama](https://ollama.com) is the easiest way to get up and running with large language models such as gpt-oss, Gemma 4, DeepSeek-R1, Qwen3 and more.

 <CardGroup cols={2}>
  <Card title="Quickstart" icon="rocket" href="/quickstart">
@@ -41,13 +41,27 @@ ollama launch claude --model kimi-k2.5:cloud

 - `kimi-k2.5:cloud`
 - `glm-5:cloud`
- `minimax-m2.5:cloud`
+- `minimax-m2.7:cloud`
 - `qwen3.5:cloud`
 - `glm-4.7-flash`
 - `qwen3.5`

 Cloud models are also available at [ollama.com/search?c=cloud](https://ollama.com/search?c=cloud).

+## Non-interactive (headless) mode
+
+Run Claude Code without interaction for use in Docker, CI/CD, or scripts:
+
+```shell
+ollama launch claude --model kimi-k2.5:cloud --yes -- -p "how does this repository work?"
+```
+
+The `--yes` flag auto-pulls the model, skips selectors, and requires `--model` to be specified. Arguments after `--` are passed directly to Claude Code.
+
+## Web search
+
+Claude Code can search the web through Ollama's web search API. See the [web search documentation](/capabilities/web-search) for setup and usage.
+
 ## Scheduled Tasks with `/loop`

 The `/loop` command runs a prompt or slash command on a recurring schedule inside Claude Code. This is useful for automating repetitive tasks like checking PRs, running research, or setting reminders.
@@ -82,6 +96,18 @@ The `/loop` command runs a prompt or slash command on a recurring schedule insid
 /loop 1h Remind me to review the deploy status
 ```

+## Telegram
+
+Chat with Claude Code from Telegram by connecting a bot to your session. Install the [Telegram plugin](https://github.com/anthropics/claude-plugins-official), create a bot via [@BotFather](https://t.me/BotFather), then launch with the channel flag:
+
+```shell
+ollama launch claude -- --channels plugin:telegram@claude-plugins-official
+```
+
+Claude Code will prompt for permission on most actions. To allow the bot to work autonomously, configure [permission rules](https://code.claude.com/docs/en/permissions) or pass `--dangerously-skip-permissions` in isolated environments.
+
+See the [plugin README](https://github.com/anthropics/claude-plugins-official/tree/main/external_plugins/telegram) for full setup instructions including pairing and access control.
+
 ## Manual setup

 Claude Code connects to Ollama using the Anthropic-compatible API.
@@ -0,0 +1,13 @@
+---
+title: Claude Desktop
+---
+
+Claude Desktop is no longer supported by `ollama launch`.
+
+Existing installations can be restored to the usual Claude profile:
+
+```shell
+ollama launch claude-desktop --restore
+```
+
+Use [Claude Code](/integrations/claude-code) for Anthropic-compatible coding workflows with Ollama.
@@ -0,0 +1,98 @@
+---
+title: Cline CLI
+---
+
+Cline CLI is an autonomous coding agent for interactive terminal sessions.
+
+<img
+  src="/images/cline-cli.png"
+  alt="Cline CLI launched with Ollama selected as the provider"
+  style={{ borderRadius: "12px" }}
+/>
+
+## Install
+
+Install the [Cline CLI](https://docs.cline.bot/usage/cli-overview). For the IDE extension, see [Cline](/integrations/cline).
+
+```bash
+npm install -g cline
+```
+
+<Note>If Cline CLI is not installed and `npm` is available, `ollama launch cline` will prompt to install `cline@latest`.</Note>
+
+## Usage with Ollama
+
+### Quick setup
+
+```bash
+ollama launch cline
+```
+
+When launched through `ollama launch cline`, Ollama sets Cline's provider to Ollama, points it at the local Ollama endpoint, and selects the model you choose.
+
+To configure without launching:
+
+```shell
+ollama launch cline --config
+```
+
+### Run directly with a model
+
+```shell
+ollama launch cline --model qwen3.5
+```
+
+To use a cloud model:
+
+```shell
+ollama launch cline --model kimi-k2.6:cloud
+```
+
+
+### Pass a prompt to Cline
+
+Arguments after `--` are passed directly to Cline:
+
+```shell
+ollama launch cline -- "summarize this repository"
+```
+
+To open Cline's Kanban board:
+
+```shell
+ollama launch cline -- kanban
+```
+
+<img
+  src="/images/cline-kanban.png"
+  alt="Cline Kanban board opened from the CLI"
+  style={{ borderRadius: "12px" }}
+/>
+
+### Manual setup
+
+To configure Cline CLI manually, first make sure Ollama is running and the model you want to use is available:
+
+```shell
+ollama pull qwen3.5
+```
+
+Then run Cline's interactive auth flow:
+
+```shell
+cline auth
+```
+
+Select Ollama as the provider, use `http://localhost:11434` as the base URL if prompted, and choose a model such as `qwen3.5` or `kimi-k2.6:cloud`.
+
+To check the current Cline configuration:
+
+```shell
+cline config
+```
+
+To start an interactive session:
+
+```shell
+cline
+```
@@ -0,0 +1,82 @@
+---
+title: Codex App
+---
+
+Codex App is OpenAI's desktop coding agent for macOS and Windows. Ollama configures the app to use Ollama's OpenAI-compatible endpoint, so Codex can work with local models and Ollama Cloud models in the desktop app.
+
+<img
+  src="/images/codex-app-home.png"
+  alt="Codex App with Ollama selected"
+  style={{ borderRadius: "12px" }}
+/>
+
+## Install
+
+Install the [Codex App](https://developers.openai.com/codex/quickstart/) for macOS or Windows.
+
+<Note>Codex App support is available in Ollama v0.24.0 and newer.</Note>
+
+
+## Quick setup
+
+```shell
+ollama launch codex-app
+```
+
+Once Codex App opens, start a task or open a repository as usual.
+
+## Built-in browser
+
+Codex App can open local servers and sites in its built-in browser. Annotate directly on the page to request changes.
+
+<img
+  src="/images/codex-app-annotate.png"
+  alt="Codex App browser annotations"
+  style={{ borderRadius: "12px" }}
+/>
+
+## Review mode
+
+Use review mode to inspect code changes, leave comments, and iterate on fixes without leaving the app.
+
+<img
+  src="/images/codex-app-review.png"
+  alt="Codex App review comments"
+  style={{ borderRadius: "12px" }}
+/>
+
+### Run directly with a model
+
+```shell
+ollama launch codex-app --model kimi-k2.6:cloud
+```
+
+Use a local model by passing its model name:
+
+```shell
+ollama launch codex-app --model gemma4:31b
+```
+
+Running `ollama launch codex-app` is persistent and will have your model selected next time you open Codex.
+
+
+### Restore Codex App
+
+To switch Codex App back to the profile you were using before `ollama launch codex-app`, run:
+
+```shell
+ollama launch codex-app --restore
+```
+
+Ollama restores Codex App's settings and configs. If Codex App is open, Ollama asks before restarting it.
+
+
+The Codex CLI profile managed by `ollama launch codex` is left separate from the Codex App profile.
+
+Before overwriting Codex App config files, Ollama Launch saves backups under `~/.ollama/backup/codex-app/`. On Windows, `~` resolves to your user profile directory.
+
+## Troubleshooting
+
+If Codex App does not open after setup, open Codex manually once and run `ollama launch codex-app` again.
+
+If Codex App is already running and does not switch models, allow Ollama to restart it when prompted, or quit Codex App and run `ollama launch codex-app` again.
@@ -1,11 +1,11 @@
 ---
-title: Codex
+title: Codex CLI
 ---


 ## Install

-Install the [Codex CLI](https://developers.openai.com/codex/cli/):
+Install the [Codex CLI](https://developers.openai.com/codex/cli/). For the desktop app, see [Codex App](/integrations/codex-app).

 ```
 npm install -g @openai/codex
@@ -21,12 +21,21 @@ npm install -g @openai/codex
 ollama launch codex
 ```

+When launched through `ollama launch codex`, Ollama refreshes the model catalog
+and uses a dedicated Codex profile for that session.
+
 To configure without launching:

 ```shell
 ollama launch codex --config
 ```

+To remove the Ollama launch profile and generated model catalog:
+
+```shell
+ollama launch codex --restore
+```
+
 ### Manual setup

 To use `codex` with Ollama, use the `--oss` flag:
@@ -35,36 +44,35 @@ To use `codex` with Ollama, use the `--oss` flag:
 codex --oss
 ```

-### Changing Models
-
-By default, codex will use the local `gpt-oss:20b` model. However, you can specify a different model with the `-m` flag:
+To use a specific model, pass the `-m` flag:

 ```
 codex --oss -m gpt-oss:120b
 ```

-### Cloud Models
+To use a cloud model:

 ```
 codex --oss -m gpt-oss:120b-cloud
 ```

+### Profile-based setup

-## Connecting to ollama.com
-
-
-Create an [API key](https://ollama.com/settings/keys) from ollama.com and export it as `OLLAMA_API_KEY`.
-
-To use ollama.com directly, edit your `~/.codex/config.toml` file to point to ollama.com.
+For a persistent Codex CLI configuration, create `~/.codex/ollama-launch.config.toml`:

 ```toml
 model = "gpt-oss:120b"
-model_provider = "ollama"
+model_provider = "ollama-launch"
+model_catalog_json = "/Users/you/.codex/model.json"

-[model_providers.ollama]
+[model_providers.ollama-launch]
 name = "Ollama"
-base_url = "https://ollama.com/v1"
-env_key = "OLLAMA_API_KEY"
+base_url = "http://localhost:11434/v1/"
+wire_api = "responses"
 ```

-Run `codex` in a new terminal to load the new settings.
+Then run:
+
+```
+codex --profile ollama-launch
+```
@@ -0,0 +1,93 @@
+---
+title: Copilot CLI
+---
+
+GitHub Copilot CLI is GitHub's AI coding agent for the terminal. It can understand your codebase, make edits, run commands, and help you build software faster.
+
+Open models can be used with Copilot CLI through Ollama, enabling you to use models such as `qwen3.5`, `glm-5.1:cloud`, `kimi-k2.5:cloud`.
+
+## Install
+
+Install [Copilot CLI](https://github.com/features/copilot/cli/):
+
+<CodeGroup>
+
+```shell macOS / Linux (Homebrew)
+brew install copilot-cli
+```
+
+```shell npm (all platforms)
+npm install -g @github/copilot
+```
+
+```shell macOS / Linux (script)
+curl -fsSL https://gh.io/copilot-install | bash
+```
+
+```powershell Windows (WinGet)
+winget install GitHub.Copilot
+```
+
+</CodeGroup>
+
+## Usage with Ollama
+
+### Quick setup
+
+```shell
+ollama launch copilot
+```
+
+### Run directly with a model
+
+```shell
+ollama launch copilot --model kimi-k2.5:cloud
+```
+
+## Recommended Models
+
+- `kimi-k2.5:cloud`
+- `glm-5:cloud`
+- `minimax-m2.7:cloud`
+- `qwen3.5:cloud`
+- `glm-4.7-flash`
+- `qwen3.5`
+
+Cloud models are also available at [ollama.com/search?c=cloud](https://ollama.com/search?c=cloud).
+
+## Non-interactive (headless) mode
+
+Run Copilot CLI without interaction for use in Docker, CI/CD, or scripts:
+
+```shell
+ollama launch copilot --model kimi-k2.5:cloud --yes -- -p "how does this repository work?"
+```
+
+The `--yes` flag auto-pulls the model, skips selectors, and requires `--model` to be specified. Arguments after `--` are passed directly to Copilot CLI.
+
+## Manual setup
+
+Copilot CLI connects to Ollama using the OpenAI-compatible API via environment variables.
+
+1. Set the environment variables:
+
+```shell
+export COPILOT_PROVIDER_BASE_URL=http://localhost:11434/v1
+export COPILOT_PROVIDER_API_KEY=
+export COPILOT_PROVIDER_WIRE_API=responses
+export COPILOT_MODEL=qwen3.5
+```
+
+1. Run Copilot CLI:
+
+```shell
+copilot
+```
+
+Or run with environment variables inline:
+
+```shell
+COPILOT_PROVIDER_BASE_URL=http://localhost:11434/v1 COPILOT_PROVIDER_API_KEY= COPILOT_PROVIDER_WIRE_API=responses COPILOT_MODEL=glm-5:cloud copilot
+```
+
+**Note:** Copilot requires a large context window. We recommend at least 64k tokens. See the [context length documentation](/context-length) for how to adjust context length in Ollama.
@@ -0,0 +1,28 @@
+---
+title: Hermes Desktop
+---
+
+Hermes Desktop is a native AI assistant app by Nous Research. It provides a desktop chat interface for Hermes Agent, an AI agent that can work with models, run tools, manage projects, use memory and skills, and connect to messaging gateways.
+
+![Hermes Desktop with Ollama](http://files.ollama.com/hermes-agent.png)
+
+## Quick start
+
+```bash
+ollama launch hermes-desktop
+```
+
+Ollama handles the setup flow automatically:
+
+1. **Install** - If Hermes Desktop isn't installed, Ollama prompts to install it
+2. **Model** - Pick a model from the selector
+3. **Configure** - Ollama configures Hermes Desktop to use your selected Ollama model
+4. **Launch** - Ollama opens Hermes Desktop
+
+## Run directly with a model
+
+```bash
+ollama launch hermes-desktop --model <model>
+```
+
+Run `ollama launch hermes-desktop` again to switch models later.
@@ -0,0 +1,117 @@
+---
+title: Hermes Agent
+---
+
+Hermes Agent is a self-improving AI agent built by Nous Research. It features automatic skill creation, cross-session memory, and 70+ skills that it ships with by default. 
+
+![Hermes Agent with Ollama](/images/hermes.png)
+
+## Quick start
+
+```bash
+ollama launch hermes
+```
+
+Ollama handles everything automatically:
+
+1. **Install** — If Hermes isn't installed, Ollama prompts to install it via the Nous Research install script
+2. **Model** — Pick a model from the selector (local or cloud)
+3. **Onboarding** — Ollama configures the Ollama provider, points Hermes at `http://127.0.0.1:11434/v1`, and sets your model as the primary
+4. **Gateway** — Optionally connects a messaging platform (Telegram, Discord, Slack, WhatsApp, Signal, Email) and launches the Hermes chat
+
+## Recommended models
+
+**Cloud models**:
+
+- `kimi-k2.5:cloud` — Multimodal reasoning with subagents
+- `glm-5.1:cloud` — Reasoning and code generation
+- `qwen3.5:cloud` — Reasoning, coding, and agentic tool use with vision
+- `minimax-m2.7:cloud` — Fast, efficient coding and real-world productivity
+
+**Local models:**
+
+- `gemma4` — Reasoning and code generation locally (~16 GB VRAM)
+- `qwen3.6` — Reasoning, coding, and visual understanding locally (~24 GB VRAM)
+
+More models at [ollama.com/search](https://ollama.com/search?c=cloud).
+
+## Connect messaging apps
+
+Link Telegram, Discord, Slack, WhatsApp, Signal, or Email to chat with your models from anywhere:
+
+```bash
+hermes gateway setup
+```
+
+## Reconfigure
+
+Re-run the full setup wizard at any time:
+
+```bash
+hermes setup
+```
+
+## Manual setup
+
+If you'd rather drive Hermes's own wizard instead of `ollama launch hermes`, install it directly:
+
+```bash
+curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash
+```
+
+Hermes launches the setup wizard automatically. Choose **Quick setup**:
+
+```
+How would you like to set up Hermes?
+
+ →  Quick setup — provider, model & messaging (recommended)
+    Full setup — configure everything
+```
+
+### Connect to Ollama
+
+1. Select **More providers...**
+2. Select **Custom endpoint (enter URL manually)**
+3. Set the API base URL to the Ollama OpenAI-compatible endpoint:
+
+   ```
+   API base URL [e.g. https://api.example.com/v1]: http://127.0.0.1:11434/v1
+   ```
+
+4. Leave the API key blank (not required for local Ollama):
+
+   ```
+   API key [optional]:
+   ```
+
+5. Hermes auto-detects downloaded models, confirm the one you want:
+
+   ```
+   Verified endpoint via http://127.0.0.1:11434/v1/models (1 model(s) visible)
+     Detected model: kimi-k2.5:cloud
+     Use this model? [Y/n]:
+   ```
+
+6. Leave context length blank to auto-detect:
+
+   ```
+   Context length in tokens [leave blank for auto-detect]:
+   ```
+
+### Connect messaging
+
+Optionally connect a messaging platform during setup:
+
+```
+Connect a messaging platform? (Telegram, Discord, etc.)
+
+ →  Set up messaging now (recommended)
+    Skip — set up later with 'hermes setup gateway'
+```
+
+### Launch
+
+```
+Launch hermes chat now? [Y/n]: Y
+```
+
@@ -9,17 +9,24 @@ Ollama integrates with a wide range of tools.
 Coding assistants that can read, modify, and execute code in your projects.

 - [Claude Code](/integrations/claude-code)
- [Codex](/integrations/codex)
+- [Codex App](/integrations/codex-app)
+- [Codex CLI](/integrations/codex)
+- [Copilot CLI](/integrations/copilot-cli)
+- [Cline CLI](/integrations/cline-cli)
 - [OpenCode](/integrations/opencode)
 - [Droid](/integrations/droid)
 - [Goose](/integrations/goose)
+- [Oh My Pi](/integrations/oh-my-pi)
 - [Pi](/integrations/pi)
+- [Pool](/integrations/pool)

 ## Assistants

 AI assistants that help with everyday tasks.

 - [OpenClaw](/integrations/openclaw)
+- [Hermes Agent](/integrations/hermes)
+- [Hermes Desktop](/integrations/hermes-desktop)

 ## IDEs & Editors

@@ -0,0 +1,67 @@
+---
+title: NemoClaw
+---
+
+NemoClaw is NVIDIA's open source security stack for [OpenClaw](/integrations/openclaw). It wraps OpenClaw with the NVIDIA OpenShell runtime to provide kernel-level sandboxing, network policy controls, and audit trails for AI agents. 
+
+## Quick start
+
+Pull a model:
+
+```bash
+ollama pull nemotron-3-nano:30b
+```
+
+Run the installer:
+
+```bash
+curl -fsSL https://www.nvidia.com/nemoclaw.sh | \
+  NEMOCLAW_NON_INTERACTIVE=1 \
+  NEMOCLAW_PROVIDER=ollama \
+  NEMOCLAW_MODEL=nemotron-3-nano:30b \
+  bash
+```
+
+Connect to your sandbox:
+
+```bash
+nemoclaw my-assistant connect
+```
+
+Open the TUI:
+
+```bash
+openclaw tui
+```
+
+<Note>Ollama support in NemoClaw is still experimental.</Note>
+
+## Platform support
+
+| Platform | Runtime | Status |
+|----------|---------|--------|
+| Linux (Ubuntu 22.04+) | Docker | Primary |
+| macOS (Apple Silicon) | Colima or Docker Desktop | Supported |
+| Windows | WSL2 with Docker Desktop | Supported |
+
+CMD and PowerShell are not supported on Windows — WSL2 is required.
+
+<Note>Ollama must be installed and running before the installer runs. When running inside WSL2 or a container, ensure Ollama is reachable from the sandbox (e.g. `OLLAMA_HOST=0.0.0.0`).</Note>
+
+## System requirements
+
+- CPU: 4 vCPU minimum
+- RAM: 8 GB minimum (16 GB recommended)
+- Disk: 20 GB free (40 GB recommended for local models)
+- Node.js 20+ and npm 10+
+- Container runtime (Docker preferred)
+
+## Recommended models
+
+- `nemotron-3-super:cloud` — Strong reasoning and coding
+- `qwen3.5:cloud` — 397B; reasoning and code generation
+- `nemotron-3-nano:30b` — Recommended local model; fits in 24 GB VRAM
+- `qwen3.5:27b` — Fast local reasoning (~18 GB VRAM)
+- `glm-4.7-flash` — Reasoning and code generation (~25 GB VRAM)
+
+More models at [ollama.com/search](https://ollama.com/search).
@@ -0,0 +1,35 @@
+---
+title: Oh My Pi
+---
+
+Oh My Pi (OMP) is a terminal coding agent with IDE-style tools built in. It combines chat, project context, structured code edits, language server support, debugging tools, browser access, plugins, and subagents in one terminal workflow.
+
+Ollama can configure OMP to use Ollama as its model provider and launch an interactive session.
+
+![Oh My Pi with Ollama](http://files.ollama.com/omp.png)
+
+## Quick setup
+
+```bash
+ollama launch omp
+```
+
+This configures Ollama as a provider, sets up web search tools, and starts OMP.
+
+### Run directly with a model
+
+```shell
+ollama launch omp --model <model>
+```
+
+## Plugins
+
+OMP supports plugins for extra tools and capabilities. When launching OMP through Ollama, the Ollama web search plugin is managed automatically.
+
+## Manual setup
+
+Install OMP from [omp.sh](https://omp.sh), then run:
+
+```bash
+ollama launch omp --config
+```
@@ -15,13 +15,29 @@ Ollama handles everything automatically:
 1. **Install** — If OpenClaw isn't installed, Ollama prompts to install it via npm
 2. **Security** — On the first launch, a security notice explains the risks of tool access
 3. **Model** — Pick a model from the selector (local or cloud)
-4. **Onboarding** — Ollama configures the provider, installs the gateway daemon, and sets your model as the primary
+4. **Onboarding** — Ollama configures the provider, installs the gateway daemon, sets your model as the primary, and enables OpenClaw's bundled Ollama web search
 5. **Gateway** — Starts in the background and opens the OpenClaw TUI

 <Note>OpenClaw requires a larger context window. It is recommended to use a context window of at least 64k tokens if using local models. See [Context length](/context-length) for more information.</Note>

 <Note>Previously known as Clawdbot. `ollama launch clawdbot` still works as an alias.</Note>

+## Web search and fetch
+
+OpenClaw ships with a bundled Ollama `web_search` provider that lets local or cloud-backed Ollama setups search the web through the configured Ollama host.
+
+```bash
+ollama launch openclaw
+```
+
+Ollama web search is enabled automatically when launching OpenClaw through Ollama. To configure it manually:
+
+```bash
+openclaw configure --section web
+```
+
+<Note>Ollama web search for local models requires `ollama signin`.</Note>
+
 ## Configure without launching

 To change the model without starting the gateway and TUI:
@@ -43,15 +59,27 @@ If the gateway is already running, it restarts automatically to pick up the new
 **Cloud models**:

 - `kimi-k2.5:cloud` — Multimodal reasoning with subagents
- `minimax-m2.5:cloud` — Fast, efficient coding and real-world productivity
- `glm-5:cloud` — Reasoning and code generation
+- `qwen3.5:cloud` — Reasoning, coding, and agentic tool use with vision
+- `glm-5.1:cloud` — Reasoning and code generation
+- `minimax-m2.7:cloud` — Fast, efficient coding and real-world productivity

 **Local models:**

- `glm-4.7-flash` — Reasoning and code generation locally (~25 GB VRAM)
+- `gemma4` — Reasoning and code generation locally (~16 GB VRAM)
+- `qwen3.5` — Reasoning, coding, and visual understanding locally (~11 GB VRAM)

 More models at [ollama.com/search](https://ollama.com/search?c=cloud).

+## Non-interactive (headless) mode
+
+Run OpenClaw without interaction for use in Docker, CI/CD, or scripts:
+
+```bash
+ollama launch openclaw --model kimi-k2.5:cloud --yes
+```
+
+The `--yes` flag auto-pulls the model, skips selectors, and requires `--model` to be specified.
+
 ## Connect messaging apps

 ```bash
@@ -65,4 +93,3 @@ Link WhatsApp, Telegram, Slack, Discord, or iMessage to chat with your local mod
 ```bash
 openclaw gateway stop
 ```
-
@@ -28,79 +28,4 @@ To configure without launching:
 ollama launch opencode --config
 ```

-### Manual setup
-
-Add a configuration block to `~/.config/opencode/opencode.json`:
-
-```json
-{
-  "$schema": "https://opencode.ai/config.json",
-  "provider": {
-    "ollama": {
-      "npm": "@ai-sdk/openai-compatible",
-      "name": "Ollama",
-      "options": {
-        "baseURL": "http://localhost:11434/v1"
-      },
-      "models": {
-        "qwen3-coder": {
-          "name": "qwen3-coder"
-        }
-      }
-    }
-  }
-}
-```
-
-## Cloud Models
-
-`glm-4.7:cloud` is the recommended model for use with OpenCode.
-
-Add the cloud configuration to `~/.config/opencode/opencode.json`:
-
-```json
-{
-  "$schema": "https://opencode.ai/config.json",
-  "provider": {
-    "ollama": {
-      "npm": "@ai-sdk/openai-compatible",
-      "name": "Ollama",
-      "options": {
-        "baseURL": "http://localhost:11434/v1"
-      },
-      "models": {
-        "glm-4.7:cloud": {
-          "name": "glm-4.7:cloud"
-        }
-      }
-    }
-  }
-}
-```
-
-## Connecting to ollama.com
-
-1. Create an [API key](https://ollama.com/settings/keys) from ollama.com and export it as `OLLAMA_API_KEY`.
-2. Update `~/.config/opencode/opencode.json` to point to ollama.com:
-
-```json
-{
-  "$schema": "https://opencode.ai/config.json",
-  "provider": {
-    "ollama": {
-      "npm": "@ai-sdk/openai-compatible",
-      "name": "Ollama Cloud",
-      "options": {
-        "baseURL": "https://ollama.com/v1"
-      },
-      "models": {
-        "glm-4.7:cloud": {
-          "name": "glm-4.7:cloud"
-        }
-      }
-    }
-  }
-}
-```
-
-Run `opencode` in a new terminal to load the new settings.
+<Note>`ollama launch opencode` passes its configuration to OpenCode inline via the `OPENCODE_CONFIG_CONTENT` environment variable. OpenCode deep-merges its config sources on startup, so anything you declare in `~/.config/opencode/opencode.json` is still respected and available inside OpenCode. Models declared only in `opencode.json` won't appear in `ollama launch`'s model-selection menu.</Note>
@@ -2,31 +2,83 @@
 title: Pi
 ---

-Pi is a minimal AI agent toolkit with plugin support.
+Pi is a minimal and extensible coding agent.

-## Install
-
-Install [Pi](https://github.com/badlogic/pi-mono):
-
-```bash
-npm install -g @mariozechner/pi-coding-agent
-```
-
-## Usage with Ollama
-
-### Quick setup
+## Quick setup

 ```bash
 ollama launch pi
 ```

+This installs Pi if needed, configures Ollama as a provider including web tools, and drops you into an interactive session.
+
 To configure without launching:

 ```shell
 ollama launch pi --config
 ```

-### Manual setup
+### Run directly with a model
+
+```shell
+ollama launch pi --model qwen3.5:cloud
+```
+
+Cloud models are also available at [ollama.com](https://ollama.com/search?c=cloud).
+
+## Extensions 
+
+Pi ships with four core tools: `read`, `write`, `edit`, and `bash`. All other capabilities are added through its extension system.
+
+On-demand capability packages invoked via `/skill:name` commands.  
+
+Install from npm or git:
+
+```bash
+pi install npm:@foo/some-tools
+pi install git:github.com/user/repo@v1
+```
+
+See all packages at [pi.dev](https://pi.dev/packages) 
+
+### Web search
+
+Pi can use web search and fetch tools via the `@ollama/pi-web-search` package.
+
+When launching Pi through Ollama, package install/update is managed automatically.
+To install manually:
+
+```bash
+pi install npm:@ollama/pi-web-search
+```
+
+### Autoresearch with `pi-autoresearch`
+
+[pi-autoresearch](https://github.com/davebcn87/pi-autoresearch) brings autonomous experiment loops to Pi. Inspired by Karpathy's autoresearch, it turns any measurable metric into an optimization target: test speed, bundle size, build time, model training loss, Lighthouse scores.
+
+```bash
+pi install https://github.com/davebcn87/pi-autoresearch
+```
+
+Tell Pi what to optimize. It runs experiments, benchmarks each one, keeps improvements, reverts regressions, and repeats — all autonomously. A built-in dashboard tracks every run with confidence scoring to distinguish real gains from benchmark noise.
+
+```bash
+/autoresearch optimize unit test runtime
+```
+
+Each kept experiment is automatically committed. Each failed one is reverted. When you're done, Pi can group improvements into independent branches for clean review and merge.
+
+## Manual setup
+
+### Install
+
+Install [Pi](https://github.com/earendil-works/pi):
+
+```bash
+npm install -g @earendil-works/pi-coding-agent
+```
+
+

 Add a configuration block to `~/.pi/agent/models.json`:

@@ -0,0 +1,54 @@
+---
+title: Pool
+---
+
+Pool is Poolside's software agent for the terminal, built for enterprise development workflows.
+
+## Install
+
+Install [Pool](https://github.com/poolsideai/pool):
+
+## Usage with Ollama
+
+### Quick setup
+
+```shell
+ollama launch pool
+```
+
+### Run directly with a model
+
+```shell
+ollama launch pool --model kimi-k2.6:cloud
+```
+
+### Pass arguments through to Pool
+
+Arguments after `--` are passed directly to Pool:
+
+```shell
+ollama launch pool -- --help
+```
+
+## Manual setup
+
+Pool connects to Ollama using the OpenAI-compatible API via environment variables.
+
+1. Set the environment variables:
+
+```shell
+export POOLSIDE_STANDALONE_BASE_URL=http://localhost:11434/v1
+export POOLSIDE_API_KEY=ollama
+```
+
+2. Run Pool with an Ollama model:
+
+```shell
+pool -m kimi-k2.6:cloud
+```
+
+Or run with environment variables inline:
+
+```shell
+POOLSIDE_STANDALONE_BASE_URL=http://localhost:11434/v1 POOLSIDE_API_KEY=ollama pool -m kimi-k2.6:cloud
+```
@@ -2,33 +2,84 @@
 title: VS Code
 ---

-## Install
+VS Code includes built-in AI chat through GitHub Copilot Chat. Ollama models can be used directly in the Copilot Chat model picker.

-Install [VS Code](https://code.visualstudio.com/download).

-## Usage with Ollama
+![VS Code with Ollama](/images/vscode.png)

-1. Open Copilot side bar found in top right window
+
+## Prerequisites
+
+- Ollama v0.18.3+
+- [VS Code 1.113+](https://code.visualstudio.com/download)
+- [GitHub Copilot Chat extension 0.41.0+](https://marketplace.visualstudio.com/items?itemName=GitHub.copilot-chat)
+
+<Note> VS Code requires you to be logged in to use its model selector, even for custom models. This doesn't require a paid GitHub Copilot account; GitHub Copilot Free will enable model selection for custom models.</Note>
+
+## Quick setup
+
+```shell
+ollama launch vscode
+```
+
+Recommended models will be shown after running the command. See the latest models at [ollama.com](https://ollama.com/search?c=tools).
+
+Make sure **Local** is selected at the bottom of the Copilot Chat panel to use your Ollama models.
+<div style={{ display: "flex", justifyContent: "center" }}>
+  <img
+    src="/images/local.png"
+    alt="Ollama Local Models"
+    width="60%"
+    style={{ borderRadius: "4px", marginTop: "10px", marginBottom: "10px" }}
+  />
+</div>
+
+
+## Run directly with a model
+
+```shell
+ollama launch vscode --model qwen3.5:cloud
+```
+Cloud models are also available at [ollama.com](https://ollama.com/search?c=cloud).
+
+## Manual setup
+
+To configure Ollama manually without `ollama launch`:
+
+1. Open the **Copilot Chat** side bar from the top right corner
   <div style={{ display: "flex", justifyContent: "center" }}>
     <img
       src="/images/vscode-sidebar.png"
       alt="VS Code chat Sidebar"
       width="75%"
+       style={{ borderRadius: "4px" }}
     />
   </div>
-2. Select the model dropdown > **Manage models**
+2. Click the **settings gear icon** (<Icon icon="gear" />) to bring up the Language Models window
   <div style={{ display: "flex", justifyContent: "center" }}>
     <img
-       src="/images/vscode-models.png"
+       src="/images/vscode-other-models.png"
       alt="VS Code model picker"
       width="75%"
+       style={{ borderRadius: "4px" }}
     />
   </div>
-3. Enter **Ollama** under **Provider Dropdown** and select desired models (e.g `qwen3, qwen3-coder:480b-cloud`)
+3. Click **Add Models** and select **Ollama** to load all your Ollama models into VS Code
   <div style={{ display: "flex", justifyContent: "center" }}>
     <img
-       src="/images/vscode-model-options.png"
-       alt="VS Code model options dropdown"
+       src="/images/vscode-add-ollama.png"
+       alt="VS Code model options dropdown to add ollama models"
       width="75%"
+       style={{ borderRadius: "4px" }}
+     />
+   </div>
+
+4. Click the **Unhide** button in the model picker to show your Ollama models  
+   <div style={{ display: "flex", justifyContent: "center" }}>
+     <img
+       src="/images/vscode-unhide.png"
+       alt="VS Code unhide models button"
+       width="75%"
+       style={{ borderRadius: "4px" }}
     />
   </div>
@@ -157,6 +157,7 @@ PARAMETER <parameter> <parametervalue>
 | seed           | Sets the random number seed to use for generation. Setting this to a specific number will make the model generate the same text for the same prompt. (Default: 0)                                                                                                                                                                                                               | int        | seed 42              |
 | stop           | Sets the stop sequences to use. When this pattern is encountered the LLM will stop generating text and return. Multiple stop patterns may be set by specifying multiple separate `stop` parameters in a modelfile.                                                                                                                                                              | string     | stop "AI assistant:" |
 | num_predict    | Maximum number of tokens to predict when generating text. (Default: -1, infinite generation)                                                                                                                                                                                                                                                                                    | int        | num_predict 42       |
+| draft_num_predict | Maximum number of speculative draft tokens to predict per step when a draft model is available. Separate draft models default to 4; embedded MTP tensors require setting this parameter. Set to 0 to disable speculative drafting.                                                                                                                                               | int        | draft_num_predict 4  |
 | top_k          | Reduces the probability of generating nonsense. A higher value (e.g. 100) will give more diverse answers, while a lower value (e.g. 10) will be more conservative. (Default: 40)                                                                                                                                                                                                | int        | top_k 40             |
 | top_p          | Works together with top-k. A higher value (e.g., 0.95) will lead to more diverse text, while a lower value (e.g., 0.5) will generate more focused and conservative text. (Default: 0.9)                                                                                                                                                                                         | float      | top_p 0.9            |
 | min_p          | Alternative to the top*p, and aims to ensure a balance of quality and variety. The parameter \_p* represents the minimum probability for a token to be considered, relative to the probability of the most likely token. For example, with _p_=0.05 and the most likely token having a probability of 0.9, logits with a value less than 0.045 are filtered out. (Default: 0.0) | float      | min_p 0.05           |
@@ -794,14 +794,14 @@ paths:
          label: Default
          source: |
            curl http://localhost:11434/api/generate -d '{
-              "model": "gemma3",
+              "model": "gemma4",
              "prompt": "Why is the sky blue?"
            }'
        - lang: bash
          label: Non-streaming
          source: |
            curl http://localhost:11434/api/generate -d '{
-              "model": "gemma3",
+              "model": "gemma4",
              "prompt": "Why is the sky blue?",
              "stream": false
            }'
@@ -809,7 +809,7 @@ paths:
          label: With options
          source: |
            curl http://localhost:11434/api/generate -d '{
-              "model": "gemma3",
+              "model": "gemma4",
              "prompt": "Why is the sky blue?",
              "options": {
                "temperature": 0.8,
@@ -821,7 +821,7 @@ paths:
          label: Structured outputs
          source: |
            curl http://localhost:11434/api/generate -d '{
-              "model": "gemma3",
+              "model": "gemma4",
              "prompt": "What are the populations of the United States and Canada?",
              "stream": false,
              "format": {
@@ -846,7 +846,7 @@ paths:
          label: With images
          source: |
            curl http://localhost:11434/api/generate -d '{
-              "model": "gemma3",
+              "model": "gemma4",
              "prompt": "What is in this picture?",
              "images": ["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"]
            }'
@@ -854,13 +854,13 @@ paths:
          label: Load model
          source: |
            curl http://localhost:11434/api/generate -d '{
-              "model": "gemma3"
+              "model": "gemma4"
            }'
        - lang: bash
          label: Unload model
          source: |
            curl http://localhost:11434/api/generate -d '{
-              "model": "gemma3",
+              "model": "gemma4",
              "keep_alive": 0
            }'
      requestBody:
@@ -870,7 +870,7 @@ paths:
            schema:
              $ref: "#/components/schemas/GenerateRequest"
            example:
-              model: gemma3
+              model: gemma4
              prompt: Why is the sky blue?
      responses:
        "200":
@@ -880,7 +880,7 @@ paths:
              schema:
                $ref: "#/components/schemas/GenerateResponse"
              example:
-                model: "gemma3"
+                model: "gemma4"
                created_at: "2025-10-17T23:14:07.414671Z"
                response: "Hello! How can I help you today?"
                done: true
@@ -906,7 +906,7 @@ paths:
          label: Default
          source: |
            curl http://localhost:11434/api/chat -d '{
-              "model": "gemma3",
+              "model": "gemma4",
              "messages": [
                {
                  "role": "user",
@@ -918,7 +918,7 @@ paths:
          label: Non-streaming
          source: |
            curl http://localhost:11434/api/chat -d '{
-              "model": "gemma3",
+              "model": "gemma4",
              "messages": [
                {
                  "role": "user",
@@ -931,7 +931,7 @@ paths:
          label: Structured outputs
          source: |
            curl -X POST http://localhost:11434/api/chat -H "Content-Type: application/json" -d '{
-              "model": "gemma3",
+              "model": "gemma4",
              "messages": [
                {
                  "role": "user",
@@ -1011,7 +1011,7 @@ paths:
          label: Images
          source: |
            curl http://localhost:11434/api/chat -d '{
-              "model": "gemma3",
+              "model": "gemma4",
              "messages": [
                {
                  "role": "user",
@@ -1036,7 +1036,7 @@ paths:
              schema:
                $ref: "#/components/schemas/ChatResponse"
              example:
-                model: "gemma3"
+                model: "gemma4"
                created_at: "2025-10-17T23:14:07.414671Z"
                message:
                  role: "assistant"
@@ -1148,17 +1148,17 @@ paths:
                $ref: "#/components/schemas/ListResponse"
              example:
                models:
-                  - name: "gemma3"
-                    model: "gemma3"
+                  - name: "gemma4"
+                    model: "gemma4"
                    modified_at: "2025-10-03T23:34:03.409490317-07:00"
-                    size: 3338801804
-                    digest: "a2af6cc3eb7fa8be8504abaf9b04e88f17a119ec3f04a3addf55f92841195f5a"
+                    size: 9608350245
+                    digest: "c6eb396dbd5992bbe3f5cdb947e8bbc0ee413d7c17e2beaae69f5d569cf982eb"
                    details:
                      format: "gguf"
-                      family: "gemma"
+                      family: "gemma4"
                      families:
-                        - "gemma"
-                      parameter_size: "4.3B"
+                        - "gemma4"
+                      parameter_size: "8.0B"
                      quantization_level: "Q4_K_M"
  /api/ps:
    get:
@@ -1181,17 +1181,17 @@ paths:
                $ref: "#/components/schemas/PsResponse"
              example:
                models:
-                  - name: "gemma3"
-                    model: "gemma3"
+                  - name: "gemma4"
+                    model: "gemma4"
                    size: 6591830464
-                    digest: "a2af6cc3eb7fa8be8504abaf9b04e88f17a119ec3f04a3addf55f92841195f5a"
+                    digest: "c6eb396dbd5992bbe3f5cdb947e8bbc0ee413d7c17e2beaae69f5d569cf982eb"
                    details:
                      parent_model: ""
                      format: "gguf"
-                      family: "gemma3"
+                      family: "gemma4"
                      families:
-                        - "gemma3"
-                      parameter_size: "4.3B"
+                        - "gemma4"
+                      parameter_size: "8.0B"
                      quantization_level: "Q4_K_M"
                    expires_at: "2025-10-17T16:47:07.93355-07:00"
                    size_vram: 5333539264
@@ -1205,13 +1205,13 @@ paths:
          label: Default
          source: |
            curl http://localhost:11434/api/show -d '{
-              "model": "gemma3"
+              "model": "gemma4"
            }'
        - lang: bash
          label: Verbose
          source: |
            curl http://localhost:11434/api/show -d '{
-              "model": "gemma3",
+              "model": "gemma4",
              "verbose": true
            }'
      requestBody:
@@ -1221,7 +1221,7 @@ paths:
            schema:
              $ref: "#/components/schemas/ShowRequest"
            example:
-              model: gemma3
+              model: gemma4
      responses:
        "200":
          description: Model information
@@ -1239,44 +1239,64 @@ paths:
                details:
                  parent_model: ""
                  format: "gguf"
-                  family: "gemma3"
+                  family: "gemma4"
                  families:
-                    - "gemma3"
-                  parameter_size: "4.3B"
+                    - "gemma4"
+                  parameter_size: "8.0B"
                  quantization_level: "Q4_K_M"
                model_info:
-                  gemma3.attention.head_count: 8
-                  gemma3.attention.head_count_kv: 4
-                  gemma3.attention.key_length: 256
-                  gemma3.attention.sliding_window: 1024
-                  gemma3.attention.value_length: 256
-                  gemma3.block_count: 34
-                  gemma3.context_length: 131072
-                  gemma3.embedding_length: 2560
-                  gemma3.feed_forward_length: 10240
-                  gemma3.mm.tokens_per_image: 256
-                  gemma3.vision.attention.head_count: 16
-                  gemma3.vision.attention.layer_norm_epsilon: 0.000001
-                  gemma3.vision.block_count: 27
-                  gemma3.vision.embedding_length: 1152
-                  gemma3.vision.feed_forward_length: 4304
-                  gemma3.vision.image_size: 896
-                  gemma3.vision.num_channels: 3
-                  gemma3.vision.patch_size: 14
-                  general.architecture: "gemma3"
+                  gemma4.attention.head_count: 8
+                  gemma4.attention.head_count_kv: 2
+                  gemma4.attention.key_length: 512
+                  gemma4.attention.key_length_swa: 256
+                  gemma4.attention.layer_norm_rms_epsilon: 0.000001
+                  gemma4.attention.shared_kv_layers: 18
+                  gemma4.attention.sliding_window: 512
+                  gemma4.attention.value_length: 512
+                  gemma4.attention.value_length_swa: 256
+                  gemma4.audio.attention.head_count: 8
+                  gemma4.audio.attention.layer_norm_epsilon: 0.000001
+                  gemma4.audio.block_count: 12
+                  gemma4.audio.conv_kernel_size: 5
+                  gemma4.audio.embedding_length: 1024
+                  gemma4.audio.feed_forward_length: 4096
+                  gemma4.block_count: 42
+                  gemma4.context_length: 131072
+                  gemma4.embedding_length: 2560
+                  gemma4.embedding_length_per_layer_input: 256
+                  gemma4.feed_forward_length: 10240
+                  gemma4.final_logit_softcapping: 30
+                  gemma4.rope.dimension_count: 512
+                  gemma4.rope.dimension_count_swa: 256
+                  gemma4.rope.freq_base: 1000000
+                  gemma4.rope.freq_base_swa: 10000
+                  gemma4.vision.attention.head_count: 12
+                  gemma4.vision.attention.layer_norm_epsilon: 0.000001
+                  gemma4.vision.block_count: 16
+                  gemma4.vision.embedding_length: 768
+                  gemma4.vision.feed_forward_length: 3072
+                  gemma4.vision.num_channels: 3
+                  gemma4.vision.patch_size: 16
+                  gemma4.vision.projector.scale_factor: 3
+                  general.architecture: "gemma4"
                  general.file_type: 15
-                  general.parameter_count: 4299915632
                  general.quantization_version: 2
-                  tokenizer.ggml.add_bos_token: true
+                  tokenizer.ggml.add_bos_token: false
                  tokenizer.ggml.add_eos_token: false
+                  tokenizer.ggml.add_mask_token: false
                  tokenizer.ggml.add_padding_token: false
                  tokenizer.ggml.add_unknown_token: false
                  tokenizer.ggml.bos_token_id: 2
                  tokenizer.ggml.eos_token_id: 1
+                  tokenizer.ggml.eos_token_ids:
+                    - 1
+                    - 106
+                    - 50
+                  tokenizer.ggml.mask_token_id: 4
                  tokenizer.ggml.merges: null
                  tokenizer.ggml.model: "llama"
                  tokenizer.ggml.padding_token_id: 0
-                  tokenizer.ggml.pre: "default"
+                  tokenizer.ggml.pre: "gemma4"
                  tokenizer.ggml.scores: null
                  tokenizer.ggml.token_type: null
                  tokenizer.ggml.tokens: null
@@ -1292,7 +1312,7 @@ paths:
          label: Default
          source: |
            curl http://localhost:11434/api/create -d '{
-              "from": "gemma3",
+              "from": "gemma4",
              "model": "alpaca",
              "system": "You are Alpaca, a helpful AI assistant. You only answer with Emojis."
            }'
@@ -1301,7 +1321,7 @@ paths:
          source: |
            curl http://localhost:11434/api/create -d '{
              "model": "ollama",
-              "from": "gemma3",
+              "from": "gemma4",
              "system": "You are Ollama the llama."
            }'
        - lang: bash
@@ -1320,7 +1340,7 @@ paths:
              $ref: "#/components/schemas/CreateRequest"
            example:
              model: mario
-              from: gemma3
+              from: gemma4
              system: "You are Mario from Super Mario Bros."
      responses:
        "200":
@@ -1347,8 +1367,8 @@ paths:
          label: Copy a model to a new name
          source: |
            curl http://localhost:11434/api/copy -d '{
-              "source": "gemma3",
-              "destination": "gemma3-backup"
+              "source": "gemma4",
+              "destination": "gemma4-backup"
            }'
      requestBody:
        required: true
@@ -1357,8 +1377,8 @@ paths:
            schema:
              $ref: "#/components/schemas/CopyRequest"
            example:
-              source: gemma3
-              destination: gemma3-backup
+              source: gemma4
+              destination: gemma4-backup
      responses:
        "200":
          description: Model successfully copied
@@ -1373,13 +1393,13 @@ paths:
          label: Default
          source: |
            curl http://localhost:11434/api/pull -d '{
-              "model": "gemma3"
+              "model": "gemma4"
            }'
        - lang: bash
          label: Non-streaming
          source: |
            curl http://localhost:11434/api/pull -d '{
-              "model": "gemma3",
+              "model": "gemma4",
              "stream": false
            }'
      requestBody:
@@ -1389,7 +1409,7 @@ paths:
            schema:
              $ref: "#/components/schemas/PullRequest"
            example:
-              model: gemma3
+              model: gemma4
      responses:
        "200":
          description: Pull status updates.
@@ -1457,7 +1477,7 @@ paths:
          label: Delete model
          source: |
            curl -X DELETE http://localhost:11434/api/delete -d '{
-              "model": "gemma3"
+              "model": "gemma4"
            }'
      requestBody:
        required: true
@@ -1466,7 +1486,7 @@ paths:
            schema:
              $ref: "#/components/schemas/DeleteRequest"
            example:
-              model: gemma3
+              model: gemma4
      responses:
        "200":
          description: Model successfully deleted
@@ -59,7 +59,7 @@ Use the [API](/api) to integrate Ollama into your applications:

 ```sh
 curl http://localhost:11434/api/chat -d '{
-  "model": "gemma3",
+  "model": "gemma4",
  "messages": [{ "role": "user", "content": "Hello!" }]
 }'
 ```
@@ -114,6 +114,25 @@ If you are experiencing problems getting Ollama to correctly discover or use you
 - `OLLAMA_DEBUG=1` During GPU discovery additional information will be reported
 - Check dmesg for any errors from amdgpu or kfd drivers `sudo dmesg | grep -i amdgpu` and `sudo dmesg | grep -i kfd`

+### AMD Driver Version Mismatch
+
+If your AMD GPU is not detected on Linux and the server logs contain messages like:
+
+```
+msg="failure during GPU discovery" ... error="failed to finish discovery before timeout"
+msg="bootstrap discovery took" duration=30s ...
+```
+
+This typically means the system's AMD GPU driver is too old. Ollama bundles
+ROCm 7 linux libraries which require a compatible ROCm 7 kernel driver. If the
+system is running an older driver (ROCm 6.x or earlier), GPU initialization
+will hang during device discovery and eventually time out, causing Ollama to
+fall back to CPU.
+
+To resolve this, upgrade to the ROCm v7 driver using the `amdgpu-install`
+utility from [AMD's ROCm documentation](https://rocm.docs.amd.com/projects/install-on-linux/en/latest/).
+After upgrading, reboot and restart Ollama.
+
 ## Multiple AMD GPUs

 If you experience gibberish responses when models load across multiple AMD GPUs on Linux, see the following guide.
@@ -12,10 +12,18 @@ terminal application. As usual the Ollama [API](/api) will be served on

 - Windows 10 22H2 or newer, Home or Pro
 - NVIDIA 452.39 or newer Drivers if you have an NVIDIA card
- AMD Radeon Driver https://www.amd.com/en/support if you have a Radeon card
+- AMD ROCm v7 / HIP7-capable driver stack for ROCm acceleration, or a Vulkan-capable AMD Radeon driver for Vulkan acceleration

 Ollama uses unicode characters for progress indication, which may render as unknown squares in some older terminal fonts in Windows 10. If you see this, try changing your terminal font settings.

+<Note>
+  Some RDNA2 / Radeon RX 6000 systems, including RX 6800-class cards, may not
+  expose ROCm v7 on current Windows AMD drivers. Vulkan is enabled by default
+  and is the recommended fallback for those systems. If a mixed iGPU/dGPU
+  system selects an unstable Vulkan iGPU, set `GGML_VK_VISIBLE_DEVICES` to the
+  discrete GPU index.
+</Note>
+
 ## Filesystem Requirements

 The Ollama install does not require Administrator, and installs in your home directory by default. You'll need at least 4GB of space for the binary install. Once you've installed Ollama, you'll need additional space for storing the Large Language models, which can be tens to hundreds of GB in size. If your home directory doesn't have enough space, you can change where the binaries are installed, and where the models are stored.
@@ -80,9 +88,13 @@ help you keep up to date.

 If you'd like to install or integrate Ollama as a service, a standalone
 `ollama-windows-amd64.zip` zip file is available containing only the Ollama CLI
-and GPU library dependencies for Nvidia. If you have an AMD GPU, also download
-and extract the additional ROCm package `ollama-windows-amd64-rocm.zip` into the
-same directory. This allows for embedding Ollama in existing applications, or
+and GPU library dependencies for Nvidia. Depending on your hardware, you may also
+need to download and extract additional packages into the same directory:
+
+- **AMD GPU**: `ollama-windows-amd64-rocm.zip`
+- **MLX (CUDA)**: `ollama-windows-amd64-mlx.zip`
+
+This allows for embedding Ollama in existing applications, or
 running it as a system service via `ollama serve` with tools such as
 [NSSM](https://nssm.cc/).
Author	SHA1	Message	Date
The Librarian	70feb5e845	Sync ollama docs from 12e04379 on 2026-06-12	2026-06-12 04:00:06 +00:00
The Librarian	7c43dfec77	Sync ollama docs from 421faa02 on 2026-05-12	2026-05-12 04:00:09 +00:00
The Librarian	5fb5e0e5b8	Sync ollama docs from 9330bb91 on 2026-04-12	2026-04-12 04:00:07 +00:00