docs(architecture): correct agent model to gemma4:e2b

Co-Authored-By: Claude <noreply@anthropic.com>
fix(config): default agent model to gemma4:e2b
2026-08-07 14:59:12 +02:00 · 2026-08-07 14:58:57 +02:00 · 2026-08-07 14:51:58 +02:00 · 2026-07-19 17:11:08 +02:00
6 changed files with 10 additions and 7 deletions
@@ -270,7 +270,7 @@ echo $WEBBER_API_URL  # Should be http://localhost:8095
 curl http://192.168.86.149:11434/api/tags

 # Check model is available
-curl http://192.168.86.149:11434/api/tags | grep mistral-nemo
+curl http://192.168.86.149:11434/api/tags | grep gemma4
 ```

 ### Tests failing
@@ -107,7 +107,7 @@ This project uses prefixed tags for independent release cycles:
 ## Requirements

 - Python 3.12+
- Ollama running with `mistral-nemo:latest` model
+- Ollama running with `gemma4:e2b` model
 - Docker (for production deployment)
 - SearXNG (optional, for web search)

@@ -14,7 +14,7 @@ CORS_ORIGINS=["http://localhost:3000","http://localhost:8080"]

 # LLM - Ollama (tower-of-joy)
 OLLAMA_URL=http://192.168.86.149:11434
-OLLAMA_AGENT_MODEL=mistral-nemo-large:latest
+OLLAMA_AGENT_MODEL=gemma4:e2b
 OLLAMA_EMBED_MODEL=nomic-embed-text:latest

 # Auth - Tatlock integration (optional)
@@ -7,6 +7,9 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0

 ## [Unreleased]

+### Changed
+- Default `OLLAMA_AGENT_MODEL` is now `gemma4:e2b` instead of `mistral-nemo-large:latest`, so a deployment without an explicit override no longer exhausts shared GPU memory
+
 ## [1.0.1] - 2026-07-19

 ### Changed
@@ -13,7 +13,7 @@ Webber is a FastAPI-based agent orchestration service that provides:
 **Port:** 8086
 **Runtime:** Python 3.12, FastAPI, Uvicorn
 **Agent Framework:** PydanticAI
-**Default LLM:** Ollama with mistral-nemo-large:latest
+**Default LLM:** Ollama with gemma4:e2b

 ---

@@ -205,7 +205,7 @@ All settings via environment variables or `.env`:
 | HOST | 0.0.0.0 | Server host |
 | PORT | 8086 | Server port |
 | OLLAMA_URL | http://192.168.86.149:11434 | Ollama API URL |
-| OLLAMA_AGENT_MODEL | mistral-nemo-large:latest | Agent reasoning model |
+| OLLAMA_AGENT_MODEL | gemma4:e2b | Agent reasoning model |
 | OLLAMA_EMBED_MODEL | nomic-embed-text:latest | Embedding model |
 | TATLOCK_API_URL | http://tatlock:8000 | Tatlock auth service |
 | SEARXNG_URL | http://searxng:8080 | SearXNG web search instance |
@@ -359,7 +359,7 @@ Gitea Actions workflow:

 Deployed in Portainer `agents` stack alongside Tatlock:
 - Network: `docker-dataplane`
- Registry: `git.schweitz.internal/jpmschweitzer/webber`
+- Registry: `git.schweitz.net/jpmschweitzer/webber`
 - Auto-update: Watchtower with label `com.centurylinklabs.watchtower.enable=true`

 ---
@@ -64,7 +64,7 @@ class Settings(BaseSettings):

    # LLM - Ollama (always hot in VRAM on tower-of-joy)
    ollama_url: str = "http://192.168.86.149:11434"
-    ollama_agent_model: str = "mistral-nemo-large:latest"
+    ollama_agent_model: str = "gemma4:e2b"
    ollama_embed_model: str = "nomic-embed-text:latest"

    # Auth - Tatlock integration
Author	SHA1	Message	Date
jpmschweitzerandClaude	9612c7af05	docs(architecture): correct agent model to gemma4:e2b Co-Authored-By: Claude <noreply@anthropic.com>	2026-08-07 14:59:12 +02:00
jpmschweitzerandClaude	9d58df85f3	fix(config): default agent model to gemma4:e2b mistral-nemo-large holds ~9.2 GB of the 11 GB card it shares with Speaches, which starves Whisper and breaks voice transcription. gemma4:e2b holds 1.9 GB and is faster. The deployed stack already overrides this via OLLAMA_AGENT_MODEL; this aligns the default so a deployment without that override does not reintroduce the contention. Co-Authored-By: Claude <noreply@anthropic.com>	2026-08-07 14:58:57 +02:00
jpmschweitzerandClaude	3de96fe070	docs: correct Ollama model references to gemma4:e2b The deployed agent model is gemma4:e2b; these references still named mistral-nemo, so the setup and troubleshooting steps checked for a model that is no longer expected to be present. Co-Authored-By: Claude <noreply@anthropic.com>	2026-08-07 14:51:58 +02:00
jpmschweitzerandClaude Fable 5	e2acb7de39	docs(architecture): registry is git.schweitz.net not git.schweitz.internal Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>	2026-07-19 17:11:08 +02:00