Pomera AI Commander

--- description: Workflow for updating AI model defaults in Pomera AI Commander as part of releases --- # AI Model Update Workflow This workflow ensures AI model defaults stay current with the latest available models from each provider. ## When to Run - Before each major release - Monthly maintenance updates - When a provider announces significant model changes ## Steps ### 1. Research Current Models // turbo Check each provider's documentation for latest models: ```bash # Provider documentation URLs # Google AI: https://ai.google.dev/gemini-api/docs/models # OpenAI: https://platform.openai.com/docs/models # Anthropic: https://docs.anthropic.com/en/docs/models-overview # Cohere: https://docs.cohere.com/docs/models # HuggingFace: https://huggingface.co/models?inference=warm # Groq: https://console.groq.com/docs/models # OpenRouter: https://openrouter.ai/docs/models (check :free suffix for free models) # AWS Bedrock: https://docs.aws.amazon.com/bedrock/latest/userguide/models.html # Vertex AI: https://cloud.google.com/vertex-ai/generative-ai/docs/model-reference/overview ``` ### 2. Update Settings Defaults Registry Edit `core/settings_defaults_registry.py`: 1. Find the provider section (e.g., `# Google AI - Updated ...`) 2. Update the comment date to current month/year 3. Update the `MODEL` default if there's a new recommended model 4. Update `MODELS_LIST` with latest models in order of: - Default/recommended model first - Other premium models - Free models (if available, marked with `:free` suffix for OpenRouter) ### 3. Update Free Models (OpenRouter) OpenRouter free models change frequently. Check https://openrouter.ai/models?pricing=free: - Models with `:free` suffix in OpenRouter are free to use - Popular free models to include: - `google/gemini-2.5-flash:free` - `meta-llama/llama-3.3-70b-instruct:free` - `qwen/qwen-2.5-72b-instruct:free` - `deepseek/deepseek-chat:free` ### 4. Update HuggingFace Models HuggingFace free inference changes based on model popularity: - Check https://huggingface.co/models?inference=warm for available models - Prioritize models with "Instruct" or "Chat" in the name - Include popular coding models like Qwen2.5-Coder ### 5. Verify Changes // turbo Run the app to verify changes load correctly: ```bash cd P:\Pomera-AI-Commander python pomera.py ``` 1. Open AI Tools 2. Verify each provider shows updated model list 3. Verify default model is selectable 4. Test at least one API call if possible ### 6. Update Version Notes Add model updates to release notes: ```markdown ### AI Model Updates - Updated Google AI models to Gemini 2.5 series - Added new free models for OpenRouter - Added Studio LM provider for local LLM support ``` ## Model Update Checklist | Provider | Default Model | Free Options | Last Updated | |----------|---------------|--------------|--------------| | Google AI | gemini-2.5-pro | N/A | Jan 2026 | | Azure AI | gpt-4.1 | N/A | Dec 2025 | | Anthropic AI | claude-sonnet-4-5 | N/A | Dec 2025 | | OpenAI | gpt-4.1 | N/A | Dec 2025 | | Cohere AI | command-a-03-2025 | N/A | Dec 2025 | | HuggingFace AI | Llama-3.3-70B-Instruct | All (free tier) | Jan 2026 | | Groq AI | llama-3.3-70b-versatile | N/A | Dec 2025 | | OpenRouterAI | claude-sonnet-4.5 | 7+ free models | Jan 2026 | | AWS Bedrock | claude-3-5-sonnet | N/A | Dec 2025 | | Vertex AI | gemini-2.5-pro | N/A | Dec 2025 | | Studio LM | local-model | All (local) | Jan 2026 | ## Notes - Always test API calls after updating to ensure model names are correct - OpenRouter model names use `provider/model-name` format - HuggingFace model names use `org/Model-Name` format (case-sensitive) - Free models may have rate limits or be removed without notice