agentjido / req_llm / 51450757973bd80b6aea84a3ba9b94f9df0e2e09

49%

Build Type

push

github

Committed by web-flow

Commit Message

Fix Google provider v1beta default and update test fixtures (#169)

* Refresh Anthropic Model Fixtures from a Live API call

* Refresh Cerebras model fixtures from Live API calls

* Refresh Groq Fixtures with Live API Calls

* Refresh xAI Models with Live API Calls

* Refresh zAI fixtures with Live LLM API Calls

* Refresh zai_coder models with live LLM calls

* Refresh Openrouter fixtures with Live LLM calls

* Refresh OpenAI models from live API calls

* Tested and confirmed zai_coder:glm-4.5-flash

* Add catalog_allow configuration for various LLM models

Introduce a new configuration file `catalog_allow.exs` to specify allowed models for multiple providers including Anthropic, OpenAI, Google, Groq, xAI, OpenRouter, Amazon Bedrock, and Google Vertex AI. Update main config to enable catalog and import the new configuration. Modify catalog loading logic to handle model filtering based on the new configuration and enable/disable functionality.

* Update Google provider to use v1beta API and adjust related configurations

- Changed base URL for Google provider from v1 to v1beta across multiple files.
- Updated documentation to reflect the default API version as v1beta.
- Modified tests to ensure compatibility with the new default API version.
- Adjusted model sync task documentation and JSON files to align with the updated API versioning.

* Refactor model configuration and filtering logic

- Removed hardcoded sample models from the main configuration file.
- Added sample model configurations to the test configuration file for better isolation.
- Enhanced model filtering logic in the catalog to allow for dynamic model specification based on allowed patterns.
- Updated model matrix to utilize allowed models based on the new catalog configuration.
- Introduced methods for resolving allowed model specifications and matching patterns.

* Update model configurations and refresh test fixtures

- Enhanced `catalog_allow.exs` to clarify model statuses for Anth... (continued)

Run Details

39 of 79 new or added lines in 4 files covered. (49.37%)

14 existing lines in 9 files now uncovered.

3934 of 7506 relevant lines covered (52.41%)

229.87 hits per line