* OpenRouter Gemini caching
* Fix tests
* Remove unsupported models
* Clean up the task header a bit
* Update src/api/providers/openrouter.ts
Co-authored-by: ellipsis-dev[bot] <65095814+ellipsis-dev[bot]@users.noreply.github.com>
* Remove model that doesn't seem to work
---------
Co-authored-by: ellipsis-dev[bot] <65095814+ellipsis-dev[bot]@users.noreply.github.com>
* Update validate.ts
Allow ARNs from Bedrock Marketplace, which are different because models are deployed using SageMaker Inference behind the scenes.
* Update bedrock.ts
Allow ARNs from Bedrock Marketplace, which are different because models are deployed using SageMaker Inference behind the scenes.
The FakeAI object passed by the user must be exactly the same object
that is passed to FakeAIHandler via API configuration.
Unfortunatelly, as the VSCode global state is used as configuration
storage, we lose this property (VSCode global state creates copies of
the object). Also the class of the stored object is lost, so methods
of the object are unavailable.
Therefore, we store the original objects in global variable and
use ID field of FakeAI object to identify the original object.
* feat: initialize VS Code Language Model client in constructor
* feat: add VS Code LLM models and configuration
* feat: integrate VS Code LLM models into API configuration normalization
* Fix tests
---------
Co-authored-by: Matt Rubens <mrubens@users.noreply.github.com>
* Add xAI provider
* Add model reasoning effort
* DRY this up
* Handle undefined delta
* Cleanup getModel to fix test
* Add missing translations
* Small type cleanup
* Support temperature
---------
Co-authored-by: cte <cestreich@gmail.com>
* fix(api): update cacheReadsPrice for OpenAI GPT-4.1 models (#2887)
Set correct cacheReadsPrice (cached input price) for gpt-4.1, gpt-4.1 mini, and gpt-4.1 nano based on official OpenAI pricing. No changes to cacheWritesPrice as per current OpenAI documentation. This ensures prompt caching costs are accurately reflected for these models in cost calculations.
* Update more OpenAI cache prices
* Track cache tokens and cost correctly for OpenAI
* Update tests
---------
Co-authored-by: monotykamary <tom81094@gmail.com>
Fixes Bedrock ARN validation
Updates the Bedrock ARN regex to allow alphanumeric characters, dots, hyphens, and colons in the resource ID.
This prevents validation errors when using ARNs containing those characters.
