API Web Search

Released support for web search via the API. API docs have been updated and a Postman Collection demonstrating the various calls and responses can be found here.

Deepseek Coder V2 Lite in Beta

Released Deepseek Coder V2 Lite in beta in both the API and the UI for feedback from our beta testers. If you’re interested in helping test, please join the #ai-code channel in Discord and reach out to our team.

Mistral Codestral 22B in Beta

Released Mistral Codestral 22B in beta in both the API and the UI for feedback from our beta testers. If you’re interested in helping test, please join the #ai-code channel in Discord and reach out to our team.

API Updates

• /image/generate - Fixed an issue with seed parameter on image generation not being fully random on every request.

• /image/generate - Updated API documentation to note that on the hide_watermark parameter, Venice may ignore this parameter for certain generated content.

• /image/generate - Add a request id field on the image generation JSON response. API docs are updated.

• image/upscale - Removed the previous dimension requirements on upscaled images. The API can now be used to upscale images of any dimension.

• /api/models - Beta API models are now returned in the model list endpoint. The docs have been updated.

• /api/models - Added a code filter to the type parameter on /api/models to filter models that are designed for code. The docs have been updated.

• Changed Qwen Coder API model ID to qwen-2.5-coder-32b. Ensured backwards compatibility using the Compatibility Mappings.

• Documentation for support for gzip and brotli compression has been added back to the API docs here and here. This applies to the JSON responses on /image/generate (when return_binary is false) and on /chat/completions (when stream is false).

App

• When uploading images for use with vision models or in-painting, the browser will now resize images to fit within the context of the model.