Starstruk Logo MarkStarstruk

v1.7.7

Portkey-AI/gatewayv1.7.7Oct 1, 2024by VisargD

AI Summary

A maintenance release focusing on provider compatibility and bug fixes. It includes support for a new Mistral AI /fim/completions route, adds missing logprob mapping for Azure OpenAI, and improves chat templating for Bedrock-hosted models.

Key Highlights

  • Support for Mistral AI /fim/completions route
  • Missing logprob mapping for Azure OpenAI
  • Enhanced chat templating for Llama and Mistral on Bedrock
  • Default models added for Gemini, Together AI, and Fireworks AI
  • Vertex embeddings support added

New Features

  • Mistral AI /fim/completions route
  • Azure OpenAI logprob mapping
  • Bedrock templating improvements
  • Default models for Gemini, Together AI, Fireworks AI
  • Vertex embeddings support

Full Release Notes

## What's Changed
* chore: upgrade portkey mustache package version to 2.1.2 by @VisargD in https://github.com/Portkey-AI/gateway/pull/611
* Chore/support additional params cohere embed by @narengogi in https://github.com/Portkey-AI/gateway/pull/614
* Add missing logprob mapping for Azure OpenAI provider by @rootAvish in https://github.com/Portkey-AI/gateway/pull/618
* Added Support for the new /fim/completions route on Mistral AI API by @keshavkrishna in https://github.com/Portkey-AI/gateway/pull/467
* fix: chat templating for llama and mistral hosted on bedrock by @narengogi in https://github.com/Portkey-AI/gateway/pull/619
* Add contextual embedding cookbook by @siddharthsambharia-portkey in https://github.com/Portkey-AI/gateway/pull/627
* webhook plugin header parsing logic to allow objects by @VisargD in https://github.com/Portkey-AI/gateway/pull/632
* Ends With Plugin by @shivam-pareek in https://github.com/Portkey-AI/gateway/pull/607
* fix: added max_completion_tokens for chat complete by @shivam-pareek in https://github.com/Portkey-AI/gateway/pull/617
* Chore: Debug configs for vscode by @narengogi in https://github.com/Portkey-AI/gateway/pull/602
* fix: send Portkey as user-agent by @b4s36t4 in https://github.com/Portkey-AI/gateway/pull/609
* Add default models for gemini, together ai, fireworks ai by @vrushankportkey in https://github.com/Portkey-AI/gateway/pull/630
* Upstage.ai by @shivam-pareek in https://github.com/Portkey-AI/gateway/pull/608
* fix chat templating for llama hosted on bedrock by @narengogi in https://github.com/Portkey-AI/gateway/pull/623
* improvement: Add vertex embeddings support by @elentaure in https://github.com/Portkey-AI/gateway/pull/622
* Fix media file prompting for vertex gemini and google gemini by @narengogi in https://github.com/Portkey-AI/gateway/pull/635
* feat: add json mode support for vertex-ai by @VisargD in https://github.com/Portkey-AI/gateway/pull/638
* Chore: better error handling by @VisargD in https://github.com/Portkey-AI/gateway/pull/640
* Feat: bedrock cross region model ids by @VisargD in https://github.com/Portkey-AI/gateway/pull/641
* chore: update plugin manifest with ends with handler by @VisargD in https://github.com/Portkey-AI/gateway/pull/645
* 1.7.7 by @VisargD in https://github.com/Portkey-AI/gateway/pull/646

## New Contributors
* @rootAvish made their first contribution in https://github.com/Portkey-AI/gateway/pull/618
* @keshavkrishna made their first contribution in https://github.com/Portkey-AI/gateway/pull/467
* @shivam-pareek made their first contribution in https://github.com/Portkey-AI/gateway/pull/607
* @elentaure made their first contribution in https://github.com/Portkey-AI/gateway/pull/622

**Full Changelog**: https://github.com/Portkey-AI/gateway/compare/v1.7.6...v1.7.7