Skip to content

feat(provider): add Kyma API OpenAI-compatible provider#1600

Open
sonpiaz wants to merge 1 commit intoPortkey-AI:mainfrom
sonpiaz:feat/add-kyma-provider
Open

feat(provider): add Kyma API OpenAI-compatible provider#1600
sonpiaz wants to merge 1 commit intoPortkey-AI:mainfrom
sonpiaz:feat/add-kyma-provider

Conversation

@sonpiaz
Copy link
Copy Markdown

@sonpiaz sonpiaz commented Apr 11, 2026

Description

  • Added Kyma API to the global provider registry (src/globals.ts)
  • Implemented OpenAI-compatible chat completion and streaming support
  • Added provider-specific API configuration (base URL, auth headers)
  • Included a usage notebook demonstrating Portkey + Kyma API integration

About Kyma API

Kyma API is an LLM gateway providing access to 20+ open-source and open-weight models through a single OpenAI-compatible endpoint:

  • 20+ models: DeepSeek V3/R1, Qwen 3, LLaMA 3.3 70B, Gemma 4, Kimi K2, and more
  • 4-layer fallback routing for high availability (multi-provider, never-die architecture)
  • Standard OpenAI /v1/chat/completions endpoint with streaming (SSE)
  • Tool calling support on most models
  • Free credits on signup, pay-per-token after

Tests Run/Test cases added

  • Manual verification via OpenAI client with provider="kyma" header
  • Streaming responses validated against live endpoint

Type of Change

  • Bug fix (non-breaking change which fixes an issue)
  • New feature (non-breaking change which adds functionality)
  • Breaking change (fix or feature that would cause existing functionality to not work as expected)
  • Documentation update
  • Refactoring (no functional changes)

- Added Kyma to the global provider registry (src/globals.ts)
- Implemented OpenAI-compatible chat completion and streaming support
- Added provider-specific API configuration (base_url, auth headers)
- Included a usage notebook demonstrating Portkey + Kyma API integration

Kyma API (kymaapi.com) is an LLM gateway with 20+ open-source models
(DeepSeek, Qwen, LLaMA, Gemma, Kimi, and more) via a single endpoint
with 4-layer fallback routing for high availability.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant