Skip to content

feat(ai-gateway): add Nebius Token Factory provider integration#299

Open
Arindam200 wants to merge 1 commit intoHelicone:mainfrom
Arindam200:main
Open

feat(ai-gateway): add Nebius Token Factory provider integration#299
Arindam200 wants to merge 1 commit intoHelicone:mainfrom
Arindam200:main

Conversation

@Arindam200
Copy link
Copy Markdown

Summary

Adds comprehensive Nebius Token Factory support as a new inference provider with 30+ models, full monitoring, and model mapping capabilities.

What's Included

  • Provider Integration: Nebius with Meta-Llama, NVIDIA, and open-source models
  • Direct API Access: /nebius/v1/* endpoints with OpenAI-compatible format
  • Model Mapping: Nebius models as fallbacks in 50+ model mappings
  • Testing: Unit tests + integration tests with mock infrastructure
  • Environment: NEBIUS_API_KEY support and Docker integration

Testing

cargo +nightly test nebius  # Nebius-specific tests
cargo +nightly test         # All tests

Files Changed

  • config/embedded/providers.yaml - Added Nebius config
  • config/embedded/model-mapping.yaml - Added model mappings
  • src/types/provider.rs - Provider recognition
  • src/types/model_id.rs - Model ID parsing
  • src/router/router_details.rs - Routing support
  • tests/nebius.rs - Integration tests
  • stubs/nebius/ - Mock infrastructure
  • .env.template - API key template

Checklist

  • Provider configuration
  • Model ID parsing
  • Endpoint registration
  • Mock testing
  • Unit + integration tests
  • Environment setup
  • Code formatting
  • No breaking changes

- Add Nebius provider configuration with 30+ supported models
- Implement Nebius model ID parsing with slash support
- Add Nebius to provider registry and endpoint mappings
- Create comprehensive mock testing infrastructure
- Add unit tests for model ID parsing and routing
- Add integration tests for direct proxy functionality
- Update environment template with NEBIUS_API_KEY
- Add Nebius models as fallbacks in model mappings

The Nebius Token Factory integration follows existing provider patterns
and enables users to route requests through /nebius/v1/* endpoints
with full monitoring and rate limiting support.
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