Skip to main content
The SDK auto-instruments these LLM providers with zero additional code:
  • OpenAI
  • Anthropic
  • Cohere
  • Azure OpenAI
  • Amazon Bedrock
  • Google Vertex AI
  • Replicate
  • HuggingFace
Both SDKs also auto-instrument the LangChain and LlamaIndex frameworks. Local models via Ollama and hosted OpenAI-compatible endpoints (Together, Fireworks, Groq, OpenRouter, self-hosted vLLM) work through the OpenAI SDK, and as of belvedir 0.4.0 (Node) / 0.6.0 (Python) also through raw fetch / requests / httpx calls to any OpenAI-compatible chat/completions endpoint, streamed responses included, with no client SDK at all. See Cloud Inference. Apps built on the AI SDK (the ai package) are captured through that same raw-HTTP path when calls use @ai-sdk/openai-compatible; its native provider packages (@ai-sdk/openai, @ai-sdk/anthropic) and built-in gateway routing are not captured. See the AI SDK guide. For agents you didn’t write (OpenClaw, Hermes, or anything that already speaks OpenTelemetry), see Agents without the SDK.