Integrations¶
Connect your LLM providers and agent frameworks to Tokensor.
Provider keys¶
Store a provider API key on your account (onboarding or Settings) and Tokensor injects it upstream when you route calls through the transparent proxies:
- OpenAI — route
gpt-*calls throughhttps://tokensor.dev/proxy/openai/v1. - Anthropic — route
claude-*calls throughhttps://tokensor.dev/proxy/anthropic/v1. - OpenCode — route Zen/Go calls through
https://tokensor.dev/proxy/opencode/{zen|go}/v1. See OpenCode.
Observability platforms¶
- Langfuse — configure in Settings to export traces and sync findings.
- Helicone — configure in Settings to import existing call data.
Agent frameworks¶
Connect agent frameworks via OpenTelemetry to track costs per agent and detect runaway loops, redundant tool calls, and context bloat. See Agent Frameworks.
Bring your own vendor¶
Tokensor supports OpenAI, Anthropic, and OpenCode out of the box. If you want another vendor, see Adding a Vendor for how the platform is designed to be extended.