One API, every model
A drop-in OpenAI-compatible endpoint for chat, embeddings, vision, tools and streaming — across every provider, on one bill.
Product · Blossom
One key. Every model.
Blossom is a token-aware API gateway for every major LLM — OpenAI, Anthropic, Google, Mistral, Meta, DeepSeek and 300+ open-source models behind a single OpenAI-compatible endpoint. Change the base URL, keep your SDK, and route across the whole frontier with one key and one bill.
What it does
Three primitives, every model behind them.
Smart routing & fallback
Declarative policies pick the right model per request by cost, latency or capability. When one provider 429s, the next takes over mid-stream.
0% token markup
Pay provider list price exactly; the platform fee is separate. Semantic caching finds the cheapest model that still passes your eval.
99.99% gateway uptime
Multi-provider, multi-region failover the instant a model degrades, so your app keeps shipping through provider incidents.
Token-level tracing
Per-key budgets and full request tracing with prompt and completion bodies. Export to your warehouse or stream to OpenTelemetry.
Compliance by design
No-train by default, regional routing, bring-your-own provider keys, and SOC 2 · GDPR · ISO 27001 on Enterprise.
Zyan Zealous product