Skip to main content
deepseek.health
OpenAI-compat

Are the OpenAI-compat providers up?

Drop-in providers speaking /v1/chat/completions

Live reachability for the drop-in OpenAI-compatible LLM providers — Together AI, Fireworks AI and Groq — all speaking /v1/chat/completions.

OpenAI-compat API Status

Live, in-browser checks against OpenAI-compat's public APIs. Refreshes every 15s.

Checking…

OpenAI-compatible endpoints

All expose /v1/chat/completions with a bearer token

4 endpoints
  • Together AI
    api.together.xyz/v1
    —ms
    warming up
  • Fireworks AI
    api.fireworks.ai/inference/v1
    —ms
    warming up
  • Groq
    api.groq.com/openai/v1
    —ms
    warming up
  • OpenAI (reference)
    api.openai.com/v1
    —ms
    warming up
Polling every 15s · Last updated — · Next in 15sOpen OpenAI-compat ↗

The OpenAI-compatible endpoints all require a bearer token. These checks confirm DNS/TCP/TLS reachability only — a green dot does not mean your key is valid or your model is available.

Sponsored · OpenAI-compat
Ask Otto
/v1/chat

I'm Otto — pick the right OpenAI-compatible provider. Together, Fireworks, Groq, DeepSeek, Mistral… ask me anything.

Open DeepSeek

Sponsored content. deepseek.health may earn a referral reward when you sign up via this widget. Answers are informational only — not investment or engineering advice.

About OpenAI-compat

"OpenAI-compatible" is now a de facto standard: Together, Fireworks, Groq, DeepSeek, Mistral and many others expose an /v1/chat/completions endpoint that speaks the same wire format as OpenAI's API. Point the OpenAI SDK's base_url at any of them and swap models freely.

OpenAI-compat at a glance

Wire protocol
OpenAI /v1/chat/completions
Providers tracked
Together, Fireworks, Groq, OpenAI
Swap cost
1 line: base_url + model name

Common OpenAI-compat issues

  • Model name drift

    Each provider names the same open-weight model differently (e.g. meta-llama/Llama-3.3-70B vs accounts/fireworks/models/llama-v3p3-70b). Keep a mapping table.

  • Streaming SSE quirks

    Some providers send an extra trailing [DONE] frame or omit the usage block. Test each provider's stream shape before shipping.

Frequently asked

Which provider should I default to?
Groq for lowest latency (Llama, Mixtral, DeepSeek on LPUs), Together and Fireworks for widest open-weight model catalog, DeepSeek for reasoning at the lowest per-token price.
Does OpenAI's SDK really work with all of them?
Yes for chat completions, streaming, and JSON mode. Advanced OpenAI-only features (Assistants API, Realtime, some tool-use variants) are not portable.

About this check

All data is provided on a best-effort basis with no guarantees about accuracy or availability. This is an unofficial status page and is not affiliated with, endorsed by, or sponsored by OpenAI-compat.

Checks run client-side from your browser against OpenAI-compat's public endpoints. Latency reflects your network path, not the server's health alone. See our resources for more on how to read these checks.