Skip to content
TapPass

Mistral, DeepSeek, Groq, Qwen, Moonshot, MiniMax

These six providers share one integration path: store the key once (per org, in the vault), then select the provider by prefixing the model name. No extra configuration, no separate endpoints.

Provider Model prefix (example) Platform env fallback
Mistral mistral/mistral-large-latest MISTRAL_API_KEY
DeepSeek deepseek/deepseek-chat DEEPSEEK_API_KEY
Groq groq/llama-3.3-70b-versatile GROQ_API_KEY
Qwen (Alibaba DashScope) dashscope/qwen-plus DASHSCOPE_API_KEY
Moonshot / Kimi moonshot/kimi-k2 MOONSHOT_API_KEY
MiniMax minimax/abab6.5-chat MINIMAX_API_KEY

Set the key

Section titled “Set the key”

Per organization, in the vault:

  • Dashboard → Settings → LLM providers, or
  • CLI: tappass model key set mistral (prompts for the key)

A platform-wide fallback key from the server environment is used only when the operator enables hybrid fallback (off by default — a missing org key blocks the call).

Call it

Section titled “Call it”
response = agent.chat("Summarize Q4 revenue", model="groq/llama-3.3-70b-versatile")

or zero-code with the OpenAI SDK pointed at the gateway — same model string, same governance.

Notes

Section titled “Notes”
  • Qwen / DashScope defaults to the international endpoint (dashscope-intl.aliyuncs.com); repoint with DASHSCOPE_API_BASE if your key is issued for another region. The same <PROVIDER>_API_BASE override pattern works for every provider on this page.
  • Provider display names (Qwen, Moonshot, …) come from the server’s canonical provider-label map — what you set here is what appears in dashboards, cost reports, and audit facets.

Next steps

Section titled “Next steps”