Text and chat

Send an ordered messages array and a live text model ID. temperature and either max_tokens or max_completion_tokens are supported; send only one token-limit field. Grok, DeepSeek V4.1 Flash, GLM 5.3, and GLM 5.3 Flash accept OpenAI-compatible tools, tool_choice, parallel_tool_calls, and response_format. Grok 4.5 supports reasoning_effort as low, medium, or high; GLM 5.3 and GLM 5.3 Flash support low, high, or max. Papaya 1.0 (modelrush/papaya-1.0) is for text chat and streaming. It is not available for tool or function calling, or in ModelRush Agent. A Papaya request that includes tools, tool_choice, parallel_tool_calls, functions, or function_call, even when empty or false, or tool-call history (tool or function role messages, tool_calls, tool_call_id, or function_call) returns 400 unsupported_parameter with the offending param before any charge. It also rejects image parts, response_format, and reasoning_effort with 400 before any charge, and caps either token-limit field at 8,192. Omit stream for a JSON completion, or set stream: true for OpenAI-compatible server-sent events; stream_options.include_usage is accepted. Compatibility clients may send store: false.

bash
curl https://api.modelrush.ai/v1/chat/completions \
  -H "Authorization: Bearer $MODELRUSH_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "modelrush/deepseek-v4.1-flash",
    "messages": [{"role": "user", "content": "Return exactly: ModelRush connected"}]
  }'

bash
curl https://api.modelrush.ai/v1/chat/completions \
  -H "Authorization: Bearer $MODELRUSH_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"modelrush/deepseek-v4.1-flash","messages":[{"role":"user","content":"Stream a short answer"}],"stream":true}'

The response follows the Chat Completions shape with id, model, choices, and usage. Persist the response ID for support and reconciliation.