NovitaAI is a hosted AI infrastructure provider with an OpenAI-compatible model API. In OpenClaw it is a bundled model provider, so the provider id isDocumentation Index
Fetch the complete documentation index at: https://docs2.openclaw.ai/llms.txt
Use this file to discover all available pages before exploring further.
novita, credentials go through the normal model auth flow, and model refs look
like novita/deepseek/deepseek-v3-0324.
Use Novita when you want hosted access to open-weight and third-party model
routes without running your own inference server. The bundled catalog focuses on
chat models that are practical for agent turns, including DeepSeek, Moonshot,
MiniMax, GLM, and Qwen routes exposed by Novita.
This provider uses Novita’s OpenAI-compatible endpoint. OpenClaw handles
provider registration, auth, aliases, model ref normalization, and base URL
selection; Novita controls live model availability, account permissions,
pricing, and rate limits.
Setup
Create an API key at novita.ai/settings/key-management, then run:Defaults
- Provider:
novita - Aliases:
novita-ai,novitaai - Base URL:
https://api.novita.ai/openai/v1 - Env var:
NOVITA_API_KEY - Default model:
novita/deepseek/deepseek-v3-0324
When to choose Novita
- You want hosted open-weight model access with an OpenAI-compatible API.
- You want DeepSeek, Kimi, MiniMax, GLM, or Qwen-family routes through a single provider account.
- You want another hosted fallback path beside OpenRouter, GMI, DeepInfra, or direct vendor APIs.
- You prefer provider-side model hosting over maintaining vLLM, SGLang, LM Studio, or Ollama infrastructure.
Models
The bundled catalog seeds commonly available NovitaAI route ids, including:novita/moonshotai/kimi-k2.5novita/minimax/minimax-m2.7novita/zai-org/glm-5novita/deepseek/deepseek-v3-0324novita/deepseek/deepseek-r1-0528novita/qwen/qwen3-235b-a22b-fp8
Troubleshooting
401or403: verify the key in Novita’s key management page and re-runopenclaw onboard --auth-choice novita-api-keyif the stored profile is stale.- Unknown model errors: use the exact
novita/<route-id>returned byopenclaw models list --provider novita. - Slow or failed routes: try another Novita model route or set Novita as a fallback provider for workloads that can tolerate provider-specific variance.