Provider Matrix

Nimi Runtime routes AI requests through a unified API. Use this page as the single source for provider capability coverage and rollout status.

For onboarding and quickstart, prefer the high-level targeting surface:

• nimi run "<prompt>" for the local default text target
• nimi run "<prompt>" --provider <provider> for a provider default cloud target
• nimi run "<prompt>" --cloud for the saved machine default cloud target

This matrix documents advanced provider-qualified prefixes exposed by lower-level runtime and SDK surfaces.

GA/Beta/Planned on this page describe provider capability rollout inside current runtime contracts. They do not mean the entire Nimi platform is stable overall; Nimi is still in a rapid development phase.

Status Legend

• GA: available now for production usage in the current provider/runtime contract slice
• Beta: available now with limited or policy-gated modalities
• Planned: tracked for future delivery, not available in current runtime contracts

Capability Legend

• ✅ available now
• - not available in current runtime contract
• 🟡 planned

Local Engines

Legacy localai/, nexa/, and nimi_media/ namespaces are removed from the public runtime contract. Canonical local routing uses the built-in engine families below.

Engine	Runtime Target	Status	Text	Embed	Image	Video	TTS	STT	Notes
llama (llama.cpp / llama-server)	llama	GA	✅	✅	-	-	-	-	Supervised bootstrap downloads official ggml-org/llama.cpp release packs and runs llama-server
media	media	GA	-	-	✅	✅	-	-	Nimi-managed local image/video generation engine
speech	speech	GA	-	-	-	-	✅	✅	Nimi-managed local speech engine family

Cloud Providers

Provider	SDK Prefix	Status	Text	Embed	Image	Video	TTS	STT	Notes
OpenAI	openai/	GA	✅	✅	-	-	-	-
Anthropic	anthropic/	GA	✅	-	-	-	-	-
Google Gemini	gemini/	GA	✅	✅	✅	✅	✅	✅
DeepSeek	deepseek/	GA	✅	-	-	-	-	-
OpenRouter	openrouter/	GA	✅	-	-	-	-	-
OpenAI-Compatible (BYO endpoint)	openai_compatible/	GA	✅	-	-	-	-	-	Supports Ollama, vLLM, LM Studio, LiteLLM, Xinference, and similar endpoints
Alibaba DashScope	dashscope/	GA	✅	✅	✅	✅	✅	✅
Volcengine ARK (Doubao)	volcengine/	GA	✅	✅	✅	✅	-	-
Volcengine OpenSpeech	volcengine_openspeech/	GA	-	-	-	-	✅	✅
MiniMax	minimax/	GA	✅	-	✅	✅	✅	✅
Kimi (Moonshot)	kimi/	GA	✅	-	✅	-	✅	✅
GLM (Zhipu)	glm/	GA	✅	-	✅	✅	✅	✅
Azure OpenAI	azure/	GA	✅	✅	-	-	-	-
Mistral AI	mistral/	GA	✅	✅	-	-	-	-
Groq	groq/	Beta	✅	-	-	-	-	🟡	STT planned
xAI (Grok)	xai/	GA	✅	-	-	-	-	-
Baidu Qianfan (ERNIE)	qianfan/	Beta	✅	✅	🟡	-	🟡	🟡
Tencent Hunyuan	hunyuan/	Beta	✅	✅	🟡	🟡	🟡	🟡
iFlytek Spark	spark/	Beta	✅	-	-	-	🟡	🟡
AWS Bedrock	TBD	Planned	🟡	🟡	🟡	-	-	-
Cohere	TBD	Planned	🟡	🟡	-	-	-	-
Together AI	TBD	Planned	🟡	🟡	🟡	-	-	-
Replicate	TBD	Planned	🟡	-	🟡	🟡	-	-
ElevenLabs	TBD	Planned	-	-	-	-	🟡	🟡
Baichuan AI	TBD	Planned	🟡	🟡	-	-	-	-
Yi (01.AI)	TBD	Planned	🟡	-	-	-	-	-
Step AI	TBD	Planned	🟡	-	🟡	🟡	-	-
Perplexity AI	TBD	Planned	🟡	-	-	-	-	-
Stability AI	TBD	Planned	-	-	🟡	🟡	-	-
AssemblyAI	TBD	Planned	-	-	-	-	-	🟡
Runway	TBD	Planned	-	-	-	🟡	-	-

Validation In Your Environment

nimi provider list
nimi doctor
nimi provider test gemini
nimi run "Hello from Nimi" --provider gemini
nimi provider set gemini --api-key-env NIMI_RUNTIME_CLOUD_GEMINI_API_KEY --default
nimi run "Hello from Nimi" --cloud

Runnable provider examples: examples/sdk/providers/