Quick Reference for All Providers

Cherry Studio built-in 60+ providers, this page provides an overview table; after finding the target provider just fill in the key according to the guide to use it. For providers with dedicated topic docs, there are jump links; the rest follow the general steps (Provider Overview) to configure.

Usage steps

1. Find the target provider(you can use Ctrl/⌘+F for quick search)
2. Click Official website Register an account and get an API key
3. In Cherry Studio Settings → Model Services find the corresponding provider, fill in the key, then click "Get model list"
4. Configuration complete

One-line decision

Your needs	Recommended direction
Quick start for beginners, avoiding complex processes	CherryIN or CherryAI
Most convenient for access within China	DeepSeek / Moonshot / SiliconFlow / Zhipu
Strongest models overseas	OpenAI / Anthropic / Gemini
One key for 200+ providers	OpenRouter
Fully local, privacy-sensitive	Ollama / LM Studio
Enterprise compliance	Azure OpenAI / AWS Bedrock
Use agent	Anthropic / CherryIN(must support the Anthropic protocol)

Self-operated models from major domestic tech companies

No VPN needed, Chinese-language advantage, relatively low prices.

Provider	One-line feature	Official website	Dedicated docs
DeepSeek	The king of cost-performance for coding and reasoning	deepseek.com	→
Moonshot AI (Kimi)	Ultra-long context (up to 2 million Chinese characters)	moonshot.cn	→
ZhiPu (Zhipu)	GLM series, multimodal, Anthropic-compatible for running agents	bigmodel.cn	→
doubao (Doubao/Volcano Engine)	Made by ByteDance, affordable pricing	volcengine.com	→
Baidu Cloud (Wenxin Yiyan)	Baidu ERNIE series	cloud.baidu.com	—
Bailian (Alibaba Bailian)	Qwen series, with a huge number of models	bailian.console.aliyun.com	→
BAICHUAN AI	Baichuan large model	baichuan-ai.com	—
MiniMax	Domestic multimodal (voice, video)	minimaxi.com	→
StepFun	StepFun	stepfun.com	—
LongCat	Meituan LongCat series	longcat.chat	—
Xiaomi MiMo	Xiaomi large model	mimo.mi.com	—

Self-operated models from major overseas tech companies

Top-tier performance; access from China usually requires a proxy.

Provider	One-line feature	Official website	Dedicated docs
OpenAI	GPT series	openai.com	→
Anthropic	Claude series, the top choice for agents	anthropic.com	→
Gemini (Google)	Google large models	aistudio.google.com	→
Azure OpenAI	Microsoft-hosted OpenAI, enterprise compliant	portal.azure.com	→
VertexAI	Hosted on Google Cloud	cloud.google.com	→
AWS Bedrock	Amazon-hosted models from multiple providers	aws.amazon.com/bedrock	—
Mistral	Representative European open-source models	mistral.ai	→
Grok (xAI)	Musk's xAI, with built-in web access	x.ai	→
Perplexity	search-augmented chat	perplexity.ai	→

Gateway / aggregation

One key accesses multiple models, with centralized account management.

Provider	One-line feature	Official website	Dedicated docs
CherryAI	Cherry official free trial	—	→
CherryIN	Cherry official paid gateway, dual endpoints (OpenAI + Anthropic)	open.cherryin.cc	→
OpenRouter	Largest overseas aggregator, 200+ models	openrouter.ai	→
AiHubMix	Overseas aggregator	aihubmix.com	—
DMXAPI	Domestic aggregator	dmxapi.cn	—
302.AI	Domestic aggregator	302.ai	—
NewAPI	Self-hosted gateway (open source)	newapi.pro	→
OneAPI	Self-hosted gateway (open source)	—	→
PPIO	Domestic cloud compute + models	ppio.com	→
BurnCloud	Domestic aggregator	burncloud.com	—
AIOnly	Domestic aggregator	aiionly.com	—
ocoolAI	Domestic aggregator	ocoolai.com	—
Poe	AI marketplace under Quora	poe.com	—
Vercel AI Gateway	Gateway under Vercel	vercel.com/ai	—

Ultra-low-latency / high-throughput inference service

Suitable for scenarios that need a sense of "speed" (IM bots, real-time translation, etc.).

Provider	One-line feature	Official website	Dedicated docs
Groq	LPU hardware, millisecond-level response	groq.com	→
Cerebras AI	Self-developed chips, ultra-large context	cerebras.ai	—
Together	Centralized hosting for open-source models	together.ai	—
Fireworks	Inference optimization for open-source models	fireworks.ai	—

Domestic cloud + compute services

Provider	One-line feature	Official website	Dedicated docs
Silicon (SiliconFlow)	China's largest open-source model hosting	siliconflow.cn	→
ModelScope (Mota)	Alibaba's open-source model platform	modelscope.cn	→
AlayaNew	Domestic inference service	alayanew.com	—
Qiniu (Qiniu)	Qiniu Cloud AI	qiniu.com	—
LANYUN	Domestic inference	lanyun.net	—
Xirang	Tianyi Cloud Xirang	ctyun.cn	—

Embeddings / reranking only

Used only for embeddings or reranking, for use with knowledge bases / global memory.

Provider	One-line feature	Official website	Dedicated docs
Jina	Embeddings, reranking, CLIP, generous free quota	jina.ai	—
VoyageAI	Embedding / reranking specialist	voyageai.com	—

Local inference

Completely offline, privacy-protecting.

Provider	One-line feature	Official website	Dedicated docs
Ollama	Local command-line inference, the most popular	ollama.com	→
LM Studio	GUI local inference, friendly to Apple Silicon	lmstudio.ai	→
GPUStack	Enterprise-grade local inference	gpustack.ai	—
OpenVINO Model Server	Intel-accelerated local inference	openvino.ai	—

Model platforms / others

Provider	One-line feature	Official website	Dedicated docs
Hugging Face	The world's largest open-source model community	huggingface.co	—
GitHub Copilot	Microsoft GitHub coding assistant	github.com/features/copilot	→
GitHub Models	GitHub Model Marketplace (Beta)	github.com/marketplace/models	—
MiniMax Global	MiniMax overseas version	minimax.io	—
SophNet	Domestic model hosting	sophnet.com	—
PH8	Domestic inference	ph8.co	—
Z.ai	Zhipu international version	z.ai	—
nvidia	NVIDIA NIM inference	nvidia.com	—

Custom providers

If the service you use is not in the list above, but provides OpenAI-compatible / Anthropic-compatible / Gemini-compatible any one of these protocols, you can add it via Custom providers Add it.

Still don't know which one to choose?

Just go with CherryIN or CherryAI — best for beginners to get started quickly. Switch later when you need something more advanced.

---

Get help and submit feedback

If you encounter any questions, bugs, or have suggestions for feature improvements during configuration or use, please refer to Feedback and Suggestions for the official channels provided.