Free ModelScope API Key, Base URL & Rate Limits
Model ProviderModelScope is Alibaba's model hub and inference platform, providing free API access to Qwen, DeepSeek, and other models popular in the Chinese AI ecosystem.
How to get a free ModelScope API key
- 1
- 2 Go to Access Tokens
- 3 Create an access token
- 4 Choose a model Qwen3.5 and DeepSeek available. 2,000 RPD total.
- 5 Configure OpenAI client Base URL: https://api-inference.modelscope.cn/v1
Provider Snapshot
Supported Models 35 models
View in directory →| Model ID | Developer | Context | Availability | Free Tier | Use cases |
|---|---|---|---|---|---|
| Qwen/Qwen3.8-Flash-Next Qwen/Qwen3.8-Flash-Next | Alibaba | 8K | Online | Free tier | chat |
| Qwen/Qwen3.8-27B Qwen/Qwen3.8-27B | Alibaba | 8K | Online | Free tier | chat |
| deepseek-ai/DeepSeek-V4-Pro deepseek-ai/DeepSeek-V4-Pro | DeepSeek | 8K | Online | Free tier | chat |
| deepseek-ai/DeepSeek-V4-Pro-0813 deepseek-ai/DeepSeek-V4-Pro-0813 | DeepSeek | 8K | Online | Free tier | chat |
| deepseek-ai/DeepSeek-V4-Flash-0731 deepseek-ai/DeepSeek-V4-Flash | DeepSeek | 8K | Online | Free tier | chat |
| ZhipuAI/GLM-5.2 GLM-5.1 | Z AI | 1.0M | Online | Free tier | chat |
| deepseek-ai/DeepSeek-V4.1-Flash deepseek-ai/DeepSeek-V4.1-Flash | DeepSeek | 8K | Online | Free tier | chat |
| MiniMax/MiniMax-M3 MiniMax-M3 | MiniMax | 512K | Online | Free tier | chat |
| stepfun-ai/Step-3.7-Flash stepfun-ai/Step-3.7-Flash | StepFun | 8K | Online | Free tier | chat |
| Qwen/Qwen3.5-397B-A17B Qwen/Qwen3.5-397B-A17B | Alibaba | 8K | Online | Free tier | chat |
| nex-agi/Nex-N2.5-Pro nex-agi/Nex-N2.5-Pro | ModelScope | 8K | Online | Free tier | chat |
| Qwen/Qwen3.5-122B-A10B Qwen/Qwen3.5-122B-A10B | Alibaba | 8K | Online | Free tier | chat |
| Qwen/Qwen3.5-27B Qwen/Qwen3.5-27B | Alibaba | 8K | Online | Free tier | chat |
| Qwen/Qwen3.5-35B-A3B Qwen/Qwen3.5-35B-A3B | Alibaba | 8K | Online | Free tier | chat |
| stepfun-ai/Step-3.5-Flash stepfun-ai/Step-3.5-Flash | StepFun | 8K | Online | Free tier | chat |
| opencompass/CompassJudger-1-32B-Instruct opencompass/CompassJudger-1-32B-Instruct | ModelScope | 8K | Online | Free tier | chat |
| XGenerationLab/XiYanSQL-QwenCoder-32B-2412 XGenerationLab/XiYanSQL-QwenCoder-32B-2412 | ModelScope | 8K | Online | Free tier | chat |
| XGenerationLab/XiYanSQL-QwenCoder-32B-2504 XGenerationLab/XiYanSQL-QwenCoder-32B-2504 | ModelScope | 8K | Online | Free tier | chat |
| ZhipuAI/GLM-4.7-Flash GLM-4.7-FlashX | Z AI | 200K | Online | Free tier | chat |
| MedAIBase/AntAngelMed MedAIBase/AntAngelMed | ModelScope | 8K | Online | Free tier | chat |
| MusePublic/Qwen-Image-Edit MusePublic/Qwen-Image-Edit | ModelScope | 8K | Online | Free tier | chat |
| OpenGVLab/InternVL3_5-241B-A28B OpenGVLab/InternVL3_5-241B-A28B | ModelScope | 8K | Online | Free tier | chat |
| PaddlePaddle/ERNIE-4.5-21B-A3B-PT PaddlePaddle/ERNIE-4.5-21B-A3B-PT | ModelScope | 8K | Online | Free tier | chat |
| PaddlePaddle/ERNIE-4.5-300B-A47B-PT PaddlePaddle/ERNIE-4.5-300B-A47B-PT | ModelScope | 8K | Online | Free tier | chat |
| PaddlePaddle/ERNIE-4.5-VL-28B-A3B-PT PaddlePaddle/ERNIE-4.5-VL-28B-A3B-PT | ModelScope | 8K | Online | Free tier | chat |
| Qwen/Qwen-Image-Edit Qwen/Qwen-Image-Edit | ModelScope | 8K | Online | Free tier | chat |
| Shanghai_AI_Laboratory/Intern-S1 Shanghai_AI_Laboratory/Intern-S1 | ModelScope | 8K | Online | Free tier | chat |
| Shanghai_AI_Laboratory/Intern-S2-Preview Shanghai_AI_Laboratory/Intern-S2-Preview | ModelScope | 8K | Online | Free tier | chat |
| early-access/EA-29B-A4B early-access/EA-29B-A4B | ModelScope | 8K | Online | Free tier | chat |
| Shanghai_AI_Laboratory/Intern-S1-mini Shanghai_AI_Laboratory/Intern-S1-mini | ModelScope | 8K | Online | Free tier | chat |
| nex-agi/Nex-N2.5-mini nex-agi/Nex-N2.5-mini | ModelScope | 8K | Online | Free tier | chat |
| meituan-longcat/LongCat-Flash-Lite meituan-longcat/LongCat-Flash-Lite | LongCat | 8K | Online | Free tier | chat |
| MiniMax/MiniMax-M1-80k MiniMax/MiniMax-M1-80k | MiniMax | 8K | Online | Free tier | chat |
| PaddlePaddle/ERNIE-4.5-0.3B-PT PaddlePaddle/ERNIE-4.5-0.3B-PT | ModelScope | 8K | Online | Free tier | chat |
| mistralai/Mistral-Large-Instruct-2407 mistralai/Mistral-Large-Instruct-2407 | Mistral | 8K | Online | Free tier | chat |
Developer Tools
ModelScope FreeLLM Score free API access score
How we score →What is ModelScope?
Alibaba ModelScope — Qwen3.5, DeepSeek at 2,000 RPD free.
ModelScope is Alibaba's model hub and inference platform, providing free API access to Qwen, DeepSeek, and other models popular in the Chinese AI ecosystem. The free tier allows 2,000 requests/day total with a per-model cap of 500 RPD. Best latency for China-based users.
- 2,000 RPD total, 500 RPD per model
- Qwen3.5, DeepSeek available
- Best latency in China
- OpenAI-compatible endpoint
API Compatibility: Live-tested OpenAI-compatible Chat Completions and Responses
ModelScope Free Tier Limits & Pricing
ModelScope API Setup Tutorials
ModelScope is fully compatible with popular AI coding assistants like Cursor, Claude Code, and more. To see step-by-step API configuration instructions for your favorite tool, please visit our Global Configuration Guide →
ModelScope Model Use Cases
What ModelScope's free models are best for, based on aggregated model capabilities:
ModelScope Limitations & Caveats
- China-hosted — high latency outside Asia-Pacific
- 500 RPD per-model cap, 2,000 RPD total across all models
- Alibaba account or Chinese phone required
ModelScope FAQ
What's the difference between ModelScope and SiliconFlow?
ModelScope is Alibaba's official model hub with Qwen3.5 as the flagship. SiliconFlow is an independent platform. Both are China-hosted, but ModelScope has official Qwen models while SiliconFlow offers a broader mix including DeepSeek.
Does ModelScope have the latest Qwen models?
Yes — as Alibaba's own platform, ModelScope gets the latest Qwen releases (currently Qwen3.5) before they appear on other providers.
Can international users use ModelScope?
ModelScope requires an Alibaba account or Chinese phone for registration. International users may face difficulty signing up. Consider OpenRouter or Groq as alternatives for Qwen models.