SiliconFlow
China-based inference platform offering three permanently free open-weight models alongside 200+ paid models.
- Requests / day
- 1K
- Requests / min
- 30
- Card required
- No
SiliconFlow scores 4.7/10 for free value in LLM. The main constraint: 3 permanently free models (Qwen3-8B, DeepSeek-R1-Distill-Qwen-7B, DeepSeek-OCR) at 30 RPM / 60K TPM, capped at 50 requests/day; a one-time lifetime purchase of >=10 CNY raises the cap to 1,000/day.
Free value score
4.7 / 10Show the math
Key metrics
Models
2 free · 200+ totalLimits may be stale — last re-verified outside this section's freshness window.
| Model | Family | AA | Access |
|---|---|---|---|
| Qwen3-8B | Qwen | 15 | free |
| Qwen2.5-7B-Instruct | Qwen | — | free |
| GLM-5.2 | GLM | 51.1 | paid |
| MiniMax-M3 | MiniMax | 44.4 | paid |
| DeepSeek-V4-Pro | DeepSeek | 44.3 | paid |
| Kimi-K2.6 | Kimi | 42.8 | paid |
| Kimi-K2.7-Code | Kimi | 42 | paid |
| DeepSeek-V4-Flash | DeepSeek | 40.3 | paid |
| GLM-5.1 | GLM | 40.2 | paid |
| GLM-5 | GLM | 39.5 | paid |
| Kimi-K2.5 | Kimi | 38.1 | paid |
| GLM-5V-Turbo | GLM | 34.5 | paid |
| GLM-4.7 | GLM | 33.8 | paid |
| Qwen3.5-122B-A10B | Qwen | 32.3 | paid |
| Qwen3.5-397B-A17B | Qwen | 32 | paid |
| Qwen3.6-35B-A3B | Qwen | 31.6 | paid |
| gemma-4-26B-A4B-it | Gemma | 25.7 | paid |
| DeepSeek-V3.2 | DeepSeek | 24.7 | paid |
| DeepSeek-V3.2-Exp | DeepSeek | 24.7 | paid |
| gpt-oss-120b | GPT | 23.8 | paid |
| DeepSeek-V3.1 | DeepSeek | 21.1 | paid |
| DeepSeek-V3.1-Terminus | DeepSeek | 21.1 | paid |
| DeepSeek-R1 | DeepSeek | 20.1 | paid |
| GLM-4.5-Air | GLM | 16.5 | paid |
| gpt-oss-20b | GPT | 14.9 | paid |
| DeepSeek-V3 | DeepSeek | 10.4 | paid |
| gemma-4-31B-it | Gemma | — | paid |
| GLM-4.1V-9B-Thinking | GLM | — | paid |
| Hunyuan-A13B-Instruct | Hunyuan | — | paid |
| Ling-flash-2.0 | Ling | — | paid |
| MiniMax-M2.5 | MiniMax | — | paid |
| Qwen2.5-72B-Instruct | Qwen | — | paid |
| Qwen3-14B | Qwen | — | paid |
| Qwen3-30B-A3B-Instruct-2507 | Qwen | — | paid |
| Qwen3-32B | Qwen | — | paid |
| Qwen3-Coder-30B-A3B-Instruct | Qwen | — | paid |
| Qwen3-Coder-480B-A35B | Qwen | — | paid |
| Qwen3-VL-32B-Instruct | Qwen | — | paid |
| Seed-OSS-36B-Instruct | Seed | — | paid |
| Step-3.5-Flash | Step | — | paid |
Showing first 2 of a larger catalog.
What you get
Three genuinely permanent free models
Includes a free DeepSeek-OCR vision endpoint
OpenAI SDK-compatible
Optional tiny lifetime purchase lifts daily cap 20x
Gotchas
No gotchas documented yet. Contribute one if you know a constraint we missed.
Sources
-
"SiliconFlow 🇨🇳 ... Qwen/Qwen3-8B 131K 131K Text 30 RPM, 60K TPM ; deepseek-ai/DeepSeek-R1-Distill-Qwen-7B 131K Configurable Text (reasoning) 30 RPM, 60K TPM. Permanently free models, no credit card required. 200+ paid models also available."
https://github.com/mnfst/awesome-free-llm-apis/blob/main/README.md -
"Free models default to 200 RPD per model. A one-time purchase of $10+ in credits unlocks 1,000 RPD for free models. ... (this footnote is under the OpenRouter section, NOT SiliconFlow)"
https://github.com/mnfst/awesome-free-llm-apis/blob/main/README.md -
"免费模型：1 Qwen/Qwen3-8B; 2 THUDM/GLM-Z1-9B-0414; 3 THUDM/GLM-4-9B-0414; 4 deepseek-ai/DeepSeek-R1-Distill-Qwen-7B; 5 deepseek-ai/DeepSeek-R1-0528-Qwen3-8B; 6 Qwen/Qwen2.5-7B-Instruct; ... 17 deepseek-ai/DeepSeek-OCR ... (新人注册可用免费模型 — ~20 models)"
https://zhuanlan.zhihu.com/p/2000257698017608153 -
"Qwen3 8B (Reasoning) scores 7 on the Artificial Analysis Intelligence Index, placing it below average among comparable models (averaging 9)."
https://artificialanalysis.ai/models/qwen3-8b-instruct-reasoning -
"注册福利：注册即送 2000万 Tokens（约合14元平台配额）; 有效期：永久有效 ... 备注：部分小模型标价0元永久免费"
https://github.com/guihuashaoxiang/FreeLLM-API-KeyHub -
"The platform, with over three million users and billions of token calls processed daily, provides developers with competitive pricing"
https://en.tmtpost.com/post/7463667
Alternatives
Same category · similar quota
Pre-paid credits routed to 300+ models at provider list prices; credits never expire; Kimi K2 Thinking is $0.60/$2.50 per 1M on OpenRouter; 1M free BYOK requests/mo then 5% BYOK fee
