Skip to main content

SiliconFlow

CNInference host

SiliconFlow runs open and licensed models on its own hardware and sells them by usage, often faster or cheaper than the maker.

Other region:SiliconFlowGlobal

PricingDocscontact@siliconflow.com
Visit siliconflow.cn
Models59
Cheapest for
15 models
Serves
5 model types
Visits in September 20261.5M−34.9%Unique visitors 494K
  1. Apr
  2. May
  3. Jun
  4. Jul
  5. Aug
  6. Sep
Models by type

Models on SiliconFlow

ModelInputOutputvs officialContext
Qwen3-VL Embedding 8BCheapest—$0.097/1M——
Qwen3-VL-32B-InstructOnly here$0.139/1M$0.556/1M——
Qwen3-VL-32B-ThinkingOnly here$0.139/1M$1.39/1M——
Qwen3-VL-Reranker-8BOnly here—$0.097/1M——
Qwen3.5 122B-A10B$0.278/1M$2.22/1M
from $0.901 at AIHubMix
-31%—
Seed Oss 36B Instruct$0.208/1M$0.556/1M
from $0.534 at AIHubMix
+4%—
Step 3.5 FlashCheapest$0.097/1M$0.292/1M——
Wan2.2-I2V-A14BCheapest—$0.278/video——
Wan2.2-T2V-A14BOnly here—$0.278/video——

SiliconFlow at a glance

59 models in total: 45 text, 4 image, 4 embedding, 4 reranking, 2 video. The cheapest is Qwen3 Embedding 0.6B at $0.01 /1M. LongCat-2.0 has the largest context window, 1.05M tokens. The most searched is GLM-5.3, with about 276K Google searches a month. The newest is DeepSeek V4.1 Flash, released Sep 10, 2026.

Cheapest

  1. Qwen3 Embedding 0.6B$0.01/1M
  2. Qwen3 Reranker 0.6B$0.01/1M
  3. Qwen 3 Embedding 4B$0.019/1M
  4. Qwen3-Reranker-4B$0.019/1M
  5. Qwen3 Embedding 8B$0.039/1M

Largest context

  1. LongCat-2.01.05M
  2. GLM-5.31.05M
  3. DeepSeek V4.1 Flash1.05M
  4. DeepSeek V4 Flash1.05M
  5. DeepSeek V4 Pro1.05M

Most searched

  1. GLM-5.3276K
  2. DeepSeek V4.1 Flash138K
  3. Qwen3.8 27B120K
  4. GLM-5.2109K
  5. DeepSeek V4 Flash51K

Newest

  1. DeepSeek V4.1 FlashSep 10, 2026
  2. Hy4 previewAug 27, 2026
  3. GLM-5.3Aug 25, 2026
  4. Qwen3.8 27BAug 5, 2026
  5. LongCat-2.0Jul 5, 2026