Skip to the form

APIs, 60% off list price

We’re a Singapore company selling the leading Chinese models. Your OpenAI or Anthropic SDK keeps working — swap base_url and the key. One USD invoice, and you never sign with a China-based entity.

Get 100M free tokens to test

Any text model, valid for 30 days, one per company. We email your key and pricing, usually the same day.

Two short steps. Key and pricing by email, usually the same day.

Step 1 of 2
Step 1 of 2: email and company
You are
DeepSeek-V4-Pro
$1.32$0.53
$3.96$1.58
DeepSeek-V4-Flash
$0.44$0.18
$1.32$0.53
DeepSeek-V3.2
$0.28$0.11
$0.42$0.17
USD per 1M tokensSee all 66 models
0.00TTokens served
0BTokens per day
0Businesses and developers
0%Uptime SLA

Pricing proof

All 66 Chinese models, up to 60% off list price

Qwen, DeepSeek, Kimi, GLM and MiniMax — every model, 60% off list price. Other text models are 40% off; other image and video models 10% to 20%. Image and video discounts are smaller because the vendors’ own margins are.

Type

66 of 66 models

List prices from each provider’s official pricing page  ·  Last check: 2026-09-04

ModelList priceSufy priceSavings
Text models37
Kimi-K3moonshotai/kimi-k3Moonshot (Kimi)
List$3.00$15.00
Sufy$1.20$6.00
−60%
Kimi-K2.7-Codemoonshotai/kimi-k2.7-codeMoonshot (Kimi)
List$0.95$4.00
Sufy$0.38$1.60
−60%
Kimi-K2.6moonshotai/kimi-k2.6Moonshot (Kimi)
List$0.95$4.00
Sufy$0.38$1.60
−60%
Qwen3-Next-80B-A3B-Thinkingqwen3-next-80b-a3b-thinkingAlibaba (Qwen)
List$0.15$1.20
Sufy$0.06$0.48
−60%
Qwen3.5-397B-A17Bqwen3.5-397b-a17bAlibaba (Qwen)
List$0.60$3.60
Sufy$0.24$1.44
−60%
Qwen3.5-Plusqwen/qwen3.5-plusAlibaba (Qwen)
List$0.40$2.40
Sufy$0.16$0.96
−60%
Qwen3.6-27Bqwen/qwen3.6-27bAlibaba (Qwen)
List$0.60$3.60
Sufy$0.24$1.44
−60%
Qwen3.6-Plusqwen/qwen3.6-plusAlibaba (Qwen)
List$0.50$3.00
Sufy$0.20$1.20
−60%
Qwen3.7-Maxqwen/qwen3.7-maxAlibaba (Qwen)
List$2.50$7.50
Sufy$1.00$3.00
−60%
Qwen3.7-Plusqwen/qwen3.7-plusAlibaba (Qwen)
List$0.40$1.60
Sufy$0.16$0.64
−60%
Qwen3.8 Flashqwen/qwen3.8-flash-nextAlibaba (Qwen)
List$0.15$0.47
Sufy$0.06$0.19
−60%
Qwen3.8-Maxqwen/qwen3.8-maxAlibaba (Qwen)
List$2.00$6.00
Sufy$0.80$2.40
−60%
Doubao-Seed-2.0-Codedoubao-seed-2.0-codeByteDance (Doubao/Seed)
List$0.50$3.00
Sufy$0.30$1.80
−40%
Doubao-Seed-2.0-Litedoubao-seed-2.0-liteByteDance (Doubao/Seed)
List$0.25$2.00
Sufy$0.15$1.20
−40%
Doubao-Seed-2.0-Minidoubao-seed-2.0-miniByteDance (Doubao/Seed)
List$0.10$0.40
Sufy$0.06$0.24
−40%
Doubao-Seed-2.0-Prodoubao-seed-2.0-proByteDance (Doubao/Seed)
List$0.50$3.00
Sufy$0.30$1.80
−40%
Doubao-Seed-2.1-probytedance/doubao-seed-2-1-proByteDance (Doubao/Seed)
List$0.87$4.35
Sufy$0.52$2.61
−40%
Doubao-Seed-2.1-turbobytedance/doubao-seed-2-1-turboByteDance (Doubao/Seed)
List$0.50$2.50
Sufy$0.30$1.50
−40%
DeepSeek-V3deepseek-v3DeepSeek
List$0.290$1.159
Sufy$0.116$0.464
−60%
DeepSeek-V3.2deepseek/deepseek-v3.2-251201DeepSeek
List$0.28$0.42
Sufy$0.11$0.17
−60%
DeepSeek-V4-Flashdeepseek/deepseek-v4-flashDeepSeek
List$0.44$1.32
Sufy$0.18$0.53
−60%
DeepSeek-V4-Flash-20260731deepseek/deepseek-v4-flash-20260731DeepSeek
List$0.44$1.32
Sufy$0.18$0.53
−60%
DeepSeek-V4-Flash-Vision-Expdeepseek/deepseek-v4-flash-vision-expDeepSeek
List$0.44$1.32
Sufy$0.18$0.53
−60%
DeepSeek-V4-Prodeepseek/deepseek-v4-proDeepSeek
List$1.32$3.96
Sufy$0.53$1.58
−60%
DeepSeek-V4-Pro-0813deepseek/deepseek-v4-pro-0813DeepSeek
List$1.32$3.96
Sufy$0.53$1.58
−60%
MiniMax-M1MiniMax-M1MiniMax
List$0.58$2.32
Sufy$0.23$0.93
−60%
MiniMax-M2.5minimax/minimax-m2.5MiniMax
List$0.30$1.20
Sufy$0.12$0.48
−60%
MiniMax-M2.5-Highspeedminimax/minimax-m2.5-highspeedMiniMax
List$0.60$2.40
Sufy$0.24$0.96
−60%
MiniMax-M2.7minimax/minimax-m2.7MiniMax
List$0.30$1.20
Sufy$0.12$0.48
−60%
MiniMax-M3minimax/minimax-m3MiniMax
List$0.30$1.20
Sufy$0.12$0.48
−60%
Step-3.7-Flashstepfun/step-3.7-flashStepFun
List$0.20$1.17
Sufy$0.12$0.70
−40%
Hy4 Previewtencent/hy4-previewTencent (Hunyuan)
List$0.83$2.50
Sufy$0.50$1.50
−40%
GLM-5z-ai/glm-5Z.ai (Zhipu GLM)
List$1.00$3.20
Sufy$0.40$1.28
−60%
GLM-5.1z-ai/glm-5.1Z.ai (Zhipu GLM)
List$1.40$4.40
Sufy$0.56$1.76
−60%
GLM-5.2z-ai/glm-5.2Z.ai (Zhipu GLM)
List$1.40$4.40
Sufy$0.56$1.76
−60%
GLM-5.3z-ai/glm-5.3Z.ai (Zhipu GLM)
List$1.40$4.40
Sufy$0.56$1.76
−60%
GLM-5.3-Flashz-ai/glm-5.3-flashZ.ai (Zhipu GLM)
List$0.15$0.50
Sufy$0.06$0.20
−60%
Image models5
Kling-Image-O1kling-image-o1Kuaishou (Kling)
List$0.028per image
Sufy$0.022per image
−20%
Kling-V1kling-v1Kuaishou (Kling)
List$0.0035per image
Sufy$0.0028per image
−20%
Kling-V1-5kling-v1-5Kuaishou (Kling)
List$0.014per image
Sufy$0.011per image
−20%
Kling-V2kling-v2Kuaishou (Kling)
List$0.014per image
Sufy$0.011per image
−20%
Kling-V2-Newkling-v2-newKuaishou (Kling)
List$0.0145per image
Sufy$0.0116per image
−20%
Video models24
Doubao-seedance-2-0-fast-260128bytedance/doubao-seedance-2-0-fast-260128ByteDance (Doubao/Seed)
List$3.30per 1M video tokens
Sufy$2.97per 1M video tokens
−10%
Doubao-seedance-2-0-mini-260615bytedance/doubao-seedance-2-0-mini-260615ByteDance (Doubao/Seed)
List$2.10per 1M video tokens
Sufy$1.89per 1M video tokens
−10%
Doubao-seedance-2-5-260628bytedance/doubao-seedance-2-5-260628ByteDance (Doubao/Seed)
List$6.40per 1M video tokens
Sufy$5.76per 1M video tokens
−10%
Doubao-Seedance-2.0bytedance/doubao-seedance-2-0-260128ByteDance (Doubao/Seed)
List$4.30per 1M video tokens
Sufy$3.87per 1M video tokens
−10%
Dreamina-seedance-2-0-260128byteplus/dreamina-seedance-2-0-260128BytePlus (ByteDance)
List$4.30per 1M video tokens
Sufy$3.87per 1M video tokens
−10%
Dreamina-seedance-2-0-fast-260128byteplus/dreamina-seedance-2-0-fast-260128BytePlus (ByteDance)
List$3.30per 1M video tokens
Sufy$2.97per 1M video tokens
−10%
Dreamina-seedance-2-0-mini-260615byteplus/dreamina-seedance-2-0-mini-260615BytePlus (ByteDance)
List$2.10per 1M video tokens
Sufy$1.89per 1M video tokens
−10%
Dreamina-seedance-2-5-260628byteplus/dreamina-seedance-2-5-260628BytePlus (ByteDance)
List$6.40per 1M video tokens
Sufy$5.76per 1M video tokens
−10%
Kling-3.0-Turbokling-video/kling-3.0-turboKuaishou (Kling)
List$0.11per second
Sufy$0.09per second
−20%
Kling-v2-1kling-v2-1Kuaishou (Kling)
List$0.056per second
Sufy$0.045per second
−20%
Kling-V2-5-Turbokling-v2-5-turboKuaishou (Kling)
List$0.042per second
Sufy$0.034per second
−20%
Kling-V2-6kling-v2-6Kuaishou (Kling)
List$0.070per second
Sufy$0.056per second
−20%
Kling-V3kling-v3Kuaishou (Kling)
List$0.084per second
Sufy$0.067per second
−20%
Kling-V3-Omnikling-v3-omniKuaishou (Kling)
List$0.084per second
Sufy$0.067per second
−20%
Kling-Video-O1kling-video-o1Kuaishou (Kling)
List$0.084per second
Sufy$0.067per second
−20%
MiniMax-H3minimax/minimax-h3MiniMax
List$0.130per second
Sufy$0.052per second
−60%
Minimax-H3-Maxminimax/minimax-h3-maxMiniMax
List$0.05per second
Sufy$0.02per second
−60%
Viduq1viduq1Shengshu (Vidu)
List$0.080per second
Sufy$0.064per second
−20%
Viduq2viduq2Shengshu (Vidu)
List$0.05per second
Sufy$0.04per second
−20%
Viduq2-Providuq2-proShengshu (Vidu)
List$0.040per second
Sufy$0.032per second
−20%
Viduq2-Turboviduq2-turboShengshu (Vidu)
List$0.030per second
Sufy$0.024per second
−20%
Viduq3vidu/viduq3Shengshu (Vidu)
List$0.035per second
Sufy$0.028per second
−20%
Viduq3-Providuq3-proShengshu (Vidu)
List$0.045per second
Sufy$0.036per second
−20%
Viduq3-Turboviduq3-turboShengshu (Vidu)
List$0.035per second
Sufy$0.028per second
−20%
Units
USD per million tokens. Image and video models price per output unit rather than per token.
Discount
Qwen, DeepSeek, Kimi, GLM and MiniMax 60% off list price; other text models 40%; other image and video 10% to 20%.
Rates
Headline prices are the standard on-demand input and output. Cache-hit, cache-write and batch rates are under All rates on each row.
CNY lists
Where a provider publishes only in CNY, we convert at 6.9 CNY per USD.

Migration

You change two values in the OpenAI or Anthropic SDK code you already have, and that’s the whole migration.

from openai import OpenAI
-client = OpenAI(api_key="sk-your-current-key")
+client = OpenAI(api_key="sk-your-sufy-key",
+ base_url="https://api.sufy.com/v1")

Your model names don’t change, and neither does anything you’re doing with streaming or tool calls.

Objections

Data handling

We don’t train on your prompts or outputs, and we don’t store their content. The open-weight models — DeepSeek, Qwen, GLM, Kimi, MiniMax — run on Sufy, so your content stays with us. Closed models work differently: a call to ByteDance Seed, Seedance, Seedream, or Kling reaches that vendor’s own API. What we do keep is the metadata billing needs, meaning which model you called and how many tokens it used, with a timestamp.

The entity you contract with

Your contract is with Superstify Technology Pte. Ltd., a Singapore company that’s the international brand of Qiniu Cloud. Nothing you sign involves a China-based entity.

Regions and latency

Requests land on Sufy nodes in North America — US West, US Central, US East, and Canada. If you want measured latency for the models you’re planning to run, ask and we’ll send the numbers.

Billing

You can pay in USD by credit card, ACH, or wire transfer. Everything you used lands on one invoice, broken out per model so your finance team doesn’t have to guess.

Get 100M free tokens to test

Any text model, valid for 30 days, one per company. We email your key and pricing, usually the same day.

Step 1 of 2
Step 1 of 2: email and company
You are

FAQ

Do I need a Sufy account to get pricing?

No. Fill in the form and we’ll email your rate and a test key with 100M free tokens. You’d only create an account once you’ve decided to go ahead.

What’s included in the free tokens?

100M tokens on any text model we serve, good for 30 days from the day we send your key. It’s one allowance per company. Image and video models aren’t included.

What happens when a model provider ships a new version?

We track each provider’s release schedule and retire versions on the timeline they do. You’ll hear from us before a retirement date, not after it.

Can I use my existing OpenAI or Anthropic SDK code?

Yes — the API speaks both protocols. Change base_url and api_key and your existing calls keep working, streaming and tool calls included.

What is covered by the SLA?

99% monthly uptime on the inference API, with service credits when we don’t hit it. Under 99% but at 95% or above, you get a 10% credit. Under 95%, you get 20%. Credits are issued as token balance, and the full terms come with your contract. Running in production since March 2025.