DeepSeek V4 Pro
DeepSeek-V4 series incorporate several key upgrades in architecture and optimization: (1) a hybrid attention architecture that combines Compressed Sparse Attention (CSA) and Heavily Compressed Attention (HCA) to improve long-context efficiency; (2) ManifoldConstrained Hyper-Connections (mHC) that enhance conventional residual connections; (3) and the Muon optimizer for faster convergence and greater training stability.
text → text · made by DeepSeek
$1.32→$3.96per Mtok in / out
DeepSeek · 21 sellers
Sold by 54 ways
| Seller | Lane | Rate |
|---|---|---|
| Writer api | standard | $0.44per Mtok in$0.87per Mtok outdev.writer.com · read 2026-08-25 |
| DigitalOcean cloud | standard | $0.87per Mtok in$1.74per Mtok out$0.17per Mtok cachedmodels.dev · read 2026-09-16 |
| DeepSeek api | standard | $1.32per Mtok in$3.96per Mtok out$0.044per Mtok cachedapi-docs.deepseek.com · read 2026-08-24 |
| Microsoft Azure AI cloud | standard | $1.74per Mtok in$3.48per Mtok out$0.14per Mtok cachedmodels.dev · read 2026-09-16 |
| Alibaba api | standard | $2.4per Mtok in$4.8per Mtok outalibabacloud.com · read 2026-08-24 |
| Hugging Face Inference Providers 4 ways | $0.43per Mtok in$0.87per Mtok out$0.0036per Mtok cached | |
| Hugging Face Inference Providers aggregator | standard | $0.43per Mtok in$0.87per Mtok out$0.0036per Mtok cachedmodels.dev · read 2026-09-16 |
| Hugging Face Inference Providers aggregator | deepinfra | $1.3per Mtok in$2.6per Mtok outrouter.huggingface.co · read 2026-08-25 |
| Hugging Face Inference Providers aggregator | novita | $1.6per Mtok in$3.2per Mtok outrouter.huggingface.co · read 2026-08-25 |
| Hugging Face Inference Providers aggregator | baseten | $1.74per Mtok in$3.48per Mtok outrouter.huggingface.co · read 2026-08-25 |
| Hugging Face Inference Providers aggregatornot on the seller's list today | fireworks-ai | $1.74per Mtok in$3.48per Mtok outrouter.huggingface.co · read 2026-08-25 |
| Hugging Face Inference Providers aggregatornot on the seller's list today | together | $1.74per Mtok in$3.48per Mtok outrouter.huggingface.co · read 2026-08-25 |
| Vercel AI Gateway aggregator | standard | $0.66per Mtok in$1.98per Mtok out$0.022per Mtok cachedvercel.com · read 2026-08-25 |
| Vercel AI Gateway aggregatornot on the seller's list today | sandbox | $0.66per Mtok in$1.98per Mtok outvercel.com · read 2026-08-25 |
| Requesty 0 ways | $0.94per Mtok in$1.87per Mtok out$0.072per Mtok cached | |
| Requesty aggregatornot on the seller's list today | deepinfra flex | $0.94per Mtok in$1.87per Mtok out$0.072per Mtok cachedrouter.requesty.ai · read 2026-08-25 |
| Requesty aggregatornot on the seller's list today | deepinfra | $1.04per Mtok in$2.08per Mtok out$0.08per Mtok cachedrouter.requesty.ai · read 2026-08-30 |
| Requesty aggregatornot on the seller's list today | doubleword flex | $1.18per Mtok in$2.48per Mtok outrouter.requesty.ai · read 2026-08-25 |
| Requesty aggregatornot on the seller's list today | doubleword | $1.31per Mtok in$2.75per Mtok outrouter.requesty.ai · read 2026-08-30 |
| Requesty aggregatornot on the seller's list today | fireworks | $1.74per Mtok in$3.48per Mtok out$0.15per Mtok cachedrouter.requesty.ai · read 2026-08-28 |
| Requesty aggregatornot on the seller's list today | nebius | $1.75per Mtok in$3.5per Mtok out$1.75per Mtok cachedrouter.requesty.ai · read 2026-08-28 |
| Requesty aggregatornot on the seller's list today | tensorx | $1.75per Mtok in$3.5per Mtok out$0.44per Mtok cachedrouter.requesty.ai · read 2026-08-28 |
| Nous Research aggregator | standard | $0.94per Mtok in$1.89per Mtok out$0.08per Mtok cachedinference-api.nousresearch.com · read 2026-09-16 |
| OpenRouter 16 ways | $1.6per Mtok in$3.2per Mtok out$0.14per Mtok cached | |
| OpenRouter aggregator | streamlake/fp8 | $0.96per Mtok in$1.91per Mtok out$0.08per Mtok cachedopenrouter.ai · read 2026-09-15 |
| OpenRouter aggregator | gmicloud/fp8 | $0.96per Mtok in$1.91per Mtok out$0.08per Mtok cachedopenrouter.ai · read 2026-09-08 |
| OpenRouter aggregator | digitalocean | $1.04per Mtok in$2.09per Mtok out$0.21per Mtok cachedopenrouter.ai · read 2026-09-15 |
| OpenRouter aggregatornot on the seller's list today | ionstream/fp4 | $1.13per Mtok in$2.26per Mtok out$0.094per Mtok cachedopenrouter.ai · read 2026-08-25 |
| OpenRouter aggregator | cloudflare | $1.15per Mtok in$2.55per Mtok out$0.2per Mtok cachedopenrouter.ai · read 2026-09-14 |
| OpenRouter aggregatornot on the seller's list today | coreweave/fp8 | $1.15per Mtok in$2.55per Mtok out$0.2per Mtok cachedopenrouter.ai · read 2026-08-25 |
| OpenRouter aggregator | deepinfra/fp8 | $1.3per Mtok in$2.6per Mtok out$0.1per Mtok cachedopenrouter.ai · read 2026-08-25 |
| OpenRouter aggregator | alibaba/fp8 | $1.42per Mtok in$2.83per Mtok out$0.12per Mtok cachedopenrouter.ai · read 2026-08-25 |
| OpenRouter aggregator | siliconflow/fp8 | $1.5per Mtok in$3.13per Mtok out$0.14per Mtok cachedopenrouter.ai · read 2026-08-25 |
| OpenRouter aggregator | standard | $1.6per Mtok in$3.2per Mtok out$0.14per Mtok cachedopenrouter.ai · read 2026-09-13 |
| OpenRouter aggregator | novita/fp8 | $1.6per Mtok in$3.2per Mtok out$0.14per Mtok cachedopenrouter.ai · read 2026-08-25 |
| OpenRouter aggregator | venice | $1.65per Mtok in$3.3per Mtok out$0.33per Mtok cachedopenrouter.ai · read 2026-08-25 |
| OpenRouter aggregator | atlas-cloud/fp4 | $1.68per Mtok in$3.38per Mtok out$0.13per Mtok cachedopenrouter.ai · read 2026-08-25 |
| OpenRouter aggregator | baidu/fp8 | $1.69per Mtok in$3.38per Mtok out$0.14per Mtok cachedopenrouter.ai · read 2026-09-14 |
| OpenRouter aggregator | baseten/fp4 | $1.74per Mtok in$3.48per Mtok out$0.14per Mtok cachedopenrouter.ai · read 2026-08-25 |
| OpenRouter aggregatornot on the seller's list today | fireworks | $1.74per Mtok in$3.48per Mtok out$0.14per Mtok cachedopenrouter.ai · read 2026-08-25 |
| OpenRouter aggregator | parasail/fp8 | $1.74per Mtok in$3.48per Mtok out$0.1per Mtok cachedopenrouter.ai · read 2026-08-25 |
| OpenRouter aggregatornot on the seller's list today | together | $1.74per Mtok in$3.48per Mtok out$0.2per Mtok cachedopenrouter.ai · read 2026-08-25 |
| OpenRouter aggregator | azure/us | $1.91per Mtok in$3.83per Mtok out$0.16per Mtok cachedopenrouter.ai · read 2026-08-25 |
| OpenRouter aggregator | nextbit/fp8 | $2.8per Mtok in$5.5per Mtok out$0.6per Mtok cachedopenrouter.ai · read 2026-09-16 |
| Fireworks AI 2 ways | $1.74per Mtok in$3.48per Mtok out$0.14per Mtok cached | |
| Fireworks AI aggregator | priority | $1.2per Mtok in$1.2per Mtok out$0.6per Mtok cachedraw.githubusercontent.com · read 2026-09-16 |
| Fireworks AI aggregator | standard | $1.74per Mtok in$3.48per Mtok out$0.14per Mtok cacheddocs.fireworks.ai · read 2026-08-24 |
| DeepInfra aggregator | standard | $1.3per Mtok in$2.6per Mtok out$0.1per Mtok cachedapi.deepinfra.com · read 2026-08-25 |
| Avian aggregator | standard | $1.3per Mtok in$2.61per Mtok out$0.11per Mtok cachedapi.avian.io · read 2026-09-16 |
| ElectronHub aggregator | standard | $1.5per Mtok in$3per Mtok outapi.electronhub.ai · read 2026-09-16 |
| Novita AI aggregator | standard | $1.6per Mtok in$3.2per Mtok out$0.14per Mtok cachedapi.novita.ai · read 2026-08-25 |
| Venice AI aggregator | standard | $1.65per Mtok in$3.3per Mtok out$0.33per Mtok cachedmodels.dev · read 2026-09-16 |
| Baseten aggregator | standard | $1.74per Mtok in$3.48per Mtok out$1.74per Mtok cachedbaseten.co · read 2026-08-25 |
| SiliconFlow aggregator | standard | $1.74per Mtok in$3.48per Mtok out$0.14per Mtok cachedmodels.dev · read 2026-09-16 |
| Together AI aggregator | standard | $1.74per Mtok in$3.48per Mtok out$0.2per Mtok cacheddocs.together.ai · read 2026-08-24 |
| Nebius Token Factory aggregator | standard | $1.75per Mtok in$3.5per Mtok outtokenfactory.nebius.com · read 2026-08-25 |
| Inference.net aggregator | standard | $2.1per Mtok in$4.2per Mtok out$0.0036per Mtok cachedapi.inference.net · read 2026-09-16 |
| PPIO aggregator | standard | $9per Mtok in$27per Mtok outapi.ppinfra.com · read 2026-09-16 |
Measured 29 standings
| Place | Board | Metric | Score |
|---|---|---|---|
| 6thof 32 | AIME 2026 | Accuracy (± 95% CI) (max) | 96.67 |
| 6thof 35 | Swe bench verified — Epoch AI | mean_score (max) | 77.64 |
| 9thof 17 | Cl bench life — Epoch AI | Overall (high) | 0.135 |
| 14thof 17 | SWE-rebench Leaderboard (default window 15/05/2026–01/07/2026, 111 problems) | Resolved Rate (%) (high) | 40.2 |
| 21stof 128 | LMArena · WebDev | Score (Elo) (high) | 1581 |
| 21stof 26 | Surface evolver bench — Epoch AI | Mean score (high) | 0.4 |
| 25thof 30 | Frontiercode — Epoch AI | Main score (none) | 0.176 |
| 26thof 290 | OTIS Mock AIME 2024-2025 | mean_score (max) | 96.667 |
| 28thof 161 | Dtbench — Epoch AI | Accuracy (max) | 90.67 |
| 28thof 46 | LMArena · Agent | Net Improvement (%) (high) | 4.14 |
| 30thof 80 | Simpleqa verified — Epoch AI | mean_score (max) | 46.994 |
| 33rdof 313 | GPQA Diamond | mean_score (high) | 90.909 |
| 33rdof 123 | Lmca — Epoch AI | Score (max) | 41.2 |
| 39thof 60 | Vending bench 2 — Epoch AI | Score | 3284.52 |
| 42ndof 112 | Ale bench — Epoch AI | Performance (high) | 1006.08 |
| 43rdof 106 | Frontiermath tiers 1 3 v2 — Epoch AI | mean_score (max) | 45.263 |
| 43rdof 64 | Proofbench — Epoch AI | Accuracy (max) | 16 |
| 43rdof 105 | Vectara Hallucination Leaderboard | Hallucination Rate | 8.6 |
| 50thof 402 | LMArena · Text | Score (Elo) (high) | 1463 |
| 52ndof 121 | Webdev arena — Epoch AI | Arena Score (high) | 1463.8 |
| 56thof 62 | Frontiermath tier 4 v2 — Epoch AI | mean_score (max) | 2.439 |
| 57thof 177 | Critpt — Epoch AI | Accuracy (max) | 12.857 |
| 60thof 168 | Scicode — Epoch AI | Score (max) | 50 |
| 65thof 302 | Artificial Analysis · Intelligence Index | Artificial Analysis Intelligence Index (max) | 31 |
| 70thof 127 | Mystery game puzzles — Epoch AI | mean_score (none) | 17 |
| 72ndof 170 | Weirdml — Epoch AI | Accuracy (max) | 48.9 |
| 74thof 81 | Forecastbench — Epoch AI | Overall score | 55.3 |
| 90thof 222 | Chess puzzles — Epoch AI | mean_score (max) | 20 |
| 242ndof 656 | Epoch capabilities index — Epoch AI | ECI Score (max) | 149.2 |
About
DeepSeek V4 Pro — a text model from DeepSeek, sold by 21 companies from $0.43 in and $0.87 out per million tokens, placed 26th of 290 on OTIS Mock AIME 2024-2025.
It takes text and returns text, with a context window of 1,000,000 tokens. It was published in April 2026, trained on material up to May 2025. Its sellers say it can reason step by step and call a tool. The catalogue files it under extract, reasoning, code and agents. 21 companies sell it. The cheapest is $0.43 in and $0.87 out per million tokens at Hugging Face Inference Providers. Beside the standard rate there are faster, separately routed and regional lanes. It has been measured on 29 boards, and stands best at 26th of 290 on OTIS Mock AIME 2024-2025.
Every current figure
- Maker
- DeepSeek
- Register
- model
- Takes
- text
- Returns
- text
- Context
- 1,000,000 tokens
- Published
- April 2026
- Knowledge to
- May 2025
- Parameters
- 1599 billion
- Licence
- not read
- Sellers
- 21
- Price
- $0.43 in and $0.87 out per million tokens — Hugging Face Inference Providers
- Boards
- 29
- Best place
- 26th of 290 — OTIS Mock AIME 2024-2025
Known as 16 names
DeepSeek-v4-Pro (Max)DeepSeek v4 (max)DeepSeek V4 Pro (High) (0813)deepseek-v4-pro-high-20260813DeepSeek-V4 Pro [high]deepseek-v4-proV4 Prodeepseek-ai/DeepSeek-V4-Profireworks/deepseek-v4-prodeepseek/deepseek-v4-prodeepinfra/deepseek-ai/DeepSeek-V4-Prodeepinfra/deepseek-ai/DeepSeek-V4-Pro:flexdoubleword/deepseek-v4-prodoubleword/deepseek-v4-pro:flexnebius/deepseek-ai/deepseek-v4-protensorx/deepseek-v4-pro