DeepSeek
DeepSeek
DeepSeek's API models, tier by tier: DeepSeek V4 Pro and DeepSeek V4.1 Flash, with official prices and benchmark scores.
This page compares the API models behind DeepSeek. The DeepSeek app may use other models or different limits.
DeepSeek V4 Pro · top
US$ 1.32 / US$ 3.96 peak hoursInput US$/1M / Output US$/1M · Official pricing
DeepSeek V4.1 Flash · budget
US$ 0.30 / US$ 1.20 peak hoursInput US$/1M / Output US$/1M · Official pricing
DeepSeek V4.1 Flash · open
US$ 0.30 / US$ 1.20 peak hoursInput US$/1M / Output US$/1M · Official pricing
01
Brand tiers
| tier | Current model | Reference model for scores | GPQA Diamond | FrontierMath (tiers 1 to 3) | SimpleQA Verified | Input US$/1M |
|---|---|---|---|---|---|---|
| top | DeepSeek V4 Pro | DeepSeek V4 Pro | 90.9% between 34th and 60th of 250 | 45.3% between 56th and 58th of 85 | 47% between 36th and 41st of 87 | US$ 1.32 / US$ 3.96 Official pricing peak hours |
| budget | DeepSeek V4.1 Flash | DeepSeek V4.1 Flash | 89.8% between 46th and 71st of 250 | 67.4% between 33rd and 35th of 85 | 42.9% between 46th and 48th of 87 | US$ 0.30 / US$ 1.20 Official pricing peak hours |
| open | DeepSeek V4.1 Flash | DeepSeek V4.1 Flash | 89.8% between 46th and 71st of 250 | 67.4% between 33rd and 35th of 85 | 42.9% between 46th and 48th of 87 | US$ 0.30 / US$ 1.20 Official pricing peak hours |
02
Version history
Compare with
How to read
- Each row is a DeepSeek tier: top, mid, economy and budget, from most to least expensive. «Open» sits apart and describes the weights, not the price.
- «Current model» is the one listed on DeepSeek's official pricing page.
- When the current model has no published score yet, the tier's scores come from the «reference model for scores», the newest version in the line with a score.
- Prices are in US$ per 1M tokens and carry the source label: «Official lab price», «Third-party provider» or «No current price».
- Every number comes with its source and date. «No data» means no measurement exists in the open sources this section uses.
Sources
- Epoch AI, CC BY 4.0, latest version we checked.
- OpenRouter Benchmarks, CC BY 4.0, latest version we checked.
- OpenRouter, dados públicos da API; exports de benchmarks e rankings em CC BY 4.0, latest version we checked.
Dates come from the sources: OpenRouter listing or Hugging Face repo creation, which may differ from the official launch. Family and version were grouped by rule from the name.
How to cite
If you use this data, cite the page and the date you accessed it.
Adriano Meirinho. "DeepSeek". meirinho.com.br. https://meirinho.com.br/en/ia/modelos/marca/deepseek/. Accessed November 21, 2026.
Third-party data follows the license of each source. The full table is available as CSV and JSON.