DeepSeek API pricing
DeepSeek API prices per million tokens, context windows, open-weights status and independent benchmark scores.
Prices updated . Benchmarks updated .
Prices are the current OpenRouter rate, refreshed about hourly. This page does not chart older prices.
| Model | You send / 1M | It replies / 1M | Context | ECI | Launched |
|---|---|---|---|---|---|
| DeepSeek V4.1 Flash 1017Open weights | $0.150 | $0.600 | 1.05M | 158.8 | 2026-10-17 |
| DeepSeek V4.1 FlashOpen weights | $0.045 | $0.500 | 1M | 156.1 | 2026-09-10 |
| DeepSeek V4 Pro 0813Open weights | $0.660 | $1.98 | 1M | 155.9 | 2026-08-13 |
| DeepSeek V4 Flash 0731Open weights | $0.119 | $0.238 | 1M | 154.9 | 2026-07-31 |
| DeepSeek V4 Pro 0423Open weights | $1.60 | $3.20 | 1.05M | 149.1 | 2026-08-13 |
| DeepSeek V4 Flash 0423Open weights | $0.091 | $0.182 | 1.05M | 146.2 | 2026-10-17 |
| DeepSeek V3.2Open weights | $0.260 | $0.380 | 128K | 146.1 | 2025-12-01 |
| DeepSeek V3.2 ExpOpen weights | $0.270 | $0.410 | 131K | 145.0 | 2025-01-01 |
| DeepSeek V3.1Open weights | $0.250 | $0.950 | 128K | 139.8 | 2025-08-21 |
| DeepSeek V3Open weights | $0.257 | $1.03 | 128K | 132.3 | 2025-12-01 |
| DeepSeek V4 Flash Vision ExpOpen weights | $0.216 | $0.647 | 1.05M | n/a | 2026-10-17 |
| DeepSeek V3.1 TerminusClosed | $0.270 | $1.00 | 128K | n/a | 2025-09-22 |
| R1 0528Open weights | $0.500 | $2.15 | 128K | n/a | 2025-05-28 |
| DeepSeek V3 0324Open weights | $0.290 | $1.14 | 128K | n/a | 2025-03-24 |
Which DeepSeek model ranks highest?
DeepSeek V4.1 Flash 1017 has the highest Epoch Capabilities Index among DeepSeek models on this page, at 158.8. DeepSeek V4.1 Flash is next at 156.1. DeepSeek V4.1 Flash 1017 costs $0.600 per million output tokens and reads 1.05M of context. The index is an independent score from Epoch AI. Lab-reported numbers are kept off this ranking.
Which DeepSeek model is cheapest?
DeepSeek V4 Flash 0423 has the lowest published output price among DeepSeek models here, at $0.182 per million tokens. Input tokens cost $0.091 per million and the context window is 1.05M. DeepSeek V4 Flash 0731 is next at $0.238 per million output tokens. Its Epoch Capabilities Index is 146.2. Prices are the current public rate and can change.
Which DeepSeek open-weight models lead?
DeepSeek V4.1 Flash 1017 leads DeepSeek open-weight models that have a published Epoch Capabilities Index, at 158.8. It is from DeepSeek, costs $0.600 per million output tokens, and reads 1.05M. DeepSeek V4.1 Flash from DeepSeek follows at 156.1. Open weights means the lab published the weights. The license on the model page says what you can do with them.
FAQ
Which DeepSeek model ranks highest?
DeepSeek V4.1 Flash 1017 has the highest Epoch Capabilities Index among DeepSeek models on this page, at 158.8. DeepSeek V4.1 Flash is next at 156.1. DeepSeek V4.1 Flash 1017 costs $0.600 per million output tokens and reads 1.05M of context. The index is an independent score from Epoch AI. Lab-reported numbers are kept off this ranking.
Which DeepSeek model is cheapest?
DeepSeek V4 Flash 0423 has the lowest published output price among DeepSeek models here, at $0.182 per million tokens. Input tokens cost $0.091 per million and the context window is 1.05M. DeepSeek V4 Flash 0731 is next at $0.238 per million output tokens. Its Epoch Capabilities Index is 146.2. Prices are the current public rate and can change.
How many DeepSeek models are listed?
This page lists 14 DeepSeek models with a public price or an independent score. Open the comparison table to filter the full catalogue.
Sources and licenses
- OpenRouter: Prices, context windows, modalities and the model list come from the OpenRouter public API. Data: OpenRouter.
- Vercel AI Gateway: The no-training flag comes from the Vercel AI Gateway model list.
- models.dev: Lab launch dates, open-weights flags and knowledge cutoffs come from models.dev. Copyright (c) 2025 models.dev, MIT License.
- Epoch AI: Independent benchmark scores and the Epoch Capabilities Index (ECI) come from Epoch AI's Benchmarking Hub, used under CC BY 4.0. Epoch AI, Capabilities & benchmarking, epoch.ai/benchmarks. Scores Epoch marks as external keep their original license, and the source is named on each score.
- LMArena: Arena Elo (text, webdev and vision) comes from the LMArena leaderboard dataset (lmarena-ai/leaderboard-dataset), CC BY 4.0, leaderboard date 2026-10-28.
- Lab-reported scores, when an editor has added them, are the lab's own numbers. Each one links to the source and is labeled lab-reported. They are not used to rank the lists.
- What a score measures: the benchmark guide.