DeepSeekDeepSeek V3.1
DeepSeek V3.1 is a DeepSeek open-weights reasoning model, released 21 Aug 2025. Its best configuration (thinking reasoning effort) ranks #435 of 785 on the BenchLeader Index at 48.1 ±3.5, in the lower half. The ± is the point: 296 other configurations score within that range, so they and this one cannot be told apart on quality alone — price and speed are what separate them. It scores highest in agents & tools (56) and lowest in coding (42). At $0.855 per million tokens blended it is mid-priced. Output speed of 120 tokens per second puts it faster than most, with a first answer in 19.4 s. Last measured 21 Aug 2025.stale: no new result in six months
- Blended price
- $0.855/M
- $0.580 in · $1.68 out
- Output speed
- 120 tok/s
- measured by Artificial Analysis
- First answer
- 19 s
- Context
- 128k
- Full answer
- 24 s
- median, reasoning included
- Cost per run
- –
- Released
- 21 Aug 2025
- Overall index48
- Reasoning44
- Coding42
- Agents & tools56
- Knowledge48
- Instruction following44
- Long context53
- Composite45
Versions
DeepSeek has shipped 2 models under this name. Each is ranked on its own results; a newer version often has fewer results so far, which holds its index nearer the average until more arrive.
| Model | Released | Index | Rank |
|---|---|---|---|
| DeepSeek V3.2 | 1 Dec 2025 | 56.6 | #197 |
| DeepSeek V3.1thinkingthis page | 21 Aug 2025 | 48.1 | #435 |
How its index has moved
9 Sept 2026 to 23 Oct 2026The index is recomputed from scratch every day, so a line moves when a new benchmark result lands, when a publisher revises a score, or when the models it is normalised against change. Early movement usually means the score is still settling.
Benchmark results
See all LLM benchmarks. One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.
Reasoning
| Benchmark | thinking | Source |
|---|---|---|
| GPQA Diamond (AA)not in index | 77.9%#216 | Artificial Analysis |
| Humanity's Last Exam (AA)not in index | 14.3%#258 | Artificial Analysis |
| CritPt | 2.0%#201 | Artificial Analysis |
| LMCAnot in index | 24.3%#151 | Epoch AI Benchmarking Hub |
| DTBenchnot in index | 82.7%#117 | Epoch AI Benchmarking Hub |
Agents & tools
| Benchmark | thinking | Source |
|---|---|---|
| Terminal-Bench Hard | 25.0%#138 | Artificial Analysis |
| τ²-Bench Telecom (AA)not in index | 37.4%#226 | Artificial Analysis |
Knowledge
| Benchmark | thinking | Source |
|---|---|---|
| AA-Omniscience | -29.6#278 | Artificial Analysis |
| AA-Omniscience: accuracynot in index | 29.0%#192 | Artificial Analysis |
| AA-Omniscience: non-hallucinationnot in index | 17.5%#313 | Artificial Analysis |
Instruction following
| Benchmark | thinking | Source |
|---|---|---|
| IFBench | 41.5%#246 | Artificial Analysis |
Long context
| Benchmark | thinking | Source |
|---|---|---|
| AA-LCR | 56.7%#294 | Artificial Analysis |
Composite
| Benchmark | thinking | Source |
|---|---|---|
| AA Intelligence Index v4.3.2 | 13.5#307 | Artificial Analysis |
Price history
The price above, per 1M tokens, every day it changed.
What a task costs
Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.0007 | – | 21.9 s |
| Summarize a 30-page report | 12,000 / 600 | $0.0080 | – | 24.4 s |
| Code edit | 6,000 / 1,500 | $0.0060 | – | 31.8 s |
| Agentic coding session | 60,000 / 4,000 | $0.042 | – | 52.6 s |
| Structured extraction | 2,000 / 200 | $0.0015 | – | 21.0 s |
More from DeepSeek
See also
Questions people ask
- What are DeepSeek V3.1's benchmark scores?
- As of 25 Oct 2026, DeepSeek V3.1 (thinking effort) scores 48.1 ±3.5 on the BenchLeader Index, strongest in agents & tools (56) and weakest in coding (42). That rests on 7 independent benchmark results, and every score on this page links to its publisher.
- How does DeepSeek V3.1 rank?
- DeepSeek V3.1 (thinking effort) is #435 of 785 ranked configurations on 25 Oct 2026. 296 other configurations score within its margin, so the exact position matters less than the band. The leader is Claude Opus 5.5 (max effort) at 72.6.
- How much does DeepSeek V3.1 cost per 1M tokens?
- As of 25 Oct 2026, DeepSeek V3.1 costs $0.580 per 1M input tokens and $1.68 per 1M output tokens, $0.855 blended (three input tokens per output token). The list price last changed on 2 Oct 2026.
- When was DeepSeek V3.1 released?
- DeepSeek released DeepSeek V3.1 on 21 Aug 2025.
Data as of 25 Oct 2026. Compare with another model.
Cite as: BenchLeader, “DeepSeek V3.1: benchmarks, pricing, speed and rank”, data as of 25 Oct 2026, https://www.benchleader.com/models/deepseek-v3-1.