DeepSeek, open weights
DeepSeek-V3.1
DeepSeek-V3.1 by DeepSeek ranks 108th of 354 ranked models on the Noometry Index as of October 2026, with a score of 42.8. Its strongest category is knowledge, where it ranks 90th. API pricing starts at $0.25 per million input tokens and $0.95 per million output tokens, with a 164K-token context window.
Last verified
Specifications
- Noometry rank
- #108 of 354
- Index score
- 42.8
- Evidence
- Confirmed 27 results
- Provider
DeepSeek- Released
- August 21, 2025
- Weights
- Open weights
- Reasoning
- Yes
- Context window
- 164K
- Max output
- 8K
- Input price
- $0.25 / M
- Output price
- $0.95 / M
- Blended price
- $0.42 / M
- Output speed
- 328 tokens/s Kagi
- Value
- #59 of 219
- Knowledge cutoff
- August 2025
- Input
- text
- Hugging Face
- deepseek-ai/DeepSeek-V3.1
Category scores
Each category score combines every public result we have in that category.
- Coding 40.3
- Reasoning 27.9
- Math 38.9
- Knowledge 43.7
- Multilingual 51.6
- Instruction Following 73.9
- Long Context 36.3
- Writing & Preference 60.3
| Category | Score | Rank | Results |
|---|---|---|---|
| Coding | 40.3 | #144 | 2 |
| Reasoning | 27.9 | #110 | 5 |
| Math | 38.9 | #122 | 1 |
| Knowledge | 43.7 | #90 | 2 |
| Multilingual | 51.6 | #106 | 1 |
| Instruction Following | 73.9 | #110 | 1 |
| Long Context | 36.3 | #232 | 2 |
| Writing & Preference | 60.3 | #98 | 4 |
Strengths and weaknesses
Categories where DeepSeek-V3.1 places highest and lowest among the models ranked in each, with its score against that category's median.
Strongest categories
| Category | Score | vs median | Rank |
|---|---|---|---|
| Knowledge | 43.7 | +6.4 | #90 of 314, top 29% |
| Writing & Preference | 60.3 | +6.5 | #98 of 312, top 32% |
| Reasoning | 27.9 | +4.3 | #110 of 350, top 32% |
Weakest categories
| Category | Score | vs median | Rank |
|---|---|---|---|
| Long Context | 36.3 | −4.7 | #232 of 296, top 79% |
| Coding | 40.3 | +1.6 | #144 of 340, top 43% |
| Math | 38.9 | +2.3 | #122 of 327, top 38% |
Closest competitors
The models ranked just above and below DeepSeek-V3.1. When scores are this close, price and speed are often the better way to choose.
| Model | Rank | Score | Blended $/M | Speed | |
|---|---|---|---|---|---|
| Amazon Nova Experimental Chat 12 10 | #104 | 42.9 | — | — | Compare |
| o3-pro | #105 | 42.9 | $35 | 1 | Compare |
| Qwen3.5 Plus | #106 | 42.9 | $0.90 | — | Compare |
| Amazon Nova Experimental Chat 26 01 10 | #107 | 42.8 | — | — | Compare |
| GPT-5.3 Chat | #109 | 42.8 | $4.81 | — | Compare |
| GPT-5.5 Instant | #110 | 42.7 | — | — | Compare |
| GPT-5.2 Codex | #111 | 42.6 | $4.81 | — | Compare |
| Qwen3.5-Flash | #112 | 42.5 | $0.18 | — | Compare |
Sponsored placements are available on pages like this one. Advertise on Noometry
Benchmark results
Every published result we track, with its source. Bold rows are the ones used for ranking; where several exist we prefer independent runs over self-reported numbers.
Coding
| Benchmark | Score | Position | Setting | Source | Date |
|---|---|---|---|---|---|
| WeirdML | 37.5% | Epoch AI | |||
| WeirdML | 38.4% | #83 of 119, top 70% | thinking | Epoch AI | |
| LMArena Coding | 1417 | #119 of 294, top 41% | thinking | LMArena | 2026-10-08 |
Reasoning
| Benchmark | Score | Position | Setting | Source | Date |
|---|---|---|---|---|---|
| SimpleBench | 40% | #54 of 77, top 71% | Epoch AI | ||
| Kagi LLM Benchmark | 53.2% | #55 of 99, top 56% | Kagi LLM Benchmark | ||
| LMArena Hard Prompts | 1417 | #108 of 297, top 37% | thinking | LMArena | 2026-10-08 |
| DTBench | 82.7% | #61 of 151, top 41% | thinking | Epoch AI | |
| LMCA | 24.3% | #87 of 125, top 70% | thinking | Epoch AI | |
| Epoch Capabilities Index | 139.92 | #109 of 213, top 52% | Epoch AI | 2025-08-21 | |
| ForecastBench | 58 | #52 of 72, top 73% | Epoch AI |
Math
| Benchmark | Score | Position | Setting | Source | Date |
|---|---|---|---|---|---|
| LMArena Math | 1420 | #101 of 285, top 36% | LMArena | 2026-10-08 |
Knowledge
| Benchmark | Score | Position | Setting | Source | Date |
|---|---|---|---|---|---|
| Vectara Hallucination Rate (lower is better) | 5.5% | #16 of 96, top 17% | Vectara Hallucination Leaderboard | ||
| LMArena Expert | 1405 | #119 of 273, top 44% | thinking | LMArena | 2026-10-08 |
Multilingual
| Benchmark | Score | Position | Setting | Source | Date |
|---|---|---|---|---|---|
| LMArena Non-English | 1400 | #107 of 297, top 37% | thinking | LMArena | 2026-10-08 |
| LMArena Chinese | 1469 | #75 of 285, top 27% | thinking | LMArena | 2026-10-08 |
| LMArena French | 1447 | #76 of 223, top 35% | LMArena | 2026-10-08 | |
| LMArena German | 1411 | #86 of 231, top 38% | LMArena | 2026-10-08 | |
| LMArena Japanese | 1378 | #81 of 211, top 39% | LMArena | 2026-10-08 | |
| LMArena Korean | 1337 | #111 of 213, top 53% | thinking | LMArena | 2026-10-08 |
| LMArena Russian | 1405 | #100 of 283, top 36% | LMArena | 2026-10-08 | |
| LMArena Spanish | 1431 | #82 of 226, top 37% | LMArena | 2026-10-08 |
Instruction Following
| Benchmark | Score | Position | Setting | Source | Date |
|---|---|---|---|---|---|
| LMArena Instruction Following | 1400 | #104 of 298, top 35% | thinking | LMArena | 2026-10-08 |
Long Context
| Benchmark | Score | Position | Setting | Source | Date |
|---|---|---|---|---|---|
| Fiction.LiveBench | 52.8% | #33 of 47, top 71% | Epoch AI | ||
| LMArena Longer Query | 1422 | #90 of 291, top 31% | thinking | LMArena | 2026-10-08 |
Writing & Preference
| Benchmark | Score | Position | Setting | Source | Date |
|---|---|---|---|---|---|
| LMArena Text | 1420 | #93 of 297, top 32% | LMArena | 2026-10-08 | |
| LMArena Creative Writing | 1401 | #78 of 295, top 27% | thinking | LMArena | 2026-10-08 |
| EQ-Bench Creative Writing | 1436 | #64 of 115, top 56% | EQ-Bench | ||
| LMArena Multi-Turn | 1408 | #114 of 295, top 39% | thinking | LMArena | 2026-10-08 |
API pricing by provider
| Route | Input $/M | Output $/M | Cached input $/M | Checked |
|---|---|---|---|---|
| bedrock | $0.58 | $1.68 | — | 2026-10-10 |
| deepinfra | $0.25 | $0.95 | $0.13 | 2026-10-10 |
| openrouter | $0.25 | $0.95 | $0.13 | 2026-10-10 |
| together | $0.60 | $1.70 | — | 2026-10-10 |
| vertex | $0.60 | $1.70 | $0.06 | 2026-10-10 |
Compare DeepSeek-V3.1
- DeepSeek-V3.1 vs DeepSeek-R1
- DeepSeek-V3.1 vs Amazon Nova Experimental Chat 26 01 10
- DeepSeek-V3.1 vs GPT-5.3 Chat
- DeepSeek-V3.1 vs Qwen3.5 Plus
- DeepSeek-V3.1 vs GPT-5.5 Instant
- DeepSeek-V3.1 vs o3-pro
- DeepSeek-V3.1 vs GPT-5.2 Codex
- DeepSeek-V3.1 vs GPT-6 Astra
- DeepSeek-V3.1 vs Claude Fable 5.1
- DeepSeek-V3.1 vs Gemini 3.8 Flash
- DeepSeek-V3.1 vs Kimi K3
- DeepSeek-V3.1 vs Grok 4.6
- DeepSeek-V3.1 vs Qwen3.8 Max
- DeepSeek-V3.1 vs GLM-5.3
Other DeepSeek models
Frequently asked questions
How good is DeepSeek-V3.1?
DeepSeek-V3.1 by DeepSeek ranks 108th of 354 ranked models on the Noometry Index as of October 2026, with a score of 42.8. Its strongest category is knowledge, where it ranks 90th. API pricing starts at $0.25 per million input tokens and $0.95 per million output tokens, with a 164K-token context window.
How much does DeepSeek-V3.1 cost?
DeepSeek-V3.1 costs $0.25 per million input tokens and $0.95 per million output tokens on deepinfra, with cached input at $0.13.
What is DeepSeek-V3.1's context window?
DeepSeek-V3.1 accepts up to 164K tokens of input and can write up to 8K tokens in one response.
Is DeepSeek-V3.1 open source?
Yes. DeepSeek-V3.1's weights are downloadable from Hugging Face (deepseek-ai/DeepSeek-V3.1); check the license for commercial terms.
How fast is DeepSeek-V3.1?
DeepSeek-V3.1 generated about 328 output tokens per second in the Kagi LLM Benchmark's timed runs. Speed varies by provider, load and reasoning effort.
What are DeepSeek-V3.1's strengths and weaknesses?
Relative to other ranked models, DeepSeek-V3.1 places best in knowledge, writing & preference, reasoning and lowest in long context, coding, math.
What is DeepSeek-V3.1 best at?
Its best category is knowledge, where it ranks 90th on Noometry.