DeepSeek model product
DeepSeek-V3.1 is a hybrid model supporting both thinking and non-thinking modes through different chat templates. Built on DeepSeek-V3.1-Base with a two-phase long context extension (32K phase: 630B tokens, 128K phase: 209B tokens), it features 671B total parameters with 37B activated. Key improvements include smarter tool calling through post-training optimization, higher thinking efficiency achieving comparable quality to DeepSeek-R1-0528 while responding more quickly, and UE8M0 FP8 scale data format for model weights and activations. The model excels in both reasoning tasks (thinking mode) and practical applications (non-thinking mode), with particularly strong performance in code agent tasks, math competitions, and search-based problem solving.
Updated Aug 10, 2026. Default version: DeepSeek-V3.1
Structured fields from the published default version.
This profile uses the latest version under this unique model that has a calculated LLMBoard score. Arena and price are excluded.
Published benchmark records for the scored version DeepSeek-V3.1.
| SimpleQA | 0.9 | 3 | 46 | 95.6% | C | |
| Aider-Polyglot | 0.7 | 8 | 22 | 66.7% | C | |
| BrowseComp-zh | 0.5 | 10 | 13 | 25.0% | C | |
| CodeForces | 0.7 | 12 | 16 | 26.7% | C | |
| Terminal-Bench | 0.3 | 18 | 25 | 29.2% | C | |
| MMLU-Redux | 0.9 | 21 | 48 | 57.5% | C | |
| SWE-bench Multilingual | 0.5 | 29 | 34 | 15.2% | C | |
| MMLU-Pro | 0.8 | 30 | 129 | 77.3% | C | |
| HMMT 2025 | 0.3 | 32 | 33 | 3.1% | C | |
| LiveCodeBench | 0.6 | 36 | 73 | 51.4% | C | |
| AIME 2024 | 0.7 | 43 | 53 | 19.2% | C | |
| BrowseComp | 0.3 | 55 | 58 | 5.3% | C | |
| Humanity's Last Exam | 0.2 | 68 | 92 | 26.4% | C | |
| SWE-Bench Verified | 0.7 | 71 | 104 | 32.0% | C | |
| AIME 2025 | 0.5 | 100 | 114 | 12.4% | C | |
| GPQA | 0.7 | 108 | 233 | 53.9% | C |
Preference and agent-evaluation signals from published Arena datasets.
| text | french | 54 | 1457.3 | 246 | N/A | |
| text style control | french | 84 | 1452.8 | 246 | N/A | |
| text style control | japanese | 87 | 1376.0 | 308 | N/A | |
| text | japanese | 89 | 1367.8 | 308 | N/A | |
| text | german | 93 | 1409.4 | 374 | N/A | |
| text | industry medicine and healthcare | 93 | 1432.4 | 935 | N/A | |
| text | polish | 93 | 1409.5 | 992 | N/A | |
| text | industry writing and literature and language | 96 | 1397.8 | 3,459 | N/A | |
| text | chinese | 97 | 1458.0 | 1,110 | N/A | |
| text | english | 97 | 1430.7 | 6,773 | N/A | |
| text | industry life and physical and social science | 98 | 1433.4 | 2,313 | N/A | |
| text | overall | 99 | 1419.6 | 14,942 | N/A | |
| text style control | chinese | 100 | 1463.0 | 1,110 | N/A | |
| text | math | 101 | 1420.7 | 992 | N/A | |
| text | spanish | 101 | 1412.0 | 362 | N/A | |
| text | exclude ties | 102 | 1405.8 | 10,654 | N/A | |
| text | creative writing | 104 | 1388.2 | 2,000 | N/A | |
| text style control | german | 106 | 1405.5 | 374 | N/A | |
| text | industry business and management and financial operations | 107 | 1405.9 | 2,518 | N/A | |
| text | industry entertainment and sports and media | 107 | 1379.8 | 2,747 | N/A | |
| text style control | industry writing and literature and language | 107 | 1399.0 | 3,459 | N/A | |
| text style control | spanish | 107 | 1405.2 | 362 | N/A | |
| text style control | industry business and management and financial operations | 110 | 1414.6 | 2,518 | N/A | |
| text | industry legal and government | 111 | 1418.3 | 1,011 | N/A | |
| text style control | overall | 111 | 1417.6 | 14,942 | N/A | |
| text style control | creative writing | 111 | 1389.0 | 2,000 | N/A | |
| text style control | exclude ties | 112 | 1404.7 | 10,654 | N/A | |
| text style control | math | 112 | 1414.5 | 992 | N/A | |
| text style control | polish | 112 | 1400.1 | 992 | N/A | |
| text | industry mathematical | 113 | 1415.3 | 839 | N/A | |
| text style control | industry life and physical and social science | 113 | 1434.6 | 2,313 | N/A | |
| text | non english | 114 | 1398.8 | 8,169 | N/A | |
| text | russian | 114 | 1400.7 | 790 | N/A | |
| text style control | english | 114 | 1428.1 | 6,773 | N/A | |
| text style control | industry medicine and healthcare | 115 | 1434.9 | 935 | N/A | |
| text style control | instruction following | 116 | 1403.5 | 3,689 | N/A | |
| text | industry software and it services | 117 | 1428.3 | 4,783 | N/A | |
| text style control | hard prompts | 117 | 1433.0 | 6,801 | N/A | |
| text | hard prompts | 119 | 1415.3 | 6,801 | N/A | |
| text style control | longer query | 119 | 1421.3 | 3,210 | N/A | |
| text | instruction following | 120 | 1389.4 | 3,689 | N/A | |
| text style control | industry legal and government | 121 | 1421.0 | 1,011 | N/A | |
| text style control | industry software and it services | 121 | 1445.4 | 4,783 | N/A | |
| text style control | non english | 121 | 1397.6 | 8,169 | N/A | |
| text | hard prompts english | 122 | 1418.6 | 3,357 | N/A | |
| text style control | industry entertainment and sports and media | 122 | 1378.7 | 2,747 | N/A | |
| text style control | expert | 123 | 1426.5 | 726 | N/A | |
| text | longer query | 124 | 1400.5 | 3,210 | N/A | |
| text style control | russian | 124 | 1401.2 | 790 | N/A | |
| text | expert | 125 | 1402.8 | 726 | N/A | |
| text style control | hard prompts english | 125 | 1435.5 | 3,357 | N/A | |
| text style control | industry mathematical | 125 | 1414.0 | 839 | N/A | |
| text | coding | 128 | 1415.2 | 2,624 | N/A | |
| text | korean | 128 | 1331.9 | 472 | N/A | |
| text | multi turn | 128 | 1401.7 | 2,482 | N/A | |
| text style control | multi turn | 130 | 1406.3 | 2,482 | N/A | |
| text style control | coding | 132 | 1448.3 | 2,624 | N/A | |
| text style control | korean | 132 | 1338.2 | 472 | N/A |
Official vendor API PAYG pricing is summarized first. The table then lists individual provider offerings without treating their minimum as the official price.
The default version has no provider offering with current input or output token prices.
Official prices use only the vendor's configured official Provider and positive standard USD PAYG rates. Third-party offers remain explicitly labeled.
All published versions linked to this unique model. The score columns identify the version used by the current overall ranking.
| DeepSeek-V3.1 | 36.0 | 671B | 131.1K | 8.2K | Yes | MIT |
A concise description based on the published model registry.
DeepSeek-V3.1 is a hybrid model supporting both thinking and non-thinking modes through different chat templates. Built on DeepSeek-V3.1-Base with a two-phase long context extension (32K phase: 630B tokens, 128K phase: 209B tokens), it features 671B total parameters with 37B activated. Key improvements include smarter tool calling through post-training optimization, higher thinking efficiency achieving comparable quality to DeepSeek-R1-0528 while responding more quickly, and UE8M0 FP8 scale data format for model weights and activations. The model excels in both reasoning tasks (thinking mode) and practical applications (non-thinking mode), with particularly strong performance in code agent tasks, math competitions, and search-based problem solving.
Use the benchmark, Arena and pricing sections above as separate evidence. A missing field means the current data snapshot does not support that claim.
Data snapshot: 2026-08-07. Editorial model content is not available in the backend.
Open a comparison with the three ranked models immediately above and below this model.
Recommendations prioritize the same model type and family, then the closest published LLMBoard score.
Common questions about DeepSeek-V3.1.
DeepSeek-V3.1's default version was released on Jan 10, 2025.
No official standard PAYG price is currently available for DeepSeek-V3.1.
DeepSeek-V3.1 is published under DeepSeek in the model registry.
The default version has a 131.1K token context window.
Yes. The default version is marked as open weight under MIT.
1 published provider offerings are linked to the default version.
Nearby ranked alternatives include Nova Micro, Qwen3.5 9B, Qwen3 VL 32B.