DeepSeek model product
A powerful Mixture-of-Experts (MoE) language model with 671B total parameters (37B activated per token). Features Multi-head Latent Attention (MLA), auxiliary-loss-free load balancing, and multi-token prediction training. Pre-trained on 14.8T tokens with strong performance in reasoning, math, and code tasks.
Updated Aug 10, 2026. Default version: DeepSeek-V3 0324
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 0324.
| MATH-500 | 0.9 | 22 | 32 | 32.3% | C | |
| AIME 2024 | 0.6 | 44 | 53 | 17.3% | C | |
| MMLU-Pro | 0.8 | 44 | 129 | 66.4% | C | |
| LiveCodeBench | 0.5 | 47 | 73 | 36.1% | C | |
| GPQA | 0.7 | 135 | 233 | 42.2% | C |
Preference and agent-evaluation signals from published Arena datasets.
| text style control | creative writing | 108 | 1389.9 | 6,031 | N/A | |
| text style control | german | 111 | 1398.4 | 1,154 | N/A | |
| text style control | polish | 117 | 1395.3 | 2,689 | N/A | |
| text | japanese | 118 | 1326.8 | 961 | N/A | |
| text style control | industry entertainment and sports and media | 118 | 1382.6 | 8,044 | N/A | |
| text style control | japanese | 119 | 1337.4 | 961 | N/A | |
| text style control | french | 122 | 1420.3 | 539 | N/A | |
| text style control | multi turn | 123 | 1410.3 | 7,850 | N/A | |
| text | german | 124 | 1380.2 | 1,154 | N/A | |
| text | creative writing | 127 | 1364.5 | 6,031 | N/A | |
| text style control | russian | 127 | 1399.0 | 3,094 | N/A | |
| text style control | industry writing and literature and language | 128 | 1386.4 | 10,322 | N/A | |
| text style control | industry legal and government | 131 | 1410.6 | 2,945 | N/A | |
| text | polish | 132 | 1367.9 | 2,689 | N/A | |
| text | french | 133 | 1388.7 | 539 | N/A | |
| text style control | industry medicine and healthcare | 134 | 1427.1 | 2,662 | N/A | |
| text | industry entertainment and sports and media | 135 | 1356.9 | 8,044 | N/A | |
| text style control | korean | 135 | 1333.8 | 922 | N/A | |
| text | korean | 137 | 1316.6 | 922 | N/A | |
| text style control | industry life and physical and social science | 140 | 1414.4 | 7,671 | N/A | |
| text style control | non english | 140 | 1380.1 | 22,199 | N/A | |
| text | multi turn | 141 | 1387.9 | 7,850 | N/A | |
| text style control | overall | 141 | 1395.6 | 45,445 | N/A | |
| text style control | exclude ties | 141 | 1376.6 | 31,818 | N/A | |
| text | industry writing and literature and language | 144 | 1364.9 | 10,322 | N/A | |
| text | russian | 144 | 1371.9 | 3,094 | N/A | |
| text style control | english | 145 | 1407.6 | 23,241 | N/A | |
| text style control | industry business and management and financial operations | 147 | 1390.3 | 6,919 | N/A | |
| text style control | chinese | 148 | 1413.6 | 2,466 | N/A | |
| text | english | 149 | 1386.6 | 23,241 | N/A | |
| text | industry legal and government | 150 | 1380.5 | 2,945 | N/A | |
| text | non english | 150 | 1360.5 | 22,199 | N/A | |
| text | overall | 151 | 1375.1 | 45,445 | N/A | |
| text style control | industry software and it services | 151 | 1423.9 | 14,611 | N/A | |
| text style control | longer query | 151 | 1393.6 | 8,447 | N/A | |
| text | exclude ties | 152 | 1346.0 | 31,818 | N/A | |
| text style control | hard prompts | 152 | 1408.7 | 18,671 | N/A | |
| text | industry life and physical and social science | 153 | 1381.7 | 7,671 | N/A | |
| text style control | hard prompts english | 153 | 1418.0 | 10,217 | N/A | |
| text style control | instruction following | 154 | 1378.6 | 12,386 | N/A | |
| text | industry medicine and healthcare | 155 | 1382.4 | 2,662 | N/A | |
| text style control | expert | 157 | 1396.9 | 2,295 | N/A | |
| text style control | spanish | 158 | 1364.3 | 893 | N/A | |
| text | chinese | 159 | 1388.1 | 2,466 | N/A | |
| text | math | 161 | 1374.0 | 3,183 | N/A | |
| text style control | coding | 161 | 1428.8 | 8,358 | N/A | |
| text | hard prompts | 162 | 1365.8 | 18,671 | N/A | |
| text | spanish | 162 | 1345.9 | 893 | N/A | |
| text | instruction following | 163 | 1345.6 | 12,386 | N/A | |
| text | hard prompts english | 164 | 1376.1 | 10,217 | N/A | |
| text | industry business and management and financial operations | 164 | 1350.2 | 6,919 | N/A | |
| text | expert | 167 | 1351.5 | 2,295 | N/A | |
| text | industry software and it services | 167 | 1379.7 | 14,611 | N/A | |
| text | longer query | 168 | 1352.9 | 8,447 | N/A | |
| text | industry mathematical | 170 | 1365.3 | 2,923 | N/A | |
| text style control | math | 171 | 1369.5 | 3,183 | N/A | |
| text | coding | 174 | 1368.9 | 8,358 | N/A | |
| text style control | industry mathematical | 174 | 1367.3 | 2,923 | 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.
| NanoGPT | deepseek-v3-0324 | global | $0.20 | $0.77 | 128K | |
| submodel | deepseek-ai/DeepSeek-V3-0324 | global | $0.20 | $0.80 | 75K | |
| Meganova | deepseek-ai/DeepSeek-V3-0324 | global | $0.25 | $0.88 | 163.8K | |
| NovitaAI | deepseek/deepseek-v3-0324 | global | $0.27 | $1.12 | 163.8K | |
| OpenRouter | deepseek/deepseek-chat-v3-0324 | global | $0.27 | $1.12 | 163.8K | |
| Jiekou.AI | deepseek/deepseek-v3-0324 | global | $0.28 | $1.14 | 163.8K |
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 0324 | 38.1 | 671B | 163.8K | 163.8K | No | MIT + Model License (Commercial use allowed) | |
| DeepSeek-V3 | N/A | 671B | 131.1K | 8.2K | Yes | MIT + Model License (Commercial use allowed) |
A concise description based on the published model registry.
A powerful Mixture-of-Experts (MoE) language model with 671B total parameters (37B activated per token). Features Multi-head Latent Attention (MLA), auxiliary-loss-free load balancing, and multi-token prediction training. Pre-trained on 14.8T tokens with strong performance in reasoning, math, and code tasks.
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.
DeepSeek-V3's default version was released on Mar 25, 2025.
No official standard PAYG price is currently available for DeepSeek-V3. The lowest tracked third-party offer starts at $0.20 input and $0.77 output via NanoGPT.
DeepSeek-V3 is published under DeepSeek in the model registry.
The default version has a 163.8K token context window.
No. The default version is not marked as having publicly available weights.
7 published provider offerings are linked to the default version.
Nearby ranked alternatives include Gemma 3 12B, Jamba 1.5 Large, DiffusionGemma 26B A4B.