DeepSeek model product
DeepSeek-V3
DeepSeek-V3 is a Mixture-of-Experts language model with 671B total parameters and 37B activated per token.
Updated Sep 8, 2026. Default version: DeepSeek-V3 0324
LLMBoard Score25.0DeepSeek-V3 0324
Coverage20%5 benchmark families
Context window163.8KTokens
Official input priceN/AOfficial price unavailable
DeepSeek-V3 Capability Profile
This profile uses the model's current scored version. Arena ratings and prices are shown separately.
DeepSeek-V3 Benchmark Results
Benchmark scores for DeepSeek-V3 0324.
Sort by
Benchmark | Score | Rank | Participants | Percentile | Evidence | Evaluated |
|---|
| BenchmarkMATH-500 | Score94.00% | Rank22 | Participants32 | Percentile32.26% | EvidenceC | Evaluated |
| BenchmarkAIME 2024 | Score59.40% | Rank44 | Participants53 | Percentile17.31% | EvidenceC | Evaluated |
| BenchmarkLiveCodeBench | Score49.20% | Rank49 | Participants75 | Percentile35.14% | EvidenceC | Evaluated |
| BenchmarkMMLU-Pro | Score81.20% | Rank49 | Participants138 | Percentile64.96% | EvidenceC | Evaluated |
| BenchmarkLM Arena Text Style Control | Score1,395.69 rating | Rank112 | Participants210 | Percentile46.89% | EvidenceA | Evaluated |
| BenchmarkLM Arena Text | Score1,374.91 rating | Rank117 | Participants210 | Percentile44.50% | EvidenceA | Evaluated |
| BenchmarkGPQA | Score68.40% | Rank146 | Participants247 | Percentile41.06% | EvidenceC | Evaluated |
DeepSeek-V3 Arena Results
Preference and agent-evaluation results for the default version.
Sort by
Arena | Category | Rank | Rating / score | Votes | Observations | Result date |
|---|
| Arenatext style control | Categorycreative writing | Rank89 | Rating / score1,389.49 | Votes5,969 | ObservationsN/A | Result date |
| Arenatext style control | Categorypolish | Rank91 | Rating / score1,396.14 | Votes2,727 | ObservationsN/A | Result date |
| Arenatext | Categoryjapanese | Rank93 | Rating / score1,333.70 | Votes1,044 | ObservationsN/A | Result date |
| Arenatext style control | Categorygerman | Rank94 | Rating / score1,391.30 | Votes1,220 | ObservationsN/A | Result date |
| Arenatext style control | Categoryjapanese | Rank94 | Rating / score1,344.11 | Votes1,044 | ObservationsN/A | Result date |
| Arenatext style control | Categoryindustry entertainment and sports and media | Rank96 | Rating / score1,382.19 | Votes7,981 | ObservationsN/A | Result date |
| Arenatext | Categorypolish | Rank100 | Rating / score1,368.60 | Votes2,727 | ObservationsN/A | Result date |
| Arenatext style control | Categorymulti turn | Rank100 | Rating / score1,411.19 | Votes7,703 | ObservationsN/A | Result date |
| Arenatext | Categorycreative writing | Rank101 | Rating / score1,364.01 | Votes5,969 | ObservationsN/A | Result date |
| Arenatext | Categorygerman | Rank102 | Rating / score1,373.88 | Votes1,220 | ObservationsN/A | Result date |
| Arenatext style control | Categoryfrench | Rank102 | Rating / score1,416.21 | Votes590 | ObservationsN/A | Result date |
| Arenatext style control | Categoryindustry writing and literature and language | Rank102 | Rating / score1,386.19 | Votes10,251 | ObservationsN/A | Result date |
| Arenatext style control | Categorykorean | Rank102 | Rating / score1,339.48 | Votes949 | ObservationsN/A | Result date |
| Arenatext style control | Categoryrussian | Rank102 | Rating / score1,399.66 | Votes3,572 | ObservationsN/A | Result date |
| Arenatext style control | Categoryindustry legal and government | Rank103 | Rating / score1,411.09 | Votes2,898 | ObservationsN/A | Result date |
| Arenatext | Categorykorean | Rank105 | Rating / score1,321.12 | Votes949 | ObservationsN/A | Result date |
| Arenatext style control | Categoryindustry medicine and healthcare | Rank106 | Rating / score1,426.58 | Votes2,647 | ObservationsN/A | Result date |
| Arenatext | Categoryfrench | Rank108 | Rating / score1,386.17 | Votes590 | ObservationsN/A | Result date |
| Arenatext | Categoryindustry entertainment and sports and media | Rank108 | Rating / score1,356.46 | Votes7,981 | ObservationsN/A | Result date |
| Arenatext style control | Categoryindustry life and physical and social science | Rank109 | Rating / score1,414.77 | Votes7,614 | ObservationsN/A | Result date |
| Arenatext | Categoryindustry writing and literature and language | Rank110 | Rating / score1,364.59 | Votes10,251 | ObservationsN/A | Result date |
| Arenatext | Categoryrussian | Rank112 | Rating / score1,372.15 | Votes3,572 | ObservationsN/A | Result date |
| Arenatext style control | Categoryexclude ties | Rank112 | Rating / score1,376.67 | Votes31,354 | ObservationsN/A | Result date |
| Arenatext style control | Categorynon english | Rank112 | Rating / score1,379.12 | Votes23,444 | ObservationsN/A | Result date |
| Arenatext style control | Categoryoverall | Rank112 | Rating / score1,395.69 | Votes44,795 | ObservationsN/A | Result date |
| Arenatext | Categorymulti turn | Rank113 | Rating / score1,388.60 | Votes7,703 | ObservationsN/A | Result date |
| Arenatext style control | Categoryenglish | Rank114 | Rating / score1,408.96 | Votes21,346 | ObservationsN/A | Result date |
| Arenatext | Categoryenglish | Rank115 | Rating / score1,389.20 | Votes21,346 | ObservationsN/A | Result date |
| Arenatext style control | Categoryspanish | Rank115 | Rating / score1,378.01 | Votes983 | ObservationsN/A | Result date |
| Arenatext style control | Categoryindustry business and management and financial operations | Rank116 | Rating / score1,389.70 | Votes6,840 | ObservationsN/A | Result date |
DeepSeek-V3 Pricing
Official vendor API pricing appears first, followed by individual provider offers.
- Official API
- N/A
- Official provider
- N/A
- Lowest third-party
- From $0.20 input, $0.77 output per 1M via NanoGPT
- Tracked offerings
- 10
Sort by
Provider | Provider model ID | Region | Input / 1M | Output / 1M | Context | Updated |
|---|
| ProviderNanoGPT | Provider model IDdeepseek-v3-0324 | Regionglobal | Input / 1M$0.20 | Output / 1M$0.77 | Context128K | Updated |
| Providersubmodel | Provider model IDdeepseek-ai/DeepSeek-V3-0324 | Regionglobal | Input / 1M$0.20 | Output / 1M$0.80 | Context75K | Updated |
| ProviderDeep Infra | Provider model IDdeepseek-ai/DeepSeek-V3-0324 | Regionglobal | Input / 1M$0.24 | Output / 1M$0.90 | Context163.8K | Updated |
| ProviderMeganova | Provider model IDdeepseek-ai/DeepSeek-V3-0324 | Regionglobal | Input / 1M$0.25 | Output / 1M$0.88 | Context163.8K | Updated |
| ProviderHugging Face | Provider model IDdeepseek-ai/DeepSeek-V3-0324 | Regionglobal | Input / 1M$0.27 | Output / 1M$1.12 | Context163.8K | Updated |
| ProviderNovitaAI | Provider model IDdeepseek/deepseek-v3-0324 | Regionglobal | Input / 1M$0.27 | Output / 1M$1.12 | Context163.8K | Updated |
| ProviderJiekou.AI | Provider model IDdeepseek/deepseek-v3-0324 | Regionglobal | Input / 1M$0.28 | Output / 1M$1.14 | Context163.8K | Updated |
| ProviderOpenRouter | Provider model IDdeepseek/deepseek-chat-v3-0324 | Regionglobal | Input / 1M$0.29 | Output / 1M$1.14 | Context163.8K | Updated |
| ProviderCrusoe | Provider model IDdeepseek-ai/DeepSeek-V3-0324 | Regionglobal | Input / 1M$0.50 | Output / 1M$1.5 | Context163.8K | Updated |
Official prices use only the vendor's configured official Provider and positive standard USD PAYG rates. Third-party offers remain explicitly labeled.
DeepSeek-V3 Runtime Performance
Provider-specific output speed and catalog latency for DeepSeek-V3 0324. Runtime does not affect the capability score.
No runtime data
No provider-specific speed or latency record is linked to the default version yet.
Output Speed is generated output tokens received per second. Catalog latency is reported separately from observed provider TTFT.
DeepSeek-V3 Specifications
Technical details for the model's default version.
- Version
- DeepSeek-V3 0324
- Released
- Mar 25, 2025
- Knowledge cutoff
- Unknown
- Parameters
- 671B
- Context window
- 163.8K
- Max output
- 163.8K
- Inputs
- text
- Outputs
- text
- Open weights
- Yes
- License
- MIT + Model License (Commercial use allowed)
DeepSeek-V3 Versions
Available versions of this model. The score column identifies the version used in the overall ranking.
Sort by
Version | Released | LLMBoard | Parameters | Context | Max output | Open weights | License |
|---|
| VersionDeepSeek-V3 0324 | Released | LLMBoard24.98 | Parameters671B | Context163.8K | Max output163.8K | Open weightsYes | LicenseMIT + Model License (Commercial use allowed) |
| VersionDeepSeek-V3 | Released | LLMBoardN/A | Parameters671B | Context131.1K | Max output131.1K | Open weightsYes | LicenseMIT + Model License (Commercial use allowed) |
Models similar to DeepSeek-V3
Recommendations prioritize the same model type and family, then the closest LLMBoard score.
DE
DeepSeek-R1-Distill-Qwen
DeepSeek
DE
DeepSeek-V2.5
DeepSeek
DE
DeepSeek-R1-Distill-Llama
DeepSeek
DE
DeepSeek-V3.1
DeepSeek
DE
DeepSeek-R1
DeepSeek
DE
DeepSeek-VL2
DeepSeek
What is DeepSeek-V3?
Key information about DeepSeek-V3 and its available data.
DeepSeek-V3 is a language model from DeepSeek that uses Multi-head Latent Attention (MLA), auxiliary-loss-free load balancing, and multi-token prediction training. 8T tokens and supports reasoning, math, and code tasks.
Data as of 2026-09-08.
FAQ
Common questions about DeepSeek-V3.
When was DeepSeek-V3 released?
DeepSeek-V3's default version was released on Mar 25, 2025.
How much does DeepSeek-V3 cost?
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.
Who created DeepSeek-V3?
DeepSeek-V3 was created by DeepSeek.
What is the context window for DeepSeek-V3?
The default version has a 163.8K token context window.
Is DeepSeek-V3 open weight?
Yes. The default version is marked as open weight under MIT + Model License (Commercial use allowed).
How many API providers offer DeepSeek-V3?
10 provider offerings are linked to the default version.