Introduction
We present DeepSeek-V3, a strong Mixture-of-Experts (MoE) language model with 671B total parameters with 37B activated for each token. To achieve efficient inference and cost-effective training, DeepSeek-V3 adopts Multi-head Latent Attention (MLA) and DeepSeekMoE architectures, which were thoroughly validated in DeepSeek-V2. Furthermore, DeepSeek-V3 pioneers an auxiliary-loss-free strategy for load balancing and sets a multi-token prediction training objective for stronger performance. We pre-train DeepSeek-V3 on 14.8 trillion diverse and high-quality tokens, followed by Supervised Fine-Tuning and Reinforcement Learning stages to fully harness its capabilities. Comprehensive evaluations reveal that DeepSeek-V3 outperforms other open-source models and achieves performance comparable to leading closed-source models. Despite its excellent performance, DeepSeek-V3 requires only 2.788M H800 GPU hours for its full training. In addition, its training process is remarkably stable. Throughout the entire training process, we did not experience any irrecoverable loss spikes or perform any rollbacks.
Model Summary
Architecture: Innovative Load Balancing Strategy and Training Objective
- On top of the efficient architecture of DeepSeek-V2, we pioneer an auxiliary-loss-free strategy for load balancing, which minimizes the performance degradation that arises from encouraging load balancing.
- We investigate a Multi-Token Prediction (MTP) objective and prove it beneficial to model performance. It can also be used for speculative decoding for inference acceleration.
Pre-Training: Towards Ultimate Training Efficiency
- We design an FP8 mixed precision training framework and, for the first time, validate the feasibility and effectiveness of FP8 training on an extremely large-scale model.
- Through co-design of algorithms, frameworks, and hardware, we overcome the communication bottleneck in cross-node MoE training, nearly achieving full computation-communication overlap.
- This significantly enhances our training efficiency and reduces the training costs, enabling us to further scale up the model size without additional overhead.
- At an economical cost of only 2.664M H800 GPU hours, we complete the pre-training of DeepSeek-V3 on 14.8T tokens, producing the currently strongest open-source base model. The subsequent training stages after pre-training require only 0.1M GPU hours.
Post-Training: Knowledge Distillation from DeepSeek-R1
- We introduce an innovative methodology to distill reasoning capabilities from the long-Chain-of-Thought (CoT) model, specifically from one of the DeepSeek R1 series models, into standard LLMs, particularly DeepSeek-V3. Our pipeline elegantly incorporates the verification and reflection patterns of R1 into DeepSeek-V3 and notably improves its reasoning performance. Meanwhile, we also maintain a control over the output style and length of DeepSeek-V3.
Model Downloads
| Model | Total Params | Activated Params | Context Length | Download |
|---|
| DeepSeek-V3-Base | 671B | 37B | 128K | 🤗 HuggingFace |
| DeepSeek-V3 | 671B | 37B | 128K | 🤗 HuggingFace |
NOTE: The total size of DeepSeek-V3 models on HuggingFace is 685B, which includes 671B of the Main Model weights and 14B of the Multi-Token Prediction (MTP) Module weights.
To ensure optimal performance and flexibility, we have partnered with open-source communities and hardware vendors to provide multiple ways to run the model locally. For step-by-step guidance, check out Section 6: How_to Run_Locally.
For developers looking to dive deeper, we recommend exploring README_WEIGHTS.md for details on the Main Model weights and the Multi-Token Prediction (MTP) Modules. Please note that MTP support is currently under active development within the community, and we welcome your contributions and feedback.
4. Evaluation Results
Base Model - Standard Benchmarks
| Benchmark (Metric) | # Shots | DeepSeek-V2 | Qwen2.5 72B | LLaMA3.1 405B | DeepSeek-V3 |
|---|
| Architecture (-) | - | MoE | Dense | Dense | MoE |
| # Activated Params (-) | - | 21B | 72B | 405B | 37B |
| # Total Params (-) | - | 236B | 72B | 405B | 671B |
| English Pile-test (BPB) | - | 0.606 | 0.638 | 0.542 | 0.548 |
| English BBH (EM) | 3-shot | 78.8 | 79.8 | 82.9 | 87.5 |
| English MMLU (Acc.) | 5-shot | 78.4 | 85.0 | 84.4 | 87.1 |
| English MMLU-Redux (Acc.) | 5-shot | 75.6 | 83.2 | 81.3 | 86.2 |
| English MMLU-Pro (Acc.) | 5-shot | 51.4 | 58.3 | 52.8 | 64.4 |
| English DROP (F1) | 3-shot | 80.4 | 80.6 | 86.0 | 89.0 |
| English ARC-Easy (Acc.) | 25-shot | 97.6 | 98.4 | 98.4 | 98.9 |
| English ARC-Challenge (Acc.) | 25-shot | 92.2 | 94.5 | 95.3 | 95.3 |
| English HellaSwag (Acc.) | 10-shot | 87.1 | 84.8 | 89.2 | 88.9 |
| English PIQA (Acc.) | 0-shot | 83.9 | 82.6 | 85.9 | 84.7 |
| English WinoGrande (Acc.) | 5-shot | 86.3 | 82.3 | 85.2 | 84.9 |
| English RACE-Middle (Acc.) | 5-shot | 73.1 | 68.1 | 74.2 | 67.1 |
| English RACE-High (Acc.) | 5-shot | 52.6 | 50.3 | 56.8 | 51.3 |
| English TriviaQA (EM) | 5-shot | 80.0 | 71.9 | 82.7 | 82.9 |
| English NaturalQuestions (EM) | 5-shot | 38.6 | 33.2 | 41.5 | 40.0 |
| English AGIEval (Acc.) | 0-shot | 57.5 | 75.8 | 60.6 | 79.6 |
| Code HumanEval (Pass@1) | 0-shot | 43.3 | 53.0 | 54.9 | 65.2 |
| Code MBPP (Pass@1) | 3-shot | 65.0 | 72.6 | 68.4 | 75.4 |
| Code LiveCodeBench-Base (Pass@1) | 3-shot | 11.6 | 12.9 | 15.5 | 19.4 |
| Code CRUXEval-I (Acc.) | 2-shot | 52.5 | 59.1 | 58.5 | 67.3 |
| Code CRUXEval-O (Acc.) | 2-shot | 49.8 | 59.9 | 59.9 | 69.8 |
| Math GSM8K (EM) | 8-shot | 81.6 | 88.3 | 83.5 | 89.3 |
| Math MATH (EM) | 4-shot | 43.4 | 54.4 | 49.0 | 61.6 |
| Math MGSM (EM) | 8-shot | 63.6 | 76.2 | 69.9 | 79.8 |
| Math CMath (EM) | 3-shot | 78.7 | 84.5 | 77.3 | 90.7 |
| Chinese CLUEWSC (EM) | 5-shot | 82.0 | 82.5 | 83.0 | 82.7 |
| Chinese C-Eval (Acc.) | 5-shot | 81.4 | 89.2 | 72.5 | 90.1 |
| Chinese CMMLU (Acc.) | 5-shot | 84.0 | 89.5 | 73.7 | 88.8 |
| Chinese CMRC (EM) | 1-shot | 77.4 | 75.8 | 76.0 | 76.3 |
| Chinese C3 (Acc.) | 0-shot | 77.4 | 76.7 | 79.7 | 78.6 |
| Chinese CCPM (Acc.) | 0-shot | 93.0 | 88.5 | 78.6 | 92.0 |
| Multilingual MMMLU-non-English (Acc.) | 5-shot | 64.0 | 74.8 | 73.8 | 79.4 |
Note: Best results are shown in bold. Scores with a gap not exceeding 0.3 are considered to be at the same level. DeepSeek-V3 achieves the best performance on most benchmarks, especially on math and code tasks. For more evaluation details, please check our paper.
Context Window
Evaluation results on the Needle In A Haystack (NIAH) tests. DeepSeek-V3 performs well across all context window lengths up to 128K.
Chat Model - Standard Benchmarks
| Benchmark (Metric) | DeepSeek V2-0506 | DeepSeek V2.5-0905 | Qwen2.5 72B-Inst. | Llama3.1 405B-Inst. | Claude-3.5-Sonnet-1022 | GPT-4o 0513 | DeepSeek V3 |
|---|
| Architecture (-) | MoE | MoE | Dense | Dense | - | - | MoE |
| # Activated Params (-) | 21B | 21B | 72B | 405B | - | - | 37B |
| # Total Params (-) | 236B | 236B | 72B | 405B | - | - | 671B |
| English MMLU (EM) | 78.2 | 80.6 | 85.3 | 88.6 | 88.3 | 87.2 | 88.5 |
| English MMLU-Redux (EM) | 77.9 | 80.3 | 85.6 | 86.2 | 88.9 | 88.0 | 89.1 |
| English MMLU-Pro (EM) | 58.5 | 66.2 | 71.6 | 73.3 | 78.0 | 72.6 | 75.9 |
| English DROP (3-shot F1) | 83.0 | 87.8 | 76.7 | 88.7 | 88.3 | 83.7 | 91.6 |
| English IF-Eval (Prompt Strict) | 57.7 | 80.6 | 84.1 | 86.0 | 86.5 | 84.3 | 86.1 |
| English GPQA-Diamond (Pass@1) | 35.3 | 41.3 | 49.0 | 51.1 | 65.0 | 49.9 | 59.1 |
| English SimpleQA (Correct) | 9.0 | 10.2 | 9.1 | 17.1 | 28.4 | 38.2 | 24.9 |
| English FRAMES (Acc.) | 66.9 | 65.4 | 69.8 | 70.0 | 72.5 | 80.5 | 73.3 |
| English LongBench v2 (Acc.) | 31.6 | 35.4 | 39.4 | 36.1 | 41.0 | 48.1 | 48.7 |
| Code HumanEval-Mul (Pass@1) | 69.3 | 77.4 | 77.3 | 77.2 | 81.7 | 80.5 | 82.6 |
| Code LiveCodeBench (Pass@1-COT) | 18.8 | 29.2 | 31.1 | 28.4 | 36.3 | 33.4 | 40.5 |
| Code LiveCodeBench (Pass@1) | 20.3 | 28.4 | 28.7 | 30.1 | 32.8 | 34.2 | 37.6 |
| Code Codeforces (Percentile) | 17.5 | 35.6 | 24.8 | 25.3 | 20.3 | 23.6 | 51.6 |
| Code SWE Verified (Resolved) | - | 22.6 | 23.8 | 24.5 | 50.8 | 38.8 | 42.0 |
| Code Aider-Edit (Acc.) | 60.3 | 71.6 | 65.4 | 63.9 | 84.2 | 72.9 | 79.7 |
| Code Aider-Polyglot (Acc.) | - | 18.2 | 7.6 | 5.8 | 45.3 | 16.0 | 49.6 |
| Math AIME 2024 (Pass@1) | 4.6 | 16.7 | 23.3 | 23.3 | 16.0 | 9.3 | 39.2 |
| Math MATH-500 (EM) | 56.3 | 74.7 | 80.0 | 73.8 | 78.3 | 74.6 | 90.2 |
| Math CNMO 2024 (Pass@1) | 2.8 | 10.8 | 15.9 | 6.8 | 13.1 | 10.8 | 43.2 |
| Chinese CLUEWSC (EM) | 89.9 | 90.4 | 91.4 | 84.7 | 85.4 | 87.9 | 90.9 |
| Chinese C-Eval (EM) | 78.6 | 79.5 | 86.1 | 61.5 | 76.7 | 76.0 | 86.5 |
| Chinese C-SimpleQA (Correct) | 48.5 | 54.1 | 48.4 | 50.4 | 51.3 | 59.3 | 64.8 |
Note: All models are evaluated in a configuration that limits the output length to 8K. Benchmarks containing fewer than 1000 samples are tested multiple times using varying temperature settings to derive robust final results. DeepSeek-V3 stands as the best-performing open-source model, and also exhibits competitive performance against frontier closed-source models.
Open Ended Generation Evaluation
| Model | Arena-Hard | AlpacaEval 2.0 |
|---|
| DeepSeek-V2.5-0905 | 76.2 | 50.5 |
| Qwen2.5-72B-Instruct | 81.2 | 49.1 |
| LLaMA-3.1 405B | 69.3 | 40.5 |
| GPT-4o-0513 | 80.4 | 51.1 |
| Claude-Sonnet-3.5-1022 | 85.2 | 52.0 |
| DeepSeek-V3 | 85.5 | 70.0 |
Note: English open-ended conversation evaluations. For AlpacaEval 2.0, we use the length-controlled win rate as the metric.
