Compare LLM Models
Select 2-5 models to compare side by side.
Select models to compare
0 / 5 selected
Top downloads
Select 2-5 models
| Attribute |
~deepseek
|
|---|---|
| Company | ~deepseek |
| Release Date | 2026-08-13 |
| Parameters (B) | 1600 |
| Architecture | MoE Transformer |
| Context Window | 1000000 |
| Input Modalities | text |
| Open Source | No |
| License | |
| Score | 36 |
| VRAM (GB) | 960 |
| Compute | 8x NVIDIA B200 192GB (single-node EP/DP; ~960 GB aggregate GPU memory for FP4+FP8 weights) |
| Benchmark: AGIEval | 83.1 |
| Benchmark: Agents-Last-Exam | 25.7 |
| Benchmark: AutomationBench | 31.8 |
| Benchmark: BBH | 87.5 |
| Benchmark: BrowseComp | 83.4 |
| Benchmark: C-Eval | 93.1 |
| Benchmark: CodeForces | 3206 |
| Benchmark: CyberGym | 83.3 |
| Benchmark: DROP | 88.7 |
| Benchmark: DeepSWE | 62.7 |
| Benchmark: GDPval-AA | 1554 |
| Benchmark: GPQA-Diamond | 90.1 |
| Benchmark: GSM8K | 92.6 |
| Benchmark: HLE | 42.7 |
| Benchmark: HellaSwag | 88 |
| Benchmark: HumanEval | 76.8 |
| Benchmark: IMO-AnswerBench | 89.8 |
| Benchmark: LiveCodeBench | 93.5 |
| Benchmark: MATH | 64.5 |
| Benchmark: MMLU | 90.1 |
| Benchmark: MMLU-Pro | 87.5 |
| Benchmark: NL2Repo-Bench | 61.5 |
| Benchmark: SWE-Bench-Multilingual | 76.2 |
| Benchmark: SWE-Bench-Pro | 55.4 |
| Benchmark: SWE-Bench-Verified | 80.6 |
| Benchmark: SimpleQA | 57.9 |
| Benchmark: Terminal-Bench-2.0 | 67.9 |
| Benchmark: Terminal-Bench-2.1 | 87.9 |
| Benchmark: Toolathlon-Verified | 74.1 |
| Pricing (Input) | 0.18 |
| Pricing (Output) | 8.00 |