Compare AI models
| Specifications | ||
ParametersThe learned values that shape how a model responds. B means billion and T means trillion. For mixture-of-experts models, the total count is larger than the number active for each token. Parameter count alone does not measure quality. | 236B | — |
Context windowHow much information the model can work with in one request, measured in tokens — pieces of words or other data. This includes your question, attachments, conversation and reply. Limits can vary by model variant and provider. | 128k | 1M |
| Benchmarks | ||
LiveCodeBenchCompetitive coding — Coding problems published so recently the AI can't have seen them in training — a contamination-free test of raw programming skill. Higher is better. | 41.8% | 37.8% |
HumanEvalFunction synthesis — 164 small Python problems: the AI is given a function's description and has to write the working function. This was the coding benchmark of the GPT-3.5 and GPT-4 era, before the field moved to fixing real bugs in real repositories. Higher is better. | 89% | 86.6% |
SWE-Bench VerifiedCoding — Real coding tasks pulled from open-source projects — the AI has to find and fix actual bugs. A human-checked version of the original SWE-Bench. Higher is better. | 16.8% | — |
IFEvalAgentic tasks — following verifiable instructions in generated responses | — | 76.3% |
MATHReasoning — accuracy on the original competition mathematics test set | — | 81.1% |
MMLU-ProKnowledge — accuracy on the ten-option MMLU-Pro benchmark | — | 64.5% |
GSM8KGrade-school math — Grade-school maths word problems that take a few steps of arithmetic to work through. It separated the models of 2022 and 2023 sharply, then saturated. One caveat on the historical numbers: OpenAI included part of the GSM8K training set in GPT-4's pre-training mix, so GPT-4's score is not a clean few-shot result. Higher is better. | — | 93.8% |
| Overview | ||
| Company | DeepSeek | Qwen |
| Release date | Sep 5 2024 | Nov 15 2024 |
| Access | Open Weight | Closed |
| Model details | View model | View model |
Other comparisons
DeepSeek V2.5vsClaude Fable 5.2Qwen2.5-Turbo (1M)vsClaude Fable 5.2DeepSeek V2.5vsGPT-6.2 AstraQwen2.5-Turbo (1M)vsGPT-6.2 AstraDeepSeek V2.5vsGemini 4 NeonQwen2.5-Turbo (1M)vsGemini 4 NeonDeepSeek V2.5vsMuse EmberQwen2.5-Turbo (1M)vsMuse EmberDeepSeek V2.5vsGrok 4.7Qwen2.5-Turbo (1M)vsGrok 4.7DeepSeek V2.5vsMistral Large 4Qwen2.5-Turbo (1M)vsMistral Large 4Frequently asked questions
DeepSeek V2.5 leads Qwen2.5-Turbo (1M) on 2 of the 2 benchmarks they both report (LiveCodeBench, HumanEval). DeepSeek V2.5 shipped 71 days before Qwen2.5-Turbo (1M), so benchmark comparisons should account for the intervening progress.
Context windows are 128k (DeepSeek V2.5) vs 1M (Qwen2.5-Turbo (1M)). DeepSeek V2.5 is open weight, while Qwen2.5-Turbo (1M) is closed.
On LiveCodeBench, DeepSeek V2.5 leads at 41.8% vs Qwen2.5-Turbo (1M) at 37.8%. On HumanEval, DeepSeek V2.5 leads at 89% vs Qwen2.5-Turbo (1M) at 86.6%.