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 | 0.8B/2B/4B/9B |
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 | 256k |
| 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% | 65.6% |
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% | — |
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% | — |
IFEvalAgentic tasks — following verifiable instructions in generated responses | — | 91.5% |
GPQA DiamondScience — Graduate-level science questions in biology, physics, and chemistry — hard enough that subject-matter PhDs score around 65%. Higher is better. | — | 81.7% |
IFBenchInstruction following — Tests whether the AI can follow detailed instructions and satisfy multiple constraints at once. Higher is better. | — | 64.5% |
MMLU-ProKnowledge — accuracy on the ten-option MMLU-Pro benchmark | — | 82.5% |
OSWorld-VerifiedAgentic computer use — Can the AI actually operate a computer — clicking, typing, and using real apps — to finish tasks on its own? Higher is better. | — | 41.8% |
CharXiv ReasoningChart reasoning — Can the AI read and reason about complex charts and figures, not just text? Higher is better. | — | 73% |
MMMU-ProMultimodal reasoning — A tougher version of MMMU — college-level questions that mix images, diagrams, and text together. Higher is better. | — | 70.1% |
MMMUMultimodal — Tests the AI on understanding images and text together across many college subjects. Higher is better. | — | 78.4% |
| Overview | ||
| Company | DeepSeek | Qwen |
| Release date | Sep 5 2024 | Mar 2 2026 |
| Access | Open Weight | Open Weight |
| Model details | View model | View model |
Other comparisons
DeepSeek V2.5vsClaude Fable 5.2Qwen3.5 SmallvsClaude Fable 5.2DeepSeek V2.5vsGPT-6.2 AstraQwen3.5 SmallvsGPT-6.2 AstraDeepSeek V2.5vsGemini 4 NeonQwen3.5 SmallvsGemini 4 NeonDeepSeek V2.5vsMuse EmberQwen3.5 SmallvsMuse EmberDeepSeek V2.5vsGrok 4.7Qwen3.5 SmallvsGrok 4.7DeepSeek V2.5vsMistral Large 4Qwen3.5 SmallvsMistral Large 4Frequently asked questions
Qwen3.5 Small leads DeepSeek V2.5 on 1 of the 1 benchmark they both report (LiveCodeBench). DeepSeek V2.5 shipped 543 days before Qwen3.5 Small, so benchmark comparisons should account for the intervening progress.
DeepSeek V2.5 has 236B parameters, while Qwen3.5 Small has 0.8B/2B/4B/9B. Context windows are 128k (DeepSeek V2.5) vs 256k (Qwen3.5 Small).
On LiveCodeBench, Qwen3.5 Small leads at 65.6% vs DeepSeek V2.5 at 41.8%.