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. | 7B/14B | 4B/30B/235B |
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. | 1M | 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. | 38.6% | 74.1% |
IFEvalAgentic tasks — following verifiable instructions in generated responses | 84.3% | 87.8% |
MMLU-ProKnowledge — accuracy on the ten-option MMLU-Pro benchmark | 63.3% | 84.4% |
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. | 88.4% | — |
Humanity's Last Exam · text only, no toolsMultidisciplinary reasoning — accuracy on the 2,158 text-only questions without tools | — | 18.2% |
MATHReasoning — accuracy on the original competition mathematics test set | 79.5% | — |
AIME 2025Reasoning — single-answer accuracy on the 2025 invitational mathematics exam without tools | — | 92.3% |
GPQA DiamondScience — Graduate-level science questions in biology, physics, and chemistry — hard enough that subject-matter PhDs score around 65%. Higher is better. | — | 81.1% |
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. | 94.8% | — |
| Overview | ||
| Company | Qwen | Qwen |
| Release date | Jan 27 2025 | Jul 21 2025 |
| Access | Open Weight | Open Weight |
| Model details | View model | View model |
Other comparisons
Qwen2.5-1MvsClaude Fable 5.2Qwen3-2507vsClaude Fable 5.2Qwen2.5-1MvsGPT-6.2 AstraQwen3-2507vsGPT-6.2 AstraQwen2.5-1MvsGemini 4 NeonQwen3-2507vsGemini 4 NeonQwen2.5-1MvsMuse EmberQwen3-2507vsMuse EmberQwen2.5-1MvsGrok 4.7Qwen3-2507vsGrok 4.7Qwen2.5-1MvsDeepSeek-V4.1-Flash-1017Qwen3-2507vsDeepSeek-V4.1-Flash-1017Frequently asked questions
Qwen3-2507 leads Qwen2.5-1M on 3 of the 3 benchmarks they both report (LiveCodeBench, IFEval, MMLU-Pro). Qwen2.5-1M shipped 175 days before Qwen3-2507, so benchmark comparisons should account for the intervening progress.
Qwen2.5-1M has 7B/14B parameters, while Qwen3-2507 has 4B/30B/235B. Context windows are 1M (Qwen2.5-1M) vs 256k (Qwen3-2507).
On LiveCodeBench, Qwen3-2507 leads at 74.1% vs Qwen2.5-1M at 38.6%. On IFEval, Qwen3-2507 leads at 87.8% vs Qwen2.5-1M at 84.3%. On MMLU-Pro, Qwen3-2507 leads at 84.4% vs Qwen2.5-1M at 63.3%.