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. | 24B | — |
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. | 32k | 1M |
| API pricingUSD per 1M tokens · lower wins | ||
Cheapest inputLowest input rate across every provider on OpenRouter, the lab included. The cheapest endpoint may run a quantised build or a shorter context — see "Available from" on the model page. | $0.04DeepInfra | — |
Cheapest outputLowest output rate across every provider on OpenRouter, the lab included. May come from a different provider than the cheapest input. | $0.064DeepInfra | — |
| $14.00DeepInfra · fp8 | — | |
| Benchmarks | ||
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. | 84.8% | 86.6% |
IFEvalAgentic tasks — following verifiable instructions in generated responses | 82.9% | 76.3% |
MATHReasoning — accuracy on the original competition mathematics test set | 70.6% | 81.1% |
MMLU-ProKnowledge — accuracy on the ten-option MMLU-Pro benchmark | 66.3% | 64.5% |
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. | — | 37.8% |
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 | Mistral | Qwen |
| Release date | Jan 30 2025 | Nov 15 2024 |
| Access | Open Weight | Closed |
| Model details | View model | View model |
Other comparisons
Mistral Small 3vsClaude Fable 5.2Qwen2.5-Turbo (1M)vsClaude Fable 5.2Mistral Small 3vsGPT-6.2 AstraQwen2.5-Turbo (1M)vsGPT-6.2 AstraMistral Small 3vsGemini 4 NeonQwen2.5-Turbo (1M)vsGemini 4 NeonMistral Small 3vsMuse EmberQwen2.5-Turbo (1M)vsMuse EmberMistral Small 3vsGrok 4.7Qwen2.5-Turbo (1M)vsGrok 4.7Mistral Small 3vsDeepSeek-V4.1-Flash-1017Qwen2.5-Turbo (1M)vsDeepSeek-V4.1-Flash-1017Frequently asked questions
Mistral Small 3 and Qwen2.5-Turbo (1M) are evenly matched across the 4 benchmarks they both report (HumanEval, IFEval, MATH, MMLU-Pro). Qwen2.5-Turbo (1M) shipped 76 days before Mistral Small 3, so benchmark comparisons should account for the intervening progress.
Context windows are 32k (Mistral Small 3) vs 1M (Qwen2.5-Turbo (1M)). Mistral Small 3 is open weight, while Qwen2.5-Turbo (1M) is closed.
On HumanEval, Qwen2.5-Turbo (1M) leads at 86.6% vs Mistral Small 3 at 84.8%. On IFEval, Mistral Small 3 leads at 82.9% vs Qwen2.5-Turbo (1M) at 76.3%. On MATH, Qwen2.5-Turbo (1M) leads at 81.1% vs Mistral Small 3 at 70.6%. On MMLU-Pro, Mistral Small 3 leads at 66.3% vs Qwen2.5-Turbo (1M) at 64.5%.