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 | 2.4T |
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 |
| 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. | $0.351Cloudflare | $2.00Alibaba |
Cheapest outputLowest output rate across every provider on OpenRouter, the lab included. May come from a different provider than the cheapest input. | $0.555Cloudflare | $6.00Alibaba |
| $97.95Cloudflare | $700Alibaba | |
| Benchmarks | ||
MMMU-ProMultimodal reasoning — A tougher version of MMMU — college-level questions that mix images, diagrams, and text together. Higher is better. | 49.25% | 82.7% |
DeepSWE 1.1Agentic coding — Artificial Analysis' independent test of deep, agentic software-engineering work — the AI has to plan and carry out substantial coding tasks end to end. (Version 1.1 of the test.) Higher is better. | — | 69.3% |
NL2Repo-BenchRepo-level code generation — Tests whether the AI can turn a natural-language requirement into working code across an entire repository, not just produce a single function or patch. Higher is better. | — | 64.9% |
QwenSWEBench V2Software engineering — Qwen's in-house coding benchmark, second version, built around complex real-world software-engineering tasks. Scores are not comparable with the first version. Higher is better. | — | 70% |
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. | 88.41% | — |
Terminal-Bench 4.0Agentic terminal coding — Can the AI work in a command-line terminal — running commands and finishing technical setup tasks the way a developer would? Version 4.0 recalibrated how much time, CPU and memory each task gets, removed eight tasks and fixed nineteen, so fewer runs fail for reasons that have nothing to do with the model. Scores are not comparable with earlier versions. Higher is better. | — | 26.97% |
Terminal-Bench 3.0Agentic terminal coding — command-line task completion (v3.0, much harder task set) | — | 29% |
JobBenchProfessional tool use — Tests the AI on professional workplace tasks that require using real work tools — the kind of multi-step jobs an office worker handles. Higher is better. | — | 64% |
CoWorkBenchLong-horizon office work — Tests long-running office tasks across fields including computer science, finance, law, medicine, and other productivity work. Higher is better. | — | 76.1% |
Toolathlon-VerifiedPersonal tool use — Tests how well the AI uses everyday personal tools and apps to get things done — a human-checked version of Toolathlon. Higher is better. | — | 73.3% |
GPQA DiamondScience — Graduate-level science questions in biology, physics, and chemistry — hard enough that subject-matter PhDs score around 65%. Higher is better. | 45.96% | — |
MMLUGeneral knowledge — A 57-subject multiple-choice exam — history, law, medicine, maths — that was the standard measure of how much a model knows from 2020 until roughly 2024, when frontier scores crowded into the high 80s and labs moved on to harder tests. The scores here were published years apart under different testing setups, so read them as a historical record rather than a like-for-like ranking. Higher is better. | 80.62% | — |
AutomationBenchBusiness workflows — Tests whether the AI can run real multi-step business workflows — the kind of end-to-end office processes companies want to automate — from start to finish. Higher is better. | — | 50.8% |
MMMUMultimodal — Tests the AI on understanding images and text together across many college subjects. Higher is better. | 64% | — |
| Overview | ||
| Company | Mistral | Qwen |
| Release date | Mar 17 2025 | Sep 2 2026 |
| Access | Open Weight | Closed |
| Model details | View model | View model |
Other comparisons
Mistral Small 3.1vsClaude Haiku 5.5Qwen3.8-Max-0902vsClaude Haiku 5.5Mistral Small 3.1vsGPT-6.1 TerraQwen3.8-Max-0902vsGPT-6.1 TerraMistral Small 3.1vsGemini 4 CarbonQwen3.8-Max-0902vsGemini 4 CarbonMistral Small 3.1vsMuse Spark OpenQwen3.8-Max-0902vsMuse Spark OpenMistral Small 3.1vsGrok 4.7Qwen3.8-Max-0902vsGrok 4.7Mistral Small 3.1vsDeepSeek-V4.1-Flash-1017Qwen3.8-Max-0902vsDeepSeek-V4.1-Flash-1017Frequently asked questions
Qwen3.8-Max-0902 leads Mistral Small 3.1 on 1 of the 1 benchmark they both report (MMMU-Pro). Mistral Small 3.1 shipped 534 days before Qwen3.8-Max-0902, so benchmark comparisons should account for the intervening progress.
Mistral Small 3.1 has 24B parameters, while Qwen3.8-Max-0902 has 2.4T. Context windows are 128k (Mistral Small 3.1) vs 1M (Qwen3.8-Max-0902). Mistral Small 3.1 is open weight, while Qwen3.8-Max-0902 is closed.
On MMMU-Pro, Qwen3.8-Max-0902 leads at 82.7% vs Mistral Small 3.1 at 49.25%.