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. | 671B | 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. | 128k | 256k |
| These models have no shared benchmark scores. | ||
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. | — | 68% |
SWE-Bench MultilingualMultilingual coding — Like SWE-Bench, but the coding problems span many programming languages, not just one. Tests how broadly the AI can code. Higher is better. | — | 55.7% |
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. | 65.2% | — |
Terminal-Bench 2.0Agentic terminal coding — Can the AI work in a command-line terminal — running commands and finishing technical setup tasks the way a developer would? (Version 2.0 of the test.) Higher is better. | — | 22.5% |
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. | 87.1% | — |
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. | 89.3% | — |
| Overview | ||
| Company | DeepSeek | Mistral |
| Release date | Dec 25 2024 | Dec 9 2025 |
| Access | Open Weight | Open Weight |
| Model details | View model | View model |
Other comparisons
DeepSeek V3 BasevsClaude Haiku 5.5Devstral Small 2 (24B)vsClaude Haiku 5.5DeepSeek V3 BasevsGPT-6.1 SolDevstral Small 2 (24B)vsGPT-6.1 SolDeepSeek V3 BasevsGemini 4 ArgonDevstral Small 2 (24B)vsGemini 4 ArgonDeepSeek V3 BasevsMuse Spark OpenDevstral Small 2 (24B)vsMuse Spark OpenDeepSeek V3 BasevsGrok 4.7Devstral Small 2 (24B)vsGrok 4.7DeepSeek V3 BasevsKimi K3Devstral Small 2 (24B)vsKimi K3Frequently asked questions
DeepSeek V3 Base and Devstral Small 2 (24B) don't publish scores on any of the same benchmarks, so there's no direct head-to-head comparison. DeepSeek V3 Base shipped 349 days before Devstral Small 2 (24B), so benchmark comparisons should account for the intervening progress.
DeepSeek V3 Base has 671B parameters, while Devstral Small 2 (24B) has 24B. Context windows are 128k (DeepSeek V3 Base) vs 256k (Devstral Small 2 (24B)).
Direct benchmark comparisons are unavailable — DeepSeek V3 Base and Devstral Small 2 (24B) don't publish scores on any of the same benchmarks.