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 | 141B |
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 | 64k |
| 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. | — | $2.00Mistral |
Cheapest outputLowest output rate across every provider on OpenRouter, the lab included. May come from a different provider than the cheapest input. | — | $6.00Mistral |
| — | $700Mistral · zdr | |
| 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. | 87.2% | 45.1% |
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. | 19% | — |
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. | 39.7% | — |
ARC-ChallengeReasoning — multiple-choice science questions from the AI2 Reasoning Challenge; unrelated to ARC-AGI | — | 91.3% |
HellaSwagReasoning — commonsense completion of short scenarios | — | 88.5% |
WinoGrandeReasoning — resolving ambiguous pronouns in context | — | 84.7% |
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. | — | 77.75% |
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. | — | 78.6% |
| Overview | ||
| Company | DeepSeek | Mistral |
| Release date | Jul 24 2024 | Apr 10 2024 |
| Access | Open Weight | Open Weight |
| Model details | View model | View model |
Other comparisons
DeepSeek Coder V2-0724vsClaude Fable 5.2Mixtral 8×22BvsClaude Fable 5.2DeepSeek Coder V2-0724vsGPT-6.2 AstraMixtral 8×22BvsGPT-6.2 AstraDeepSeek Coder V2-0724vsGemini 4 HydrogenMixtral 8×22BvsGemini 4 HydrogenDeepSeek Coder V2-0724vsMuse EmberMixtral 8×22BvsMuse EmberDeepSeek Coder V2-0724vsGrok 4.7Mixtral 8×22BvsGrok 4.7DeepSeek Coder V2-0724vsKimi K3.1 ProMixtral 8×22BvsKimi K3.1 ProFrequently asked questions
DeepSeek Coder V2-0724 leads Mixtral 8×22B on 1 of the 1 benchmark they both report (HumanEval). Mixtral 8×22B shipped 105 days before DeepSeek Coder V2-0724, so benchmark comparisons should account for the intervening progress.
DeepSeek Coder V2-0724 has 236B parameters, while Mixtral 8×22B has 141B. Context windows are 128k (DeepSeek Coder V2-0724) vs 64k (Mixtral 8×22B).
On HumanEval, DeepSeek Coder V2-0724 leads at 87.2% vs Mixtral 8×22B at 45.1%.