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 | 16B |
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 | 8k |
| 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% | 48.1% |
MATHReasoning — accuracy on the original competition mathematics test set | 70.6% | 45.3% |
MMLU-ProKnowledge — accuracy on the ten-option MMLU-Pro benchmark | 66.3% | 42.4% |
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. | — | 14.3% |
IFEvalAgentic tasks — following verifiable instructions in generated responses | 82.9% | — |
WinoGrandeReasoning — resolving ambiguous pronouns in context | — | 74.6% |
GPQA DiamondScience — Graduate-level science questions in biology, physics, and chemistry — hard enough that subject-matter PhDs score around 65%. Higher is better. | — | 35.2% |
SimpleQAKnowledge — accuracy on short, fact-seeking questions in the original SimpleQA set | — | 5.6% |
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. | — | 70% |
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. | — | 77.4% |
| Overview | ||
| Company | Mistral | Moonshot AI |
| Release date | Jan 30 2025 | Feb 22 2025 |
| Access | Open Weight | Open Weight |
| Model details | View model | View model |
Other comparisons
Mistral Small 3vsClaude Fable 5.2MoonlightvsClaude Fable 5.2Mistral Small 3vsGPT-6.2 AstraMoonlightvsGPT-6.2 AstraMistral Small 3vsGemini 4 HydrogenMoonlightvsGemini 4 HydrogenMistral Small 3vsMuse EmberMoonlightvsMuse EmberMistral Small 3vsGrok 4.7MoonlightvsGrok 4.7Mistral Small 3vsDeepSeek-V4.1-Flash-1017MoonlightvsDeepSeek-V4.1-Flash-1017Frequently asked questions
Mistral Small 3 leads Moonlight on 3 of the 3 benchmarks they both report (HumanEval, MATH, MMLU-Pro). Mistral Small 3 shipped 23 days before Moonlight, so benchmark comparisons should account for the intervening progress.
Mistral Small 3 has 24B parameters, while Moonlight has 16B. Context windows are 32k (Mistral Small 3) vs 8k (Moonlight).
On HumanEval, Mistral Small 3 leads at 84.8% vs Moonlight at 48.1%. On MATH, Mistral Small 3 leads at 70.6% vs Moonlight at 45.3%. On MMLU-Pro, Mistral Small 3 leads at 66.3% vs Moonlight at 42.4%.