Side-by-side benchmark comparison across agentic, coding, multimodal, knowledge, reasoning, and math workflows.
Mistral 8x7B v0.2 is clearly ahead on the aggregate, 32 to 27. The gap is large enough that you do not need to squint at the spreadsheet to see the difference.
Mistral 8x7B v0.2's sharpest advantage is in coding, where it averages 13.3 against 6.2. The single biggest benchmark swing on the page is SWE-bench Pro, 14 to 5. Ministral 3 3B does hit back in multilingual, so the answer changes if that is the part of the workload you care about most.
Ministral 3 3B gives you the larger context window at 128K, compared with 32K for Mistral 8x7B v0.2.
Pick Mistral 8x7B v0.2 if you want the stronger benchmark profile. Ministral 3 3B only becomes the better choice if multilingual is the priority or you need the larger 128K context window.
Mistral 8x7B v0.2
27.9
Ministral 3 3B
22.9
Mistral 8x7B v0.2
13.3
Ministral 3 3B
6.2
Mistral 8x7B v0.2
32.3
Ministral 3 3B
30.4
Mistral 8x7B v0.2
35.2
Ministral 3 3B
30.1
Mistral 8x7B v0.2
28.9
Ministral 3 3B
24.5
Mistral 8x7B v0.2
67
Ministral 3 3B
67
Mistral 8x7B v0.2
58.8
Ministral 3 3B
59.7
Mistral 8x7B v0.2
42
Ministral 3 3B
36
Mistral 8x7B v0.2 is ahead overall, 32 to 27. The biggest single separator in this matchup is SWE-bench Pro, where the scores are 14 and 5.
Mistral 8x7B v0.2 has the edge for knowledge tasks in this comparison, averaging 28.9 versus 24.5. Inside this category, FrontierScience is the benchmark that creates the most daylight between them.
Mistral 8x7B v0.2 has the edge for coding in this comparison, averaging 13.3 versus 6.2. Inside this category, SWE-bench Pro is the benchmark that creates the most daylight between them.
Mistral 8x7B v0.2 has the edge for math in this comparison, averaging 42 versus 36. Inside this category, AIME 2023 is the benchmark that creates the most daylight between them.
Mistral 8x7B v0.2 has the edge for reasoning in this comparison, averaging 35.2 versus 30.1. Inside this category, LongBench v2 is the benchmark that creates the most daylight between them.
Mistral 8x7B v0.2 has the edge for agentic tasks in this comparison, averaging 27.9 versus 22.9. Inside this category, OSWorld-Verified is the benchmark that creates the most daylight between them.
Mistral 8x7B v0.2 has the edge for multimodal and grounded tasks in this comparison, averaging 32.3 versus 30.4. Inside this category, OfficeQA Pro is the benchmark that creates the most daylight between them.
Mistral 8x7B v0.2 and Ministral 3 3B are effectively tied for instruction following here, both landing at 67 on average.
Ministral 3 3B has the edge for multilingual tasks in this comparison, averaging 59.7 versus 58.8. Inside this category, MMLU-ProX is the benchmark that creates the most daylight between them.
Get notified when new models drop, benchmark scores change, or the leaderboard shifts. One email per week.
Free. No spam. Unsubscribe anytime. We only store derived location metadata for consent routing.