Ministral 3 8B
Ministral 3 8B is behind the leaders in multimodal tasks. Too few results yet to rate reasoning, coding, agentic tasks, safety, long context, math, multilingual tasks, instruction following, or factuality.
Capability profile
Bars show the model's median result as a share of the leading model's, per capability. Select a row to see its results.
Too few results yet to rate Reasoning, Coding, Agentic, Safety, Long Context, Math, Multilingual, Instruction Following or Factuality.
Price
$0.15input$0.15outputper million tokens
From Artificial Analysis · 2 providers tracked · All prices
Evidence
38results on36benchmarks
- 5 independently verified
- 19 aggregator
- 14 vendor-reported
From 5 sources · latest Oct 8, 2026 · How verification works
Ministral 3 8B benchmark results
38 results on 36 benchmarks, grouped by capability. Each bar is the result as a share of the capability leader's; every result links to its source.
41.0% behind the leader1 of 6 ranked benchmarks measured
- MMMU-Pro45.95Oct 8, 2026aa_mmmu_pro
0 of 6 ranked benchmarks measured
- 0.00Oct 8, 2026
- GPQA Diamond47.07Oct 8, 2026gpqa
- Humanity's Last Exam4.26Oct 8, 2026aa_hle
Show 1 more reasoning resultHide 1 reasoning result
- 66.80Jun 12, 2026
0 of 10 ranked benchmarks measured
- SciCode20.72Oct 8, 2026aa_scicode
- Terminal-Bench 2.14.12Oct 8, 2026terminalbenchV21
- Terminal-Bench Hard4.55Oct 8, 2026aa_terminalbench_hard
0 of 7 ranked benchmarks measured
- 0.00Oct 8, 2026
- Terminal-Bench 4.00.00Oct 8, 2026
- τ-Bench V3 · Banking3.71Oct 8, 2026tauBanking
0 of 3 ranked benchmarks measured
- 25.67Oct 8, 2026
0 of 3 ranked benchmarks measured
- IFBench29.12Oct 8, 2026aa_ifbench
0 of 4 ranked benchmarks measured
- 21.70May 2, 2026
- AA-Omniscience · Accuracy12.98Oct 8, 2026omniscienceAccuracy
- AA-Omniscience · Non-hallucination5.84Oct 8, 2026omniscienceNonHallucination
More results
Benchmarks outside the capability baskets. They are not ranked against a leader.
- AA Intelligence5.00Jun 24, 2026Artificial Analysis Intelligence Index
- vectara_avg_summary_length139.40May 2, 2026Average Summary Length (Words)
- vectara_answer_rate99.10May 2, 2026Answer Rate
- vectara_factual_consistency78.30May 2, 2026Factual Consistency Rate
- AA Intelligence5.48Oct 8, 2026aa_intelligence_index
- τ²-Bench Telecom (AA run)26.61Oct 8, 2026aa_tau2
Show 16 more resultsHide 16 results
- 0.62Sep 9, 2026
- -68.95Oct 8, 2026
- 59.10Jun 5, 2026
- AIME 202486.00Jun 5, 2026AIME24
- AIME 202578.70Jun 12, 2026AIME25
- 50.90Jun 12, 2026
- Artificial Analysis Coding Index9.69Sep 9, 2026aa_coding_index
- 61.60Jun 12, 2026
- 62.60Jun 5, 2026
- 87.60Jun 12, 2026
- 8.08Jun 12, 2026
- MMLU76.10Jun 5, 2026MMLU 5-shot
- 79.30Jun 5, 2026
- 70.60Jun 5, 2026
- 68.10Jun 5, 2026
- 66.80Jun 12, 2026
Ministral 3 8B: common questions
Who makes Ministral 3 8B?
Ministral 3 8B is made by Mistral.
When was Ministral 3 8B released?
Ministral 3 8B was released on Dec 2, 2025, according to Artificial Analysis.
What is Ministral 3 8B good at?
Ministral 3 8B is behind the leaders in multimodal tasks. Too few results yet to rate reasoning, coding, agentic tasks, safety, long context, math, multilingual tasks, instruction following, or factuality.
How much does Ministral 3 8B cost?
Ministral 3 8B costs $0.15 per million input tokens and $0.15 per million output tokens, according to Artificial Analysis. We track its price at 2 providers. At a mix of three input tokens to one output token, it is cheaper than 85% of the 331 priced models we track.
How many benchmarks has Ministral 3 8B been tested on?
We track 38 results for Ministral 3 8B on 36 benchmarks from 5 sources, 5 of them independently verified. The latest was recorded on Oct 8, 2026.
About this record
Where Ministral 3 8B's numbers come from, and every name it appears under.
- Tracked since
- Apr 27, 2026
- Newest source mention
- Apr 27, 2026
Where the results come from
Verification: 38 scores · 5 independently verified · 19 aggregator-attributed · 14 vendor-reported. How these tiers are assigned
From 5 sources on 3 sites. Artificial Analysis supplies 20 of them; the 5 independently verified results come from 2 sites. Bars are coloured by trust tier.
- artificialanalysis.ai20
- huggingface.co14
- raw.githubusercontent.com4