Models / MiniMax-M3

MiniMax

MiniMax-M3

MiniMax-M3 is best for reasoning and knowledge; a strong pick at a modest price. The cheapest model of similar intelligence is MiMo-V2.6-Flash, at $0.18 per million tokens against $0.52.

Results 16 Nov 2026 · 34 results · 5 sources

Test MiniMax-M3 on your prompt → It runs beside the models you could use instead, with every answer checked and costed.

Price per million tokens
$0.30 in · $1.20 out
MiniMax's list price, 10 Nov 2026 · compare prices$0.06 cached input
Speed
146 tokens a second; on its test prompt the answer starts after 14.6 s, thinking time included
Artificial Analysis, on its usual API
Intelligence Index
29.2
Artificial Analysis
Spring Prompt overall
Not ranked yet
needs our benchmarks and two groups of results
Developer
MiniMax, based in China
Weights published
Released
1 Jun 2026
1,048,576 tokens of context
Pricing and availability
Shown as a permanent 50% off $0.60 / $2.40

Where it stands among the models you could choose

Available models grouped by price, intelligence, speed and developer location, plus models with published weights. Retired models are excluded.

AmongIntelligence IndexSpring Prompt overallPrice, blendedSpeed
Models available today 56th of 185Claude Opus 5.5 – $0.52 blended, 38th percentilecheapest: Mistral Nemo 22nd of 65Nemotron 3.5 Super VL
Models priced $0.25–1 per million tokens 8th of 61MiMo-V2.6-Pro – sets this group 10th of 20Gemini 3.5 Flash-Lite
Models with similar intelligence sets this group – $0.52 blended, 12th percentilecheapest: Nemotron 3.5 Super VL 4th of 10Nemotron 3.5 Super VL
Models writing 100–200 tokens a second 11th of 21Claude Sonnet 5.5 – $0.52 blended, 45th percentilecheapest: Qwen3.5-9B sets this group
Models from developers based in China 18th of 75MiMo-V2.6-Pro – $0.52 blended, 47th percentilecheapest: Qwen3.5-9B 5th of 23DeepSeek-V4.1-Flash
Open-weights models 19th of 104MiMo-V2.6-Pro – $0.52 blended, 56th percentilecheapest: Mistral Nemo 14th of 43Nemotron 3.5 Super VL

The name under each rank is the leader of that group; green is the top quarter of the group and red the bottom quarter, by rank. Price is shown as a percentile: the share of the group that costs less, so lower is cheaper. "Sets this group" marks the measure the group is defined by. Spring Prompt overall is our score out of 100 across every source (how it works). Price is the developer's list price (or the typical OpenRouter host where we haven't read one) for three input tokens to one output; intelligence and speed are from Artificial Analysis (data sourced from Artificial Analysis); "similar intelligence" means within 4 points on its Intelligence Index. Developer locations are where each company is based, not where a model is served. Groups under 3 models are not ranked.

How good is it, and for what?

Each use case ranks the available models its benchmarks measured, then those priced $0.25–1 per million tokens. The bar shows its percentile in that field, best to the right. Open a row for the results behind it.

Product listingsTurning a sparse product feed and photos into listings that can go live Not measured
  • CatalogBench: has not measured this model.
Decks from an analysisTurning a finished analysis into a deck you could present as it is Not measured
  • DeckBench: has not measured this model.
User surveysPlanning a user survey and reading its results without being misled Not measured
  • SurveyBench: has not measured this model.
Marketing planningPlanning a year of ad spend without overspending Not measured
  • ROASBench: has not measured this model.
Agents and tool useMulti-step tasks with tools: support desks, coding agents, function calls 79th of 159 available · 15th of 47 at its price
BenchmarkMiniMax-M3Rank among availableBest available
OpenHands IndexOpenHands Index: average · reported 57.2% 13th of 31 Claude Fable 5 81.0%
Artificial AnalysisArtificial Analysis: Terminal-Bench 4.0 · reported 2.0% 52nd of 107 Kimi K3.1 Pro 67.2%
Artificial AnalysisArtificial Analysis: Terminal-Bench 2.1 · reported 65.2% 38th of 107 Claude Fable 5.1 91.4%
Artificial AnalysisArtificial Analysis: τ-bench banking · reported 15.3% 54th of 104 Grok 4.6 50.7%
Vending-Bench 2Vending-Bench 2: money after a year · reported $2,158 38th of 53 GPT-6 Astra $15,515
  • tau2-bench: pending. Its latest results are dated 26 May 2026, before this model was released.
  • Berkeley Function Calling Leaderboard (BFCL) V4: pending. Its latest results are dated 16 Dec 2025, before this model was released.
  • Microsoft STATE-Bench: pending. Its latest results are dated 29 May 2026, before this model was released.
Professional workReal tasks from banking, consulting and law, business documents and freelance projects 39th of 51 available · 2nd of 3 at its price
BenchmarkMiniMax-M3Rank among availableBest available
APEX-AgentsAPEX-Agents: tasks passed · reported 37.7% 30th of 37 Claude Sonnet 5.5 75.5%
  • GDP.pdf: has not measured this model.
  • Remote Labor Index: has not measured this model.
Reasoning and knowledgeHard questions across science, maths and general knowledge 45th of 187 available · 6th of 61 at its price
BenchmarkMiniMax-M3Rank among availableBest available
Artificial AnalysisArtificial Analysis: Artificial Analysis Intelligence Index · reported 29.2 56th of 185 Claude Opus 5.5 57.6
Artificial AnalysisArtificial Analysis: Humanity's Last Exam · reported 39.0% 48th of 186 Claude Opus 5.5 61.4%
Artificial AnalysisArtificial Analysis: GPQA Diamond · reported 92.9% 14th of 168 GPT-6 Astra 96.3%
WritingWhat people prefer in blind comparisons, and judged writing quality 67th of 169 available · 11th of 58 at its price
BenchmarkMiniMax-M3Rank among availableBest available
Arena (formerly LMArena)Arena Text: overall · reported 1,440 64th of 153 Gemini 4 Argon 1,525
  • UGI Leaderboard: has not measured this model.
Sticking to the factsSummarising without inventing things, and factual answers 91st of 139 available · 19th of 40 at its price
BenchmarkMiniMax-M3Rank among availableBest available
Arena (formerly LMArena)Arena Text factuality: overall · reported 1,434 70th of 109 Gemini 4 Argon 1,511
  • Vectara Hallucination Leaderboard: has not measured this model.
  • SimpleQA Verified (Epoch AI): has not measured this model.
Speed under pressureGood decisions against a real clock (fast chess) Not measured
  • BulletBench: has not measured this model.

Nearest alternatives

Against MiniMax-M3 at $0.52 blended, Intelligence Index 29.2. "Similar intelligence" means within 4 points or better.

Where to run it

9 hosts on OpenRouter at the standard service tier, with a median of $0.53 blended per million tokens; quantisation where the host states it: FP4, FP8. Cheapest first.

HostQuantisationInputOutputBlended
CoreWeaveFP4$0.23$0.96$0.41
GMICloudFP8$0.24$0.96$0.42
DeepInfraFP8$0.28$1.10$0.49
Minimaxnot stated$0.30$1.20$0.52
NovitaFP8$0.30$1.20$0.52
ParasailFP8$0.30$1.20$0.52
StreamLakeFP8$0.30$1.20$0.52
VeniceFP8$0.30$1.20$0.52

Other service tiers, not counted above: Parasail fast $0.45 in, $1.80 out; DeepInfra flex $0.22 in, $0.88 out. Flex and batch are cheaper but slower or queued; priority and fast cost more.

Prices per million tokens from OpenRouter's public endpoint list, checked 10 Nov 2026; the 1 dearer hosts are left out.

Against the alternatives

The models you would most likely weigh it against: the leaders of the groups above.

Scroll sideways to see every alternative.

MiniMax-M3MiMo-V2.6-Pro
most intelligent at $0.25–1 per million tokens
Nemotron 3.5 Super VL
cheapest with similar intelligence
Claude Opus 5.5
most intelligent available today
Price per million tokens $0.52 $0.54$0.24$8.00
Tokens a second 146 4341196
Intelligence Index 29.2 46.326.557.6
Spring Prompt overall – ––82
Rank among available models, by use case
Product listings – ––13th
Decks from an analysis – ––7th
User surveys – ––4th
Marketing planning – ––9th
Agents and tool use 79th of 159 23rd132nd3rd
Professional work 39th of 51 ––3rd
Reasoning and knowledge 45th of 187 13th74th1st
Writing 67th of 169 23rd–5th
Sticking to the facts 91st of 139 10th–3rd
Speed under pressure – –––

Shaded figures are better than MiniMax-M3's. A dash means no result.

Research and examples

Where it has been measured

8th most intelligent of the 61 models priced $0.25–1 per million tokens. Among models available today, in the top quarter (by percentile) for reasoning and knowledge; in the bottom quarter for professional work.

Every result

34 published results from 5 sources, each in the source's own units, with the configuration that produced it.

APEX-Agents · reported by APEX-Agents · 1 result
MeasureValueRankConfigurationDated
APEX-Agents: tasks passed% of tasks, higher is better 37.7% 32 of 39 MiniMax-M3 1 Oct 2026

Not shown: Not your firm's documents or tools; graded by rubric, not by a client.

Arena (formerly LMArena) · reported by Arena (formerly LMArena) · 12 results
MeasureValueRankConfigurationDated
Arena Agent: confirmed task successIPS effect estimate, higher is better -0.14
-0.16–-0.11
46 of 50 MiniMax-M3 (arena agent) 2 Oct 2026
Arena Agent: praise over complaintIPS effect estimate, higher is better -0.14
-0.17–-0.12
46 of 50 MiniMax-M3 (arena agent) 2 Oct 2026
Arena Agent: steerabilityIPS effect estimate, higher is better -0.07
-0.09–-0.06
44 of 50 MiniMax-M3 (arena agent) 2 Oct 2026
Arena Agent: tool groundingIPS effect estimate, higher is better 0.00
-0.00–0.00
34 of 50 MiniMax-M3 (arena agent) 2 Oct 2026
Arena Document: overallArena rating, higher is better 1,433
1,425–1,441
38 of 44 MiniMax-M3 13 Sep 2026
Arena Text factuality: overallArena rating, higher is better 1,434
1,431–1,437
103 of 182 MiniMax-M3 2 Oct 2026
Arena Text: business, management and financeArena rating, higher is better 1,448
1,442–1,455
78 of 406 MiniMax-M3 2 Oct 2026
Arena Text: creative writingArena rating, higher is better 1,406
1,399–1,413
99 of 411 MiniMax-M3 2 Oct 2026
Arena Text: expert promptsArena rating, higher is better 1,470
1,461–1,478
90 of 363 MiniMax-M3 2 Oct 2026
Arena Text: instruction followingArena rating, higher is better 1,434
1,429–1,440
88 of 413 MiniMax-M3 2 Oct 2026
Arena Text: overallArena rating, higher is better 1,440
1,436–1,444
96 of 413 MiniMax-M3 2 Oct 2026
Arena Text: writing, literature and languageArena rating, higher is better 1,422
1,416–1,428
90 of 412 MiniMax-M3 2 Oct 2026

Not shown: Not a success rate: the estimated effect of choosing this model as the agent's orchestrator, against an even mix of the models tested, inside Arena's own harness.

Artificial Analysis · reported by Artificial Analysis · 14 results
MeasureValueRankConfigurationDated
Artificial Analysis: Artificial Analysis Coding Indexindex score, higher is better 58.6 78 of 187 MiniMax-M3 20 Oct 2026
Artificial Analysis: Artificial Analysis Intelligence Indexindex score, higher is better 29.2 123 of 413 MiniMax-M3 20 Oct 2026
Artificial Analysis: GPQA Diamond% of questions, higher is better 92.9% 27 of 360 MiniMax-M3 20 Oct 2026
Artificial Analysis: Humanity's Last Exam% of questions, higher is better 39.0% 95 of 419 MiniMax-M3 20 Oct 2026
Artificial Analysis: IFBench% of instructions, higher is better 82.9% 3 of 288 MiniMax-M3 20 Oct 2026
Artificial Analysis: Long-context reasoning (AA-LCR)% of questions, higher is better 83.0% 23 of 407 MiniMax-M3 20 Oct 2026
Artificial Analysis: Output speedtokens per second, higher is better 146 39 of 137 MiniMax-M3 20 Oct 2026
Artificial Analysis: SciCode% of problems, higher is better 47.1% 135 of 205 MiniMax-M3 20 Oct 2026
Artificial Analysis: Terminal-Bench 2.1% of tasks, higher is better 65.2% 76 of 188 MiniMax-M3 20 Oct 2026
Artificial Analysis: Terminal-Bench 4.0% of tasks, higher is better 2.0% 125 of 204 MiniMax-M3 20 Oct 2026
Artificial Analysis: Terminal-Bench Hard% of tasks, higher is better 42.4% 45 of 282 MiniMax-M3 20 Oct 2026
Artificial Analysis: Time to first answer tokenseconds, lower is better 14.6 72 of 137 MiniMax-M3 20 Oct 2026
Artificial Analysis: τ-bench banking% of tasks, higher is better 15.3% 102 of 175 MiniMax-M3 20 Oct 2026
Artificial Analysis: τ²-bench telecom% of tasks, higher is better 88.9% 60 of 286 MiniMax-M3 20 Oct 2026

Not shown: Not your codebase or tools.

OpenHands Index · reported by OpenHands Index · 6 results
MeasureValueRankConfigurationDated
OpenHands Index: average% resolved, higher is better 57.2% 13 of 34 MiniMax-M3 (openhands) 30 Jun 2026
OpenHands Index: front end (SWE-Bench Multimodal)% resolved, higher is better 36.8% 11 of 34 MiniMax-M3 (openhands) 30 Jun 2026
OpenHands Index: greenfield (Commit0)% resolved, higher is better 25.0% 14 of 34 MiniMax-M3 (openhands) 30 Jun 2026
OpenHands Index: information gathering (GAIA)% resolved, higher is better 66.7% 15 of 34 MiniMax-M3 (openhands) 30 Jun 2026
OpenHands Index: issue resolution (SWE-Bench)% resolved, higher is better 76.4% 9 of 34 MiniMax-M3 (openhands) 30 Jun 2026
OpenHands Index: testing (SWT-Bench)% resolved, higher is better 81.1% 4 of 34 MiniMax-M3 (openhands) 30 Jun 2026

Not shown: Not the model alone: the score is for the model inside the OpenHands agent, at the SDK version shown.

Vending-Bench 2 · reported by Vending-Bench 2 · 1 result
MeasureValueRankConfigurationDated
Vending-Bench 2: money after a yearUS dollars, higher is better $2,158 45 of 63 MiniMax-M3 1 Oct 2026

Not shown: Not a real business; one simulated market with set rules.

Will MiniMax-M3 work for your product?

Benchmarks aren't your task. Paste the prompt your product uses and run it beside the models you could use instead, with every answer checked and costed.

New here? Your email starts a free trial. No credit card needed.