MiniMax M3 released
MiniMax · Model release
Source confirmed
Data as of October 14, 2026 · How the score is built
CurrentOpen WeightNon-Reasoning
Released Jun 1, 20261M context
MiniMax M3
Decision readingMiniMax M3 scores 55 out of 100 and ranks #72 of 221. This profile shows 27 source-displayable benchmark rows; its strongest eligible category is Multimodal & Grounded at #43. API pricing is $0.3 input and $1.2 output per million tokens, with cached input at $0.06.
Released Jun 1, 2026 — see all recent releases
MiniMax M3 will be repriced, updated or retired. Get each notice with its source and date. Follow model changes
Decision snapshot
Each value carries a field reference instead of floating alone. Markers compare this model with the current ranked and priced catalog; they are not absolute quality thresholds.
Capability
55/100
field median 50#72 of 221 ranked models
Public
#72of 221
Verified #44 of 89
Price
$0.30input / $1.20 output
input median $0.78cached $0.060 · blended $0.75
Speed
99tok/s
field median 86 tok/sFirst token 21.57 s
Context
1Mtokens
field median 256,000Reported for this model; direct source link not stored
Strongest published evidence
Multimodal & Grounded ranks #43. Particularly strong for screenshots, documents, charts, and grounded multimodal workflows.
Validate before choosing
27 published rows leave some tracked benchmark slots empty. Knowledge is its lowest eligible category at #80.
Source-linked · 27 displayable benchmark rows
Follow model changesCategory score record
Scores and ranks appear only where published evidence can be displayed. The table keeps the score, weight, cohort, and evidence state together.
| Category | Score | Rank | Percentile | Weight | Benchmarks | Evidence |
|---|---|---|---|---|---|---|
| AgenticRank #66 of 125Percentile 48thWeight 22%9 benchmarksVerified | 39.6 | 9 benchmarks | Verified | |||
| CodingRank #73 of 149Percentile 51stWeight 20%10 benchmarksVerified | 36.5 | 10 benchmarks | Verified | |||
| ReasoningWeight 17%0 benchmarksNot measured | Not measured | 0 benchmarks | Not measured | |||
| MultimodalRank #43 of 54Percentile 21stWeight 12%5 benchmarksVerified | 61.5 | 5 benchmarks | Verified | |||
| KnowledgeRank #80 of 181Percentile 56thWeight 12%2 benchmarksVerified | 49.6 | 2 benchmarks | Verified | |||
| MultilingualWeight 7%0 benchmarksNot measured | Not measured | 0 benchmarks | Not measured | |||
| Inst. FollowingWeight 5%0 benchmarksNot measured | Not measured | 0 benchmarks | Not measured | |||
| MathRank Not rankedWeight 5%1 benchmarkVerified | 85.7 | 1 benchmark | Verified |
27 of 671 tracked benchmark slots have displayable evidence · bars run 0–100
Coverage detailsHow much of this is verified
Coverage is split by category so a strong number never hides a thin evidence base. Verified means the row is tied to a published source; provisional rows remain visible but separate.
- Agentic9/9 verified
- Coding10/10 verified
- ReasoningNot measured
- Multimodal5/5 verified
- Knowledge2/2 verified
- MultilingualNot measured
- Inst. FollowingNot measured
- Math1/1 verified
Verified sourceProvisionalNot measured
Capability shape
Each axis shows percentile within that category’s eligible cohort. The comparison outline is the median of the six nearest public-score peers; a collapsed vertex means the category is not rank-eligible.
MiniMax M3 category percentile values
- Agentic48th percentile
- Coding51st percentile
- ReasoningNot eligible
- Multimodal21st percentile
- Knowledge56th percentile
- MultilingualNot eligible
- Instruction followingNot eligible
- MathNot eligible
The dashed outline is median of 6 nearest peers.
Eligible category ranks
- Agentic#66/125
- Coding#73/149
- ReasoningNot ranked
- Multimodal#43/54
- Knowledge#80/181
- MultilingualNot ranked
- Inst. FollowingNot ranked
- MathNot ranked
Top decileTop quartileMid-fieldNot eligible
Benchmark ledger
Coding opens by default. The marker compares each value with the best source-verified result in the catalog; provisional leaders do not set the reference. Expand the remaining categories for every published row.
Coding10 rows
| Benchmark | Score | Versus best verified row | Gap | Weight | Evidence |
|---|---|---|---|---|---|
| SWE-bench Pro | Score59% | Versus best verified row Best verified: Claude Opus 5.5 · 89.9% | Gap30.9 behind | Weight26% ref. weight | Provider exact |
| LiveCodeBench (Vals)LiveCodeBench, Vals AI run | Score82.2% | Versus best verified row Best verified: Claude Fable 5.1 · 90.5% | Gap8.3 behind | Weight8% ref. weight | |
| Terminal-Bench 2.1 | Score66.0% | Versus best verified row Best verified: SWE-2 · 92.8% | Gap26.8 behind | WeightScored in Agentic | Provider exact |
| SWE-bench VerifiedSoftware Engineering Benchmark Verified | Score80.5% | Versus best verified row Best verified: Claude Opus 5 · 96% | Gap15.5 behind | WeightDisplay only | Provider exact |
| NL2Repo | Score42.1% | Versus best verified row Best verified: DeepSeek V4.1 Flash · 65.4% | Gap23.3 behind | WeightDisplay only | Provider exact |
| VIBE V2 | Score50.1% | Versus best verified row Best verified: MiniMax M3 · 50.1% | GapBest verified | WeightDisplay only | Provider exact |
| SVG-Bench | Score63.7% | Versus best verified row Best verified: MiniMax M3 · 63.7% | GapBest verified | WeightDisplay only | Provider exact |
| KernelBench Hard | Score28.8% | Versus best verified row Best verified: MiniMax M3 · 28.8% | GapBest verified | WeightDisplay only | Provider exact |
| OpenHarmony BenchOpenHarmony Bench v1.0 | Score48.4% | Versus best verified row Best verified: Qwen3.8 Max · 60.8% | Gap12.4 behind | WeightDisplay only | Benchmark exact |
| SWE-bench (Vals)SWE-bench, Vals AI run | Score75.0% | Versus best verified row Best verified: Claude Opus 5 · 97.0% | Gap22 behind | WeightDisplay only | Verified |
Agentic9 rows
| Benchmark | Score | Versus best verified row | Gap | Weight | Evidence |
|---|---|---|---|---|---|
| Terminal-Bench 2.1 | Score66.0% | Versus best verified row Best verified: SWE-2 · 92.8% | Gap26.8 behind | Weight8% ref. weight | Provider exact |
| BrowseComp | Score83.5% | Versus best verified row Best verified: Atria Dawn Preview · 92.5% | Gap9 behind | Weight8% ref. weight | Provider exact |
| OSWorld-Verified | Score70.1% | Versus best verified row Best verified: Qwen3.8 Max · 86.1% | Gap16 behind | Weight7% ref. weight | Provider exact |
| MCP Atlas | Score74.2% | Versus best verified row Best verified: Muse Spark 1.1 · 88.1% | Gap13.9 behind | Weight4% ref. weight | Provider exact |
| Terminal-Bench 2.1 (Vals)Terminal-Bench 2.1, Vals AI run | Score53.6% | Versus best verified row Best verified: GPT-6 Astra · 87.3% | Gap33.7 behind | Weight4% ref. weight | |
| Claw-Eval | Score74.5% | Versus best verified row Best verified: Ornith-1.5-397B · 81.4% | Gap6.9 behind | WeightDisplay only | Provider exact |
| BankerToolBench | Score76.1% | Versus best verified row Best verified: Hy4 preview · 78.6% | Gap2.5 behind | WeightDisplay only | Provider exact |
| ResearchClawBench | Score19.8% | Versus best verified row Best verified: Claude Opus 4.8 · 21.1% | Gap1.3 behind | WeightDisplay only | Benchmark exact |
| OSWorld 2.0 | Score4.6% | Versus best verified row Best verified: GPT-6 Astra · 72.6% | Gap68 behind | WeightDisplay only | Benchmark exact |
Multimodal5 rows
| Benchmark | Score | Versus best verified row | Gap | Weight | Evidence |
|---|---|---|---|---|---|
| MMMU-ProMassive Multi-discipline Multimodal Understanding Pro | Score78.1% | Versus best verified row Best verified: Gemini 3.5 Flash · 83.6% | Gap5.5 behind | WeightWeighted 40% | Provider exact |
| OfficeQA Pro | Score45.1% | Versus best verified row Best verified: Claude Opus 5.5 · 67.7% | Gap22.6 behind | WeightWeighted 25% | Provider exact |
| OmniDocBench 1.5 | Score91.6% | Versus best verified row Best verified: Qwen3.8 Max · 92.1% | Gap0.5 behind | WeightDisplay only | Provider exact |
| VideoMMMU | Score84.6% | Versus best verified row Best verified: Qwen3.8 Max · 88.7% | Gap4.1 behind | WeightDisplay only | Provider exact |
| Video-MME (with subtitle)Video-MME with subtitle | Score85.4% | Versus best verified row Best verified: Qwen3.8 Max · 90.4% | Gap5 behind | WeightDisplay only | Provider exact |
Knowledge2 rows
| Benchmark | Score | Versus best verified row | Gap | Weight | Evidence |
|---|---|---|---|---|---|
| MMLU-Pro (Vals)MMLU-Pro, Vals AI run | Score84.2% | Versus best verified row Best verified: Claude Fable 5.1 · 92.4% | Gap8.2 behind | Weight6% ref. weight | Verified |
| GPQA Diamond (Vals)GPQA Diamond, Vals AI run | Score92.7% | Versus best verified row Best verified: Gemini 3.1 Pro · 95.5% | Gap2.8 behind | Weight2% ref. weight |
Math1 row
| Benchmark | Score | Versus best verified row | Gap | Weight | Evidence |
|---|---|---|---|---|---|
| USAMO 2026United States of America Mathematical Olympiad 2026 | Score85.7% | Versus best verified row Best verified: Claude Mythos 5 · 97.6% | Gap11.9 behind | WeightWeighted 10% | Provider exact |
Bars run 0–100; the dark tick marks the best source-verified value
All 27 rowsWhat it costs to get this score
Published API price against the public score. The x-axis uses a log scale; the dashed path marks models that are not beaten by a cheaper, higher-scoring option. Price uses average of published input and output rates.
Current modelExplore all models
MiniMax M3 · 55.0 score · $0.75 blended per million tokens
The chart opens on the current model. Scroll horizontally to inspect the full price axis.
Horizontal: blended price per million tokens, log scale · Vertical: public score
Lineage
The sequence follows explicit supersedes links. Each score is estimated for that model; a relative can inform a sparse estimate but never sets a floor, so a newer release can score below an earlier one. Scores and prices remain blank when the corresponding public row or first-party rate is unavailable.
Base entry
Radar
Full release historyMiniMax M3 release history
Radar confirmed these at the source. Use MiniMax M3 in your work? Explore Radar to follow supported changes and choose your alerts.
Radar
Spec sheet
Each documented value carries its source. Missing fields stay visible as not sourced or not published, rather than disappearing from the page.
- API model ID
- Not published
- Context window
- 1M
- Maximum output
- Not sourced yet
- Knowledge cutoff
- Not sourced yet
- Input modalities
- Not sourced yet
- Output modalities
- Not sourced yet
- Parameters
- Not sourced yet
- Availability
- Not sourced yet
- Cloud regions
- Not tracked yet
- Lifecycle
- Current
- API capabilities
- Tool calling, structured outputs, and batch support are not tracked yet
- Prompt caching
- Published at $0.060 per million cached input tokens
- Self-host
- Open weights available; hardware estimate not sourced
- Rate limits
- Not tracked yet
How to read this profile
The visual layer above carries the decisions. These notes preserve the model, ranking, coverage, and family context behind the numbers.
MiniMax M3 ranks #72 of 221 on the public leaderboard with a score of 55/100. Its source-verified position is #44 of 89.
MiniMax M3 is a open weight model with a 1M context window. No explicit reasoning mode is documented in this profile.
Official exact-value snapshot from MiniMax's M3 model card, June 1 launch page, OpenRouter route page, and MSA technical report. BenchLM maps comparable rows directly and stores MiniMax's M3 comparison-chart rows such as VIBE V2, SVG-Bench, KernelBench Hard, GDPval rubrics, and BankerToolBench as display-only provider scores. The model card documents 1M context, 427B total parameters, about 23B activated parameters, and both thinking and non-thinking modes.
Its explicit predecessor is MiniMax M2.7. 27 of 671 tracked benchmark slots currently have displayable evidence. Missing categories stay blank.
Its strongest eligible category is Multimodal & Grounded at #43, while its lowest eligible position is Knowledge at #80. particularly strong for screenshots, documents, charts, and grounded multimodal workflows.
Last updated October 14, 2026. Runtime fields remain blank until a sourced snapshot exists.
Deployment options
Self-host and provider-specific paths stay separate from benchmark evidence so operating constraints are visible before a score becomes the whole decision.
Published weights are available, but BenchLM does not yet have a sourced parameter and VRAM profile for this exact model. Hardware cost estimates stay unavailable until that sizing record is complete.
Estimate VRAM from known parametersQuestions
How does MiniMax M3 perform overall in AI benchmarks?
MiniMax M3 ranks #72 out of 221 models on the public BenchAlign leaderboard, with a score of 55/100. Its evidence status is Supported, and this profile shows 27 source-displayable benchmark rows. The label describes evidence depth, not a provider quality claim; inspect category rows before choosing a workload.
Is MiniMax M3 good for knowledge and understanding?
MiniMax M3 ranks #80 out of 181 eligible models for knowledge and understanding, with a public category score of 49.6/100. Higher-ranked alternatives are available for workloads where this category decides the choice. Check the underlying rows before treating the aggregate as a workload guarantee.
Is MiniMax M3 good for coding and programming?
MiniMax M3 ranks #73 out of 149 eligible models for coding and programming, with a public category score of 36.5/100. Higher-ranked alternatives are available for workloads where this category decides the choice. Check the underlying rows before treating the aggregate as a workload guarantee.
Is MiniMax M3 good for mathematics?
MiniMax M3 has source-displayable benchmark coverage for mathematics, but the public category table does not assign it a rank there. The individual rows remain available for inspection. A missing category position means the evidence threshold was not met; it does not convert the model's unmeasured work into a zero.
Is MiniMax M3 good for agentic tool use and computer tasks?
MiniMax M3 ranks #66 out of 125 eligible models for agentic tool use and computer tasks, with a public category score of 39.6/100. Higher-ranked alternatives are available for workloads where this category decides the choice. Check the underlying rows before treating the aggregate as a workload guarantee.
Is MiniMax M3 good for multimodal and grounded tasks?
MiniMax M3 ranks #43 out of 54 eligible models for multimodal and grounded tasks, with a public category score of 61.5/100. Higher-ranked alternatives are available for workloads where this category decides the choice. Check the underlying rows before treating the aggregate as a workload guarantee.
Is MiniMax M3 open source?
MiniMax M3 is an open-weight model from MiniMax. Its weights can be downloaded for local or hosted deployment, subject to the published license. Open weight does not automatically mean open source: training data and training code may remain private, and commercial restrictions can still apply.
Does MiniMax M3 have full benchmark coverage on BenchLM?
No. MiniMax M3 currently has 56 source-displayable rows across 671 tracked benchmark slots. The profile exposes published, non-generated evidence and leaves missing categories blank until an exact evaluation is available. Coverage describes how much was measured; it is not a penalty added to an individual benchmark result.
What is the context window size of MiniMax M3?
MiniMax M3 has a reported context window of 1M in the exact-model catalog record. The value stays visible, but the profile marks its source link as unavailable instead of presenting it as directly documented. Maximum output length remains separate because providers often publish a different limit.