Data as of October 29, 2026 · How the score is built
Released Mar 18, 2026200K context
MiniMax M2.7
Decision readingMiniMax M2.7 scores 47.3 out of 100 and ranks #115 of 216. This profile shows 29 source-displayable benchmark rows; its strongest eligible category is Instruction Following at #14. API pricing is $0.3 input and $1.2 output per million tokens.
Released Mar 18, 2026 — see all recent releases
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
47.3/100
field median 50.5#115 of 216 ranked models
Public
#115of 216
Verified #58 of 77
Price
$0.30input / $1.20 output
input median $0.75blended $0.75
Speed
60tok/s
field median 92 tok/sFirst token 42.32 s
Context
200Ktokens
field median 256,000Reported for this model; direct source link not stored
Strongest published evidence
Instruction Following ranks #14. A well-rounded choice across a range of tasks.
Validate before choosing
29 published rows leave some tracked benchmark slots empty. Agentic is its lowest eligible category at #99.
Source-linked · 29 displayable benchmark rows
Category 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 #99 of 135Percentile 27thWeight 22%8 benchmarksMixed sources | 33.2 | 8 benchmarks | Mixed sources | |||
| CodingRank #85 of 156Percentile 46thWeight 20%12 benchmarksMixed sources | 34.7 | 12 benchmarks | Mixed sources | |||
| ReasoningWeight 17%0 benchmarksNot measured | Not measured | 0 benchmarks | Not measured | |||
| MultimodalWeight 12%0 benchmarksNot measured | Not measured | 0 benchmarks | Not measured | |||
| KnowledgeRank #94 of 179Percentile 48thWeight 12%5 benchmarksMixed sources | 45.2 | 5 benchmarks | Mixed sources | |||
| MultilingualWeight 7%0 benchmarksNot measured | Not measured | 0 benchmarks | Not measured | |||
| Inst. FollowingRank #14 of 157Percentile 92ndWeight 5%Not listedUnlisted sources | 93.0 | Not listed | Unlisted sources | |||
| MathRank Not rankedWeight 5%4 benchmarksReported | 70.5 | 4 benchmarks | Reported |
29 of 782 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.
- Agentic7/8 verified
- Coding11/12 verified
- ReasoningNot measured
- MultimodalNot measured
- Knowledge2/5 verified
- MultilingualNot measured
- Inst. FollowingNot measured
- Math0/4 verified
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 M2.7 category percentile values
- Agentic27th percentile
- Coding46th percentile
- ReasoningNot eligible
- MultimodalNot eligible
- Knowledge48th percentile
- MultilingualNot eligible
- Instruction following92nd percentile
- MathNot eligible
The dashed outline is median of 6 nearest peers.
Eligible category ranks
- Agentic#99/135
- Coding#85/156
- ReasoningNot ranked
- MultimodalNot ranked
- Knowledge#94/179
- MultilingualNot ranked
- Inst. Following#14/157
- MathNot ranked
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.
Coding12 rows
| Benchmark | Score | Versus best verified row | Gap | Weight | Evidence |
|---|---|---|---|---|---|
| SWE-bench Pro | Score56.2% | Versus best verified row Best verified: Claude Opus 5.5 · 89.9% | Gap33.7 behind | Weight26% ref. weight | Provider exact |
| LiveCodeBench (Vals)LiveCodeBench, Vals AI run | Score79.9% | Versus best verified row Best verified: Claude Fable 5.1 · 90.5% | Gap10.6 behind | Weight8% ref. weight | |
| SWE-Rebench | Score51.9% | Versus best verified row Best verified: Claude Opus 4.6 · 65.3% | Gap13.4 behind | Weight5% ref. weight | Benchmark exact |
| SWE Multilingual | Score76.5% | Versus best verified row Best verified: Claude Opus 5.5 · 93.9% | Gap17.4 behind | Weight5% ref. weight | Provider exact |
| Terminal-Bench 2.0 | Score57.0% | Versus best verified row Best verified: GPT-5.5 · 82.0% | Gap25 behind | WeightScored in Agentic | Provider exact |
| SWE-bench Verified*SWE-bench Verified (mini-swe-agent-v2) | Score75.4% | Versus best verified row | GapNo verified comparator | WeightDisplay only | Secondary exact |
| Multi-SWE Bench | Score52.7% | Versus best verified row Best verified: MiniMax M2.7 · 52.7% | GapBest verified | WeightDisplay only | Provider exact |
| VIBE-Pro | Score55.6% | Versus best verified row Best verified: MiniMax M2.7 · 55.6% | GapBest verified | WeightDisplay only | Provider exact |
| NL2Repo | Score39.8% | Versus best verified row Best verified: DeepSeek V4.1 Flash 1017 · 68.2% | Gap28.4 behind | WeightDisplay only | Provider exact |
| Vibe Code BenchVibe Code Bench v1.1 | Score27.04% | Versus best verified row Best verified: Gemini 4 Argon · 91.90% | Gap64.9 behind | WeightDisplay only | Benchmark exact |
| React Native Evals | Score71.4% | Versus best verified row Best verified: Composer 2 · 96.1% | Gap24.7 behind | WeightDisplay only | Benchmark exact |
| SWE-bench (Vals)SWE-bench, Vals AI run | Score73.8% | Versus best verified row Best verified: Claude Opus 5 · 97.0% | Gap23.2 behind | WeightDisplay only | Verified |
Agentic8 rows
| Benchmark | Score | Versus best verified row | Gap | Weight | Evidence |
|---|---|---|---|---|---|
| Terminal-Bench 2.0 | Score57% | Versus best verified row Best verified: GPT-5.5 · 82% | Gap25 behind | Weight7% ref. weight | Provider exact |
| Terminal-Bench 2.1 (Vals)Terminal-Bench 2.1, Vals AI run | Score48.7% | Versus best verified row Best verified: GPT-6 Astra · 87.3% | Gap38.6 behind | Weight4% ref. weight | |
| Toolathlon | Score46.3% | Versus best verified row Best verified: Muse Spark 1.1 · 75.6% | Gap29.3 behind | Weight3% ref. weight | Provider exact |
| τ³-bench resultsτ³-Bench Tool-Agent-User Evaluation | Score67.6% | Versus best verified row Best verified: Mercury 2.5 · 96.0% | Gap28.4 behind | Weight2% ref. weight | Secondary exact |
| MLE-Bench Lite | Score66.6% | Versus best verified row Best verified: Atria Dawn Preview · 86.2% | Gap19.6 behind | WeightDisplay only | Provider exact |
| MM-ClawBench | Score62.7% | Versus best verified row Best verified: MiniMax M2.7 · 62.7% | GapBest verified | WeightDisplay only | Provider exact |
| Claw-Eval | Score48.7% | Versus best verified row Best verified: Ornith-1.5-397B · 81.4% | Gap32.7 behind | WeightDisplay only | Benchmark exact |
| Gert LabsGert Labs Composite Game Benchmark | Score40.40% | Versus best verified row Best verified: Claude Opus 4.8 · 72.97% | Gap32.6 behind | WeightDisplay only | Benchmark exact |
Knowledge5 rows
| Benchmark | Score | Versus best verified row | Gap | Weight | Evidence |
|---|---|---|---|---|---|
| HLEHumanity's Last Exam | Score28% | Versus best verified row Best verified: Kimi K3.1 Pro · 70% | Gap42 behind | Weight44% ref. weight | Secondary exact |
| MMLU-Pro (Vals)MMLU-Pro, Vals AI run | Score80.4% | Versus best verified row Best verified: Claude Fable 5.1 · 92.4% | Gap12 behind | Weight6% ref. weight | Verified |
| GPQA Diamond (Vals)GPQA Diamond, Vals AI run | Score86.6% | Versus best verified row Best verified: Gemini 3.1 Pro · 95.5% | Gap8.9 behind | Weight2% ref. weight | |
| GPQA-DGPQA Diamond | Score87.0% | Versus best verified row Best verified: GPT-6 Astra · 96.0% | Gap9 behind | WeightDisplay only | Secondary exact |
| MMLU-Pro (Arcee)MMLU-Pro first-party comparison snapshot | Score80.8% | Versus best verified row Best verified: Trinity-Large-Preview · 75.2% | Gap5.6 behind | WeightDisplay only | Secondary exact |
Math4 rows
| Benchmark | Score | Versus best verified row | Gap | Weight | Evidence |
|---|---|---|---|---|---|
| AIME26AIME 2026 | Score89.8% | Versus best verified row Best verified: GLM-5.2 · 99.2% | Gap9.4 behind | WeightWeighted 25% | Secondary exact |
| HMMT Feb 2026Harvard-MIT Mathematics Tournament February 2026 | Score72.7% | Versus best verified row Best verified: Qwen3.7 Max · 97.1% | Gap24.4 behind | WeightWeighted 25% | Secondary exact |
| AIME25 (Arcee)AIME25 first-party comparison snapshot | Score80.0% | Versus best verified row Best verified: Trinity-Large-Preview · 24.0% | Gap56 behind | WeightDisplay only | Secondary exact |
| HMMT Nov 2025Harvard-MIT Mathematics Tournament November 2025 | Score81.0% | Versus best verified row Best verified: Qwen3.6 Plus · 94.6% | Gap13.6 behind | WeightDisplay only | Secondary exact |
Bars run 0–100; the dark tick marks the best source-verified value
All 29 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.
MiniMax M2.7 · 47.3 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
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
- 200K
- 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
- Superseded
- API capabilities
- Tool calling, structured outputs, and batch support are not tracked yet
- Prompt caching
- Not documented in the pricing record
- Self-host
- MiniMax M2.7 needs ~160GB VRAM (2× NVIDIA H100 (80GB)).
- 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 M2.7 ranks #115 of 216 on the public leaderboard with a score of 47.29/100. Its source-verified position is #58 of 77.
MiniMax M2.7 is a open weight model with a 200K context window. No explicit reasoning mode is documented in this profile.
Official exact-value snapshot based on MiniMax's live Hugging Face model card and March 18, 2026 launch report. BenchLM only carries the exact benchmarks MiniMax publishes there directly.
Its explicit predecessor is MiniMax M2.5. 29 of 782 tracked benchmark slots currently have displayable evidence. Missing categories stay blank.
Its strongest eligible category is Instruction Following at #14, while its lowest eligible position is Agentic at #99. a well-rounded choice across a range of tasks.
Last updated October 29, 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.
Where to run MiniMax M2.7 by API
3 hosts publish a per-token price for it. Novita AI has the lowest output rate, $1.20 per million tokens. Prices read from each host’s own pages on October 20, 2026.
| Host | Input /1M | Output /1M | Cached /1M | Context |
|---|---|---|---|---|
| Novita AI | $0.30 | $1.20 | $0.06 | 205K |
| OpenRouter | $0.30 | $1.20 | $0.06 | 205K |
| Together AI | $0.30 | $1.20 | $0.06 | — |
Self-host vs API cost
Estimates at 50,000 req/day · 1000 tokens/req average.
Questions
How does MiniMax M2.7 perform overall in AI benchmarks?
MiniMax M2.7 ranks #115 out of 216 models on the public BenchAlign leaderboard, with a score of 47.29/100. Its evidence status is Supported, and this profile shows 29 source-displayable benchmark rows. The label describes evidence depth, not a provider quality claim; inspect category rows before choosing a workload.
Is MiniMax M2.7 good for knowledge and understanding?
MiniMax M2.7 ranks #94 out of 179 eligible models for knowledge and understanding, with a public category score of 45.2/100. 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 M2.7 good for coding and programming?
MiniMax M2.7 ranks #85 out of 156 eligible models for coding and programming, with a public category score of 34.7/100. 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 M2.7 good for mathematics?
MiniMax M2.7 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 M2.7 good for agentic tool use and computer tasks?
MiniMax M2.7 ranks #99 out of 135 eligible models for agentic tool use and computer tasks, with a public category score of 33.2/100. 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 M2.7 open source?
MiniMax M2.7 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 M2.7 have full benchmark coverage on BenchLM?
No. MiniMax M2.7 currently has 46 source-displayable rows. 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 M2.7?
MiniMax M2.7 has a reported context window of 200K 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.