Skip to content

MiniMax · Chat

MiniMax M2.5

MiniMax 🇨🇳ChatOpen weights. One of its sources says the weights are publicly available.The maker currently lists 1 Jul 2027 as the deprecation date. Plans can change.
From Venice's listing:
MiniMax-M2.5 is a large language model optimized for coding, agentic workflows, and modern application development with enhanced reasoning capabilities.
llminfo IQ
122.8
Maker's price
$0.525 blended / 1M
Max context
1M
Listings
105

Evidence and sources

International API storefront (minimax.io) is Singapore-based; principal operating entity 上海稀宇科技有限公司 (Shanghai).

🤗 MiniMaxAI/MiniMax-M2.5 Model architecture, as its sources state it llminfo IQ 122.8, ranked among chat models. From 192 benchmarks. Plenty of data behind it.How llminfo IQ is computedCoding IQ 100.4, 95% interval 94–106. From 16 benchmark families across 9 sources. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.How Coding IQ is computedFair value (48.7). Costs per task about what its benchmark results predict. From 29 benchmarks with measured task costs. Plenty of data behind it.How Value for money worksYap score: tokens used per coding task vs. models with the same resultsChatty (64 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate. By effort: default 64 · none 51.Median 43,543 output tokens per task at default effort.How the Yap score worksReleased 12 Feb 2026Maker HQ: CN (source)
All rankings: #175 on OpenRouter today #221 on OpenRouter this week #240 on OpenRouter this month #361 on NanoGPT recentlySources: Writer Pricing, LiteLLM data, Models.dev data, AI/ML API data, Morpheus Inference Marketplace data, Venice.ai data, Hugging FacePrimary data from Writer Pricing from LiteLLM: deprecation date from Models.dev: weights availability from AI/ML API: primary mode from Morpheus Inference Marketplace: architecture from Venice.ai: description from Hugging Face: release date
Tags from its sources:instructtoolsreasoningopen-sourcemoecodingfunction-calling

Advantages and trade-offs

Compared with 10 chat models, picked mainly by popularity and IQ

Advantages

Open weights available

The weights are public. 1 of the 9 peers that say either way has open weights.

We only make a claim when there's enough data on models of the same type. Updated 22 Nov, 04:34 UTC.Prices come from the model's maker where they sell it, otherwise their official partner, otherwise the market median. This comparison uses 10 maker's prices.

Input & output

as stated by the model's providers
Text
Images
Text

Capabilities

Supported operations and provider-stated features

VisionReasoningFunction callingStructured outputPrompt cachingWeb searchBatch API

Listings

 MiniMaxminimax/MiniMax-M2.5Sources: LiteLLM data, Models.dev data1M$0.30cache read $0.03$1.20cache write $0.375
 Aihubmixcoding-minimax-m2.5-free · coding, freeSource: AIHubMix data docs205K$0.00$0.00
 Flowdotminimax/minimax-m2.5:free · freeSource: FlowDot data197K$0.00$0.00
 DeepInfraMiniMaxAI/MiniMax-M2.5Source: Models.dev data197K$0.15cache read $0.03$1.15
 Writerminimax/minimax-m2.5Source: Writer PricingNot published by the provider$0.15$0.90
 Dinferenceminimax-m2.5Source: Models.dev data200K$0.22$0.88
 TokensmindMiniMax-M2.5Source: TokensMindNot published by the provider$0.261$1.035
 Hugging FaceInference by VeniceMiniMaxAI/MiniMax-M2.5Source: Hugging Face Inference Providers data198K$0.27$0.95
 Kilominimax/minimax-m2.5Sources: Kilo AI Gateway data, Models.dev data198K$0.27cache read $0.03$0.95
 Nous Portalminimax/minimax-m2.5Source: Nous Portal Inference Gateway data198K$0.27cache read $0.03$0.95

14 listings without a price are hidden.

Included in plans

MiniMax M Plan GoMiniMaxSubscription$22/moPlan page
MiniMax M Plan ExploreMiniMaxSubscription$55/moPlan page
MiniMax M Plan BuildMiniMaxSubscription$132/moPlan page
Coding Plan ProAlibaba CloudSubscription$50/moPlan page
Coding Plan LiteTencentSubscriptionCN¥40/moPlan page
Coding Plan ProTencentSubscriptionCN¥200/moPlan page

Benchmarks

Named effort levels run from lowest to highest. Select a result to see its source details.

Human preference

Arena ELO50 related benchmarks · 50 results
Source: LMArena data · Nov 2026
BenchmarkScore
Arena ELO
Chinese
Creative Writing
English
Show 46 more benchmarksHide 46 benchmarks
BenchmarkScore
Exclude Ties
Expert
French
German
Hard Prompts
Hard Prompts English
Industry Business and Management and Financial Operations
Industry Entertainment and Sports and Media
Industry Life and Physical and Social Science
Industry Mathematical
Industry Medicine and Healthcare
Industry Software and It Services
Industry Writing and Literature and Language
Japanese
Korean
Longer Query
Multi Turn
Non English
Polish
Russian
Spanish
Style-Controlled
Text Factuality
Text Factuality - Chinese
Text Factuality - Creative Writing
Text Factuality - English
Text Factuality - Exclude Ties
Text Factuality - Expert
Text Factuality - French
Text Factuality - Hard Prompts
Text Factuality - Hard Prompts English
Text Factuality - Industry Business and Management and Financial Operations
Text Factuality - Industry Entertainment and Sports and Media
Text Factuality - Industry Life and Physical and Social Science
Text Factuality - Industry Mathematical
Text Factuality - Industry Medicine and Healthcare
Text Factuality - Industry Software and It Services
Text Factuality - Industry Writing and Literature and Language
Text Factuality - Longer Query
Text Factuality - Multi Turn
Text Factuality - Non English
Text Factuality - Polish
Text Factuality - Russian
Text Factuality - Spanish

Agentic

Reported by MiniMax · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
main capability figure
BrowseComp3 results
Reported by MiniMax · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
with context
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
GDPval-AA · Published ScoreMiniMax M2.7 official benchmark overview · exact harness/version not stated · Reported by MiniMax · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
GDPval-MM4 results
Reported by MiniMax · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
main capability figure
official prose · average win rate
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
MLE-Bench LiteMiniMax M2.7 official benchmark overview · exact harness/version not stated · Reported by MiniMax · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
OpenHandsSource: OpenHands Index data
SWE-Bench Pro3 results
Reported by MiniMax · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
main capability figure
MiniMax M2.7 official benchmark overview · exact harness/version not stated
Source: Hugging Face Official Benchmarks data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
Source: OpenRouter benchmark exports · Aug 2026
11 Feb 2026 version · 90-day results
Across 8 providers lowest 58% median 65.3% highest 68%OpenRouter's automatic routing 64.7%
Show results from 8 providers and OpenRouter's automatic routingHide results from 8 providers and OpenRouter's automatic routing
  • SiliconFlow
  • Minimax
  • DigitalOcean
  • Friendli
  • OpenRouter's automatic routing
  • Venice
  • Novita
  • AtlasCloud
  • StreamLake

Coding

Arena ELOSource: LMArena data · Nov 2026
Reported by MiniMax · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
main capability figure
MiniMax M2.7 official benchmark overview · exact harness/version not stated
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
SciCodeReported by MiniMax · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
SWE-bench Verified4 related benchmarks · 7 results
SWE-bench Verified
Reported by MiniMax · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
official prose · default prompt
main capability figure
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
SWE-bench Multilingual
Reported by MiniMax · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
main capability figure
ValsSource: Vals AI
VIBE-Pro3 results
Reported by MiniMax · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
main capability figure · average
MiniMax M2.7 official benchmark overview · exact harness/version not stated
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated

Reasoning

CL-bench9 related benchmarks · 18 results
Overall
Reported by Tencent Hunyuan / Fudan NLP Group · via BenchBoard docs
benchmark-owner leaderboard run · site model-mode label: MiniMax M2.5 · 1899 tasks · GPT-5.1 judge · low reasoning effort · single run; no standard deviation · candidate API snapshot and exact candidate harness not specified on public leaderboard
Source: Epoch AI Benchmarking Hub data
Setting not stated
Domain Knowledge Reasoning
Reported by Tencent Hunyuan / Fudan NLP Group · via BenchBoard docs
benchmark-owner leaderboard run · site model-mode label: MiniMax M2.5 · 1899 tasks · GPT-5.1 judge · low reasoning effort · single run; no standard deviation · candidate API snapshot and exact candidate harness not specified on public leaderboard
Source: Epoch AI Benchmarking Hub data
Setting not stated
Empirical Discovery & Simulation
Reported by Tencent Hunyuan / Fudan NLP Group · via BenchBoard docs
benchmark-owner leaderboard run · site model-mode label: MiniMax M2.5 · 1899 tasks · GPT-5.1 judge · low reasoning effort · single run; no standard deviation · candidate API snapshot and exact candidate harness not specified on public leaderboard
Source: Epoch AI Benchmarking Hub data
Setting not stated
Life · Behavioral Records & Activity Trails
Reported by Tencent Hunyuan / Fudan NLP Group · via BenchBoard docs
benchmark-owner leaderboard run · site model-mode label: MiniMax M2.5 (Thinking) · 405 tasks · GPT-5.1 judge · high reasoning effort · candidate API snapshot and exact candidate harness not specified on public leaderboard
Source: Epoch AI Benchmarking Hub data
Setting not stated
Life · Communication & Social Interactions
Reported by Tencent Hunyuan / Fudan NLP Group · via BenchBoard docs
benchmark-owner leaderboard run · site model-mode label: MiniMax M2.5 (Thinking) · 405 tasks · GPT-5.1 judge · high reasoning effort · candidate API snapshot and exact candidate harness not specified on public leaderboard
Source: Epoch AI Benchmarking Hub data
Setting not stated
Life · Fragmented Information & Revisions
Reported by Tencent Hunyuan / Fudan NLP Group · via BenchBoard docs
benchmark-owner leaderboard run · site model-mode label: MiniMax M2.5 (Thinking) · 405 tasks · GPT-5.1 judge · high reasoning effort · candidate API snapshot and exact candidate harness not specified on public leaderboard
Source: Epoch AI Benchmarking Hub data
Setting not stated
Show 6 more resultsHide 6 results
Life · Overall
Reported by Tencent Hunyuan / Fudan NLP Group · via BenchBoard docs
benchmark-owner leaderboard run · site model-mode label: MiniMax M2.5 (Thinking) · 405 tasks · GPT-5.1 judge · high reasoning effort · candidate API snapshot and exact candidate harness not specified on public leaderboard
Source: Epoch AI Benchmarking Hub data
Setting not stated
Procedural Task Execution
Reported by Tencent Hunyuan / Fudan NLP Group · via BenchBoard docs
benchmark-owner leaderboard run · site model-mode label: MiniMax M2.5 · 1899 tasks · GPT-5.1 judge · low reasoning effort · single run; no standard deviation · candidate API snapshot and exact candidate harness not specified on public leaderboard
Source: Epoch AI Benchmarking Hub data
Setting not stated
Rule System Application
Reported by Tencent Hunyuan / Fudan NLP Group · via BenchBoard docs
benchmark-owner leaderboard run · site model-mode label: MiniMax M2.5 · 1899 tasks · GPT-5.1 judge · low reasoning effort · single run; no standard deviation · candidate API snapshot and exact candidate harness not specified on public leaderboard
Source: Epoch AI Benchmarking Hub data
Setting not stated
Reported by MiniMax · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
official appendix · no tools
Source: OpenEvals Leaderboard data · Reported by the model's developer. The source relaying it did not verify it. · Mar 2026
Setting not stated
MEWC3 results
Reported by MiniMax · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
main capability figure
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
Toolathlon VerifiedMiniMax M2.7 official benchmark overview · exact harness/version not stated · Reported by MiniMax · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
WideSearchReported by MiniMax · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.

Science

GPQA Diamond11 results
Reported by MiniMax · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Source: OpenEvals Leaderboard data · Reported by the model's developer. The source relaying it did not verify it. · Mar 2026
Setting not stated
Source: OpenRouter benchmark exports · Aug 2026
11 Feb 2026 version · 90-day results
Across 8 providers lowest 68.1% median 84.2% highest 84.6%OpenRouter's automatic routing 81.1%
Show results from 8 providers and OpenRouter's automatic routingHide results from 8 providers and OpenRouter's automatic routing
  • Minimax
  • SiliconFlow
  • DigitalOcean
  • Novita
  • StreamLake
  • Friendli
  • OpenRouter's automatic routing
  • AtlasCloud
  • Venice

Instruction following

IFBenchReported by MiniMax · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.

Math

AIME 2025Reported by MiniMax · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.

Factuality

Knowledge

General

Long context

Multilingual

Other

Horangi36 related benchmarks · 36 results
MM-ClawBenchMiniMax M2.7 official benchmark overview · exact harness/version not stated · Reported by MiniMax · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
RISEReported by MiniMax · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
ValsSource: Vals AI
VIBE-ProReported by MiniMax · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
WildClawBench6 related benchmarks · 6 results
WildClawBench
Source: Hugging Face Official Benchmarks data · Reported by the model's developer. The source relaying it did not verify it. claim
Multimodal (26 tasks)
Reported by WildClawBench / InternLM · via BenchBoard docs
OpenClaw · multimodal-input split · 26 tasks
Overall
Reported by WildClawBench / InternLM · via BenchBoard docs
OpenClaw · full 60-task suite
Pure-text (34 tasks)
Reported by WildClawBench / InternLM · via BenchBoard docs
OpenClaw · pure-text split · 34 tasks
Total Cost
Reported by WildClawBench / InternLM · via BenchBoard docs
OpenClaw · full 60-task suite
Total Elapsed
Reported by WildClawBench / InternLM · via BenchBoard docs
OpenClaw · full 60-task suite

Data credits: BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links. · Epoch AI Benchmarking Hub, CC BY 4.0 (licence) · LMArena, CC BY 4.0 (licence) · Data by LLM Stats · BullshitBench by Peter Gostev, MIT (licence) · ChemBench leaderboard, Jablonka group / lamalab-org. Public verified numeric result summaries; results licence not separately stated. · EQ-Bench (Samuel J. Paech) — cite paech2023eqbench · OpenEvals/leaderboard-data, MIT (licence) · OpenRouter public benchmark leaderboard and search-lane exports, CC BY 4.0. Fractional source scores converted to percentages; immutable dated snapshots. (licence) · Horangi 4, Weights & Biases Korea; public publisher-selected numeric evaluation results · Hugging Face official benchmark leaderboards; licence varies by benchmark · LLM Chess by Maxim Saplin, Apache-2.0 (Copyright [2024] [Maxim Saplin]) (licence) · MineBench, MIT (licence) · OpenHands Index (All Hands AI), Apache-2.0 (licence) · SnakeBench (Greg Kamradt / ARC Prize Foundation), MIT (licence) · SpreadsheetBench (Renmin University of China and contributors), spreadsheetbench.github.io. · Vectara Hallucination Leaderboard, Apache-2.0 (licence)

Also known as: alicloud-minimax-m2.5, azure_ai/fw-minimax-m2.5, baseten/minimaxai/minimax-m2.5, bedrock/ap-northeast-1/minimax.minimax-m2.5, bedrock/ap-south-1/minimax.minimax-m2.5, bedrock/ap-southeast-2/minimax.minimax-m2.5, bedrock/ap-southeast-3/minimax.minimax-m2.5, bedrock/eu-central-1/minimax.minimax-m2.5, bedrock/eu-north-1/minimax.minimax-m2.5, bedrock/eu-south-1/minimax.minimax-m2.5, bedrock/eu-west-1/minimax.minimax-m2.5, bedrock/eu-west-2/minimax.minimax-m2.5, bedrock/minimax-m2.5@eu-central-1, bedrock/minimax-m2.5@eu-north-1, bedrock/minimax-m2.5@eu-south-1, bedrock/minimax-m2.5@eu-west-1, bedrock/minimax-m2.5@us-east-1, bedrock/minimax-m2.5@us-east-2, bedrock/minimax-m2.5@us-west-2, bedrock/sa-east-1/minimax.minimax-m2.5, bedrock/us-east-1/minimax.minimax-m2.5, bedrock/us-east-2/minimax.minimax-m2.5, bedrock/us-west-2/minimax.minimax-m2.5, bedrock_mantle/ap-northeast-1/minimax.minimax-m2.5, bedrock_mantle/ap-south-1/minimax.minimax-m2.5, bedrock_mantle/ap-southeast-2/minimax.minimax-m2.5, bedrock_mantle/ap-southeast-3/minimax.minimax-m2.5, bedrock_mantle/eu-central-1/minimax.minimax-m2.5, bedrock_mantle/eu-north-1/minimax.minimax-m2.5, bedrock_mantle/eu-south-1/minimax.minimax-m2.5, bedrock_mantle/eu-west-1/minimax.minimax-m2.5, bedrock_mantle/eu-west-2/minimax.minimax-m2.5, bedrock_mantle/minimax.minimax-m2.5, bedrock_mantle/sa-east-1/minimax.minimax-m2.5, cc-minimax-m2.5, cc-minimax-m2.5-highspeed, coding-minimax-m2.5, coding-minimax-m2.5-free, coding-minimax-m2.5-highspeed, gmi/minimaxai/minimax-m2.5, minimax-m2-5, minimax-m2.5, minimax-m2.5-free, minimax-m2.5-highspeed, minimax-m2.5:web, minimax.minimax-m2.5, minimax/m2-5-20260218, minimax/m2-5-highspeed-20260218, minimax/minimax-m2.5, minimax/minimax-m2.5-highspeed, minimax/minimax-m2.5:free, minimaxai/minimax-m2.5, minimaxai/minimax-m2.5-fp4, minimaxi/minimax-m2.5, minimaxi/minimax-m2.5-highspeed, nebius/minimaxai/minimax-m2.5, novita/minimax-m2.5, novita/minimax/minimax-m2.5, novita/minimax/minimax-m2.5-highspeed, streamlake/minimax-m2.5, tensormesh/minimaxai/minimax-m2.5

Like this one. Tap or hover any dotted value, label or icon to see where it comes from and how it is measured. open details when tapped or hovered. How this data works