DataLearner logo
MI

MiniMax M2.5

Reasoning modelMiniMax MMiniMax M2.5

MiniMax M2.5

Release date: 2026-02-12Updated: 2026-07-17 21:57:36.3886,829
Live demoGitHubHugging FaceCompare
Parameters
229B
Context length
128K
Chinese support
Supported
Reasoning ability

MiniMax M2.5 is a reasoning model from MiniMaxAI, released on 2026-02-12. It accepts text input and returns text output. The cataloged parameter count is 229B, with 10B active parameters per inference. The recorded context window is 128K. Cataloged capabilities include Reasoning model and Multilingual. The checkpoint is listed under the Minimax Model License, with the license link included in the model references. Use the linked references to confirm current access, licensing, and provider-specific limits.

Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology

MiniMax M2.5

Model basics

Reasoning traces
Supported
Thinking modes
Thinking Mode (Default)Standard Mode
Context length
128K tokens
Max output length
No data
Model type
Reasoning model
Modality (in / out)
Text → Text
Release date
2026-02-12
Model file size
230GB
MoE architecture
Yes
Total params / Active params
229B / 10B
Knowledge cutoff
No data
MiniMax M2.5

Open source & experience

MiniMax M2.5

Official resources

Paper
Paper
DataLearnerAI blog
No blog post yet
MiniMax M2.5

API details

API speed
3/5
💡Default unit: $/1M tokens. If vendors use other units, follow their published pricing.
Learn about pricing modes
Standard
TypeConditionInputOutput
Text-$0.300/ 1M$2.40/ 1M
Turbo
TypeConditionInputOutput
Text-$0.150/ 1M$1.20/ 1M
MiniMax M2.5

Benchmark Results

MiniMax M2.5 currently shows benchmark results led by SWE-bench Verified (14 / 114, score 80.20), Claw Bench (4 / 29, score 92.10), Pinch Bench (7 / 38, score 87.80). This page also consolidates core specs, context limits, and API pricing so you can evaluate the model from benchmark results and deployment constraints together.

Thinking
Tool usage

General Knowledge

4 evaluations
Benchmark / mode
Score
Rank/total
ARC-AGI
Thinking Mode
63.70
35 / 68
LiveBench
Deep
60.14
68 / 115
HLE
Thinking Mode
19.40
130 / 181
ARC-AGI-2
Thinking Mode
4.90
47 / 62

General Evaluation

1 evaluations
Benchmark / mode
Score
Rank/total
GPQA Diamond
Thinking Mode
85.20
84 / 224

Coding and Software Engineer

2 evaluations
Benchmark / mode
Score
Rank/total
SWE-bench Verified
Thinking ModeTools
80.20
14 / 114
SWE-Bench Pro - Public
Thinking ModeTools
55.40
28 / 57

Math and Reasoning

1 evaluations
Benchmark / mode
Score
Rank/total
AIME2025
Thinking Mode
86.30
48 / 106

Writing and Creative Capabilities

1 evaluations
Benchmark / mode
Score
Rank/total
Creative Writing
Standard Mode
1358.20
63 / 99

Agent Level Benchmark

1 evaluations
Benchmark / mode
Score
Rank/total
τ²-Bench - Telecom
Thinking ModeTools
97.80
10 / 35

Instruction Following

1 evaluations
Benchmark / mode
Score
Rank/total
IF Bench
Thinking ModeTools
70
15 / 33

AI Agent - Information Search

1 evaluations
Benchmark / mode
Score
Rank/total
BrowseComp
Thinking ModeTools
76.30
23 / 54

AI Agent - Tool Usage

1 evaluations
Benchmark / mode
Score
Rank/total
Terminal Bench 2.0
Thinking ModeTools
51.70
32 / 48

Productivity Knowledge

1 evaluations
Benchmark / mode
Score
Rank/total
GDPval-AA
Thinking Mode
36
17 / 21

Long Context

1 evaluations
Benchmark / mode
Score
Rank/total
AA-LCR
Thinking Mode
69.50
6 / 18

Claw-style Agent Evaluation

2 evaluations
Benchmark / mode
Score
Rank/total
Claw Bench
Thinking ModeTools
92.10
4 / 29
Pinch Bench
Thinking ModeTools
87.80
7 / 38
View benchmark analysisCompare with other models

Compare with other models

MiniMax M2.5

Publisher

MiniMax M2.5

Model Overview

MiniMax M2.5 is a reasoning model from MiniMaxAI, released on 2026-02-12.

It accepts text input and produces text output. Its cataloged capabilities include Reasoning model and Multilingual. The cataloged parameter count is 229B, with 10B active parameters per inference. The recorded context window is 128K.

The checkpoint is listed under the Minimax Model License, with the license link included in the model references. The page records 4 API pricing rules from MiniMaxAI; current provider pricing and conditions should be checked before deployment. The evaluation section contains 16 cataloged benchmark results with their recorded modes and scores. The page links 4 release, model-card, repository, or provider references for checking the underlying claims. Specifications, availability, and prices can change; undisclosed values are intentionally left unstated.

MiniMax M2.5

FAQ

What is MiniMax M2.5?

MiniMax M2.5 is a reasoning model from MiniMaxAI, released on 2026-02-12. It accepts text input and returns text output. The cataloged parameter count is 229B, with 10B active parameters per inference. The recorded context window is 128K. Cataloged capabilities include Reasoning model and Multilingual. The checkpoint is listed under the Minimax Model License, with the license link included in the model references. Use the linked references to confirm current access, licensing, and provider-specific limits.

What input and output modalities does MiniMax M2.5 support?

The current model record lists text as input and text as output.

What are the main recorded specifications for MiniMax M2.5?

The cataloged parameter count is 229B, with 10B active parameters per inference. The recorded context window is 128K. Fields without a source-backed value remain undisclosed.

Does MiniMax M2.5 have API pricing?

The page records 4 API pricing rules from MiniMaxAI; current provider pricing and conditions should be checked before deployment.

Are benchmark results available for MiniMax M2.5?

The evaluation section contains 16 cataloged benchmark results with their recorded modes and scores. Compare only results that use the same benchmark version and evaluation mode.

Is MiniMax M2.5 open source?

The checkpoint is listed under the Minimax Model License, with the license link included in the model references. Review the linked license text before commercial or derivative use.

DataLearner on WeChat

Follow DataLearner on WeChat for AI model updates and research notes.

DataLearner WeChat QR code