MiniMax · MiniMax M3 · Reviewed 2026-10-02

MiniMax M3

MiniMax's open-weight M3 model combining sparse attention, visual input, and long-context agentic work.

All models

Overall · 0–10038.7

Status: active
Released: 2026-06-01
Context: 1,000,000 tokens
Parameters: Undisclosed
Inputs: Text · Image · Video
Type: multimodal
Max output: — tokens
API model ID: —
Reasoning: —
Knowledge cutoff: —
Training cutoff: —
Weights / license: Available · License undisclosed
Price / 1M tokens: —
Cached input: —

Independent benchmark scores

Independently evaluated

Coding

Terminal-Bench 4.0 in the Artificial Analysis evaluation harness.

Artificial Analysis · AA evaluated M3 configuration

Independently evaluated

Reasoning

39%

Humanity's Last Exam as evaluated by Artificial Analysis; the source configuration is preserved.

Artificial Analysis · AA evaluated M3 configuration

Independently evaluated

Research

1246 Elo

GDPval-AA v2.1 professional knowledge-work Elo. This is a task lens, not an assessment of every research skill.

Artificial Analysis · AA evaluated M3 configuration

Independently evaluated

Multimodal

10%

GDP.pdf professional document understanding, including visual tables, charts and diagrams; not an audio/video score.

Artificial Analysis · AA evaluated M3 configuration

Independently evaluated

Agentic

21%

AutomationBench-AA multi-step business workflows in the Artificial Analysis harness.

Artificial Analysis · AA evaluated M3 configuration

Raw benchmark results

Source-backed rows with the original evaluation context preserved.

6 results

Benchmark	Score	Source	Evaluation notes
AutomationBench-AAChecked 2026-10-02	21%	Artificial Analysis	Artificial Analysis Intelligence Index v4.3.2; configuration=AA evaluated M3 configuration. Rounded public score; checked date is not the evaluation execution date.
AA GDP.pdfChecked 2026-10-02	10%	Artificial Analysis	Artificial Analysis Intelligence Index v4.3.2; configuration=AA evaluated M3 configuration. Rounded public score; checked date is not the evaluation execution date.
GDPval-AA v2.1Checked 2026-10-02	1246	Artificial Analysis	Artificial Analysis Intelligence Index v4.3.2; configuration=AA evaluated M3 configuration. Rounded public score; checked date is not the evaluation execution date.
AA Humanity's Last ExamChecked 2026-10-02	39%	Artificial Analysis	Artificial Analysis Intelligence Index v4.3.2; configuration=AA evaluated M3 configuration. Rounded public score; checked date is not the evaluation execution date.
Artificial Analysis Intelligence Index v4.3.2Checked 2026-10-02	29	Artificial Analysis	Artificial Analysis Intelligence Index v4.3.2; configuration=AA evaluated M3 configuration. Rounded public score; checked date is not the evaluation execution date.
AA Terminal-Bench 4.0Checked 2026-10-02	2%	Artificial Analysis	Artificial Analysis Intelligence Index v4.3.2; configuration=AA evaluated M3 configuration. Rounded public score; checked date is not the evaluation execution date.

Strengths

Native image and video input
1M-token context
Open weights

Limitations

Unconfirmed specifications remain undisclosed; missing evaluation results remain unranked

Model evidence · official · Checked 2026-10-02

MiniMax M3 release

Official release or model documentation reviewed 2026-10-02; unconfirmed fields are left unset.

Capability Index source

Artificial Analysis Intelligence Index v4.3.2

Underlying raw metrics and exact model configurations remain authoritative.

Capability Index source

Artificial Analysis methodology

Underlying raw metrics and exact model configurations remain authoritative.

Task lenses preserve raw scores. Radar axes use score / current leader × 100; overall is the equal-weight average of all five indices. Missing evidence is not a zero or a low rank.