MiniMax M3

MiniMax[text][open-weights]

Open-weight 428B MoE for coding and agentic work with 1M context, native multimodal input, and computer use

[interactive architecture ↓]

Input

$0.30

per 1M tokens

Output

$1.20

per 1M tokens

Context

1.0M tokens

Max Output

512K tokens

Released

2026-05-31

Features

[open-weights][1m-tokens-context][top-open-weight-swe-bench-pro-(59.0%)][text,-image,-and-video-input][computer-use]

Technical Specifications

LicenseMiniMax Community License

Benchmarks

BrowseComp83.5%

MCP Atlas74.2%

MMMU-Pro78.1%

OSWorld-Verified70.1%

SWE-bench Pro59.0%

SWE-bench Verified80.5%

Tau2-bench88.9%

Terminal-Bench66.0%

The Lab

location: Shanghai, Chinafounded: 2022backed by: Private investment

Founded in 2022 and headquartered in Shanghai, MiniMax is a private Chinese AI lab developing foundation models for text, voice, and video. Its MiniMax M family powers conversational AI products including the Hailuo consumer assistant. The lab competes with Moonshot, Zhipu, and DeepSeek in China's crowded frontier-model market.

more about MiniMax →

Interactive Architecture

Take MiniMax M3 apart, layer by layer. The 3D teardown below shows its 60 transformer layers (0 dense + 60 MoE), GQA attention, and expert routing, with values from the public model config.

60 layers (0 dense → 60 MoE)

GQA attention · 64Q heads

4+0 of 128 experts active

428B params · 23B active

Token embedding

6144-dim vectors over a large 200K-token vocabulary, sized for broad multilingual coverage.

60 layers, mixed dense/sparse MLP

Pre-norm blocks where each layer's MLP can be dense (12,288 intermediate) or sparse MoE (3,072 intermediate) via a per-layer selector, letting MiniMax trade capacity for speed layer by layer.

MiniMax Sparse Attention

GQA (64 query / 4 KV heads) wrapped in a block-sparse scheme: a lightning indexer with 4 index heads and 128-token blocks selects which KV blocks each query attends to, making the 1M context affordable. (Rendered as the GQA block.)

128-expert MoE

A router scores 128 experts and activates 4 per token, spending ~23B of ~428B. The low active count keeps inference cheap relative to the stored knowledge.

Output head

Final RMSNorm then 200K logits. M3 is a native-multimodal model built to use its full 1M context.

What vocabulary and embedding size does MiniMax M3 use?+

6144-dim vectors over a large 200K-token vocabulary, sized for broad multilingual coverage.

How many transformer layers does MiniMax M3 have?+

Pre-norm blocks where each layer's MLP can be dense (12,288 intermediate) or sparse MoE (3,072 intermediate) via a per-layer selector, letting MiniMax trade capacity for speed layer by layer.

What attention does MiniMax M3 use?+

How does the Mixture of Experts in MiniMax M3 work?+

A router scores 128 experts and activates 4 per token, spending ~23B of ~428B. The low active count keeps inference cheap relative to the stored knowledge.

How does MiniMax M3 produce output tokens?+

Final RMSNorm then 200K logits. M3 is a native-multimodal model built to use its full 1M context.

scene values sourced from:HuggingFace config HuggingFace config

Model ID: minimax-m3

The Lab

location: Shanghai, Chinafounded: 2022backed by: Private investment

more about MiniMax →

Interactive Architecture

Take MiniMax M3 apart, layer by layer. The 3D teardown below shows its 60 transformer layers (0 dense + 60 MoE), GQA attention, and expert routing, with values from the public model config.

60 layers (0 dense → 60 MoE)

GQA attention · 64Q heads

4+0 of 128 experts active

428B params · 23B active

Token embedding

6144-dim vectors over a large 200K-token vocabulary, sized for broad multilingual coverage.

60 layers, mixed dense/sparse MLP

Pre-norm blocks where each layer's MLP can be dense (12,288 intermediate) or sparse MoE (3,072 intermediate) via a per-layer selector, letting MiniMax trade capacity for speed layer by layer.

MiniMax Sparse Attention

128-expert MoE

A router scores 128 experts and activates 4 per token, spending ~23B of ~428B. The low active count keeps inference cheap relative to the stored knowledge.

Output head

Final RMSNorm then 200K logits. M3 is a native-multimodal model built to use its full 1M context.

What vocabulary and embedding size does MiniMax M3 use?+

6144-dim vectors over a large 200K-token vocabulary, sized for broad multilingual coverage.

How many transformer layers does MiniMax M3 have?+

Pre-norm blocks where each layer's MLP can be dense (12,288 intermediate) or sparse MoE (3,072 intermediate) via a per-layer selector, letting MiniMax trade capacity for speed layer by layer.

What attention does MiniMax M3 use?+

How does the Mixture of Experts in MiniMax M3 work?+

A router scores 128 experts and activates 4 per token, spending ~23B of ~428B. The low active count keeps inference cheap relative to the stored knowledge.

How does MiniMax M3 produce output tokens?+

Final RMSNorm then 200K logits. M3 is a native-multimodal model built to use its full 1M context.