Skip to content

MiniMax M3

by Weights & Biases · family minimax-m3 · listed Jun 2026RTVW

MiniMax M3 is a multimodal MoE model with 23B active parameters optimized for coding and agentic workflows.

Current rates USD per 1M tokens

Input$0.23

Output$0.96

Cache read$0.05

Cache write—

Against the market

Input price vs. maker flagships ($/1M tokens)

Output price vs. maker flagships ($/1M tokens)

On the record

Model id: MiniMaxAI/MiniMax-M3
Context window: 262K tokens
Max output: 262K tokens
Input modalities: text, image
Output modalities: text
Reasoning: yes
Tool calls: yes
Open weights: yes
Knowledge cutoff: —
Released: 2026-06-12
Last updated: 2026-06-12
Provider docs: docs.wandb.ai/guides/integrations/inference/

Back of the envelope

Input M tokensOutput M tokens

≈

Also listed at

Nvidiafree in · free out
Kenarifree in · free out
MiniMax Token Plan (minimax.io)free in · free out
MiniMax Token Plan (minimaxi.com)free in · free out
EmpirioLabs AI$0.225 in · $0.90 out
Lilac$0.28 in · $1.10 out
Fireworks AI$0.30 in · $1.20 out
OpenCode Go$0.30 in · $1.20 out
Nebius Token Factory$0.30 in · $1.20 out
NanoGPT$0.30 in · $1.20 out
LLM Gateway$0.30 in · $1.20 out
OpenCode Zen$0.30 in · $1.20 out
ClinePass$0.30 in · $1.20 out
Pioneer$0.30 in · $1.20 out
Abacus$0.30 in · $1.20 out

← All Weights & Biases listings