ai-trend-notifier
← wiki

$ cat wiki/models/minimax-m3.md

MiniMax M3

Spec

AttributeValue
Developer[MiniMax](/wiki/entities/minimax)
ReleasedJune 1, 2026
AnnouncedJune 1, 2026 (MiniMax official blog)
ArchitectureMSA (MiniMax Sparse Attention) — sparse MoE
Total parameters~428B
Active parameters~23B per token
Context window1,000,000 tokens (1M); min. 512K guaranteed
Max output131,072 tokens
ModalitiesText, image, video (native multimodal, no post-hoc adapters)
LicenseOpen-weight (HuggingFace)
API accessGlobal (including English)
PricingAvailable via API; pricing on OpenRouter
AvailabilityAPI (global, including English) + open-weight on HuggingFace

Release Date

June 1, 2026 — MiniMax official blog and HuggingFace release. Captured by this wiki July 17 due to WAIC 2026 coverage surfacing the model.

Benchmarks

BenchmarkMiniMax M3ComparisonNotes
SWE-Bench Pro59.0%Beats GPT-5.5 (prior OpenAI frontier)Coding agent
BrowseComp83.5Beats Claude Opus 4.7 (79.3)Browser agent
OSWorld-Verified70.06%Computer use
Note on comparisons: GPT-5.6 Sol (July 9, 2026) is the current OpenAI frontier; M3's comparisons are vs. GPT-5.5. Claude Sonnet 5 (June 30, 2026) surpasses Opus 4.7 on most benchmarks. M3's BrowseComp result is nonetheless impressive for a Chinese open-weight model.

Architecture

  • MSA (MiniMax Sparse Attention): proprietary sparse attention mechanism
    • 9× prefill speedup at 1M-token context vs. M2 (prior MiniMax model)

    • 15× decoding speedup at 1M context vs. M2

    • 1/20th per-token compute vs. M2 at 1M context
  • MoE: ~428B total, ~23B active — efficient for inference relative to total parameter count

Use Cases

  • Agentic coding (SWE-Bench Pro 59.0%)
  • Browser-based agent tasks (BrowseComp 83.5)
  • Computer use (OSWorld 70.06%)
  • Long-context document analysis (1M tokens)
  • Multimodal workflows (text + image + video)

Significance

MiniMax M3 is notable as:

  1. One of the first Chinese open-weight models to surpass a leading US closed model (Claude Opus 4.7) on an agent benchmark (BrowseComp 83.5 vs. 79.3)
  2. A frontier-class open-weight model with 1M context and native multimodality from a non-Tier-1 (non-Alibaba/ByteDance) Chinese lab
  3. Available globally via API and open-weight — expanding Chinese AI model access internationally

Compared To

ModelOrgSWE-Bench ProNotes
MiniMax M3MiniMax59.0%Open-weight, 428B MoE, BrowseComp 83.5
GPT-5.5OpenAI~58.6%M3's reference at launch; now superseded by GPT-5.6
Claude Opus 4.7Anthropic~64.3%BrowseComp 79.3 (M3 beats it here)
GLM-5.2Z.aiComparable Chinese open-weight frontier
Claude Sonnet 5Anthropic63.2%Current mid-tier Anthropic (GA June 30)

Related

Referenced by

Sources