Skip to content

MiniMax News: Latest Releases & Events

MiniMax released M3.1 Flash in October 2026, revealing that the leaderboard-topping anonymous system Space Bunny was an early version of it. That followed M3 in June, an open-weight model with a 1M-token context window and native multimodality, and M2.7 in March. In February, Anthropic accused MiniMax, DeepSeek and Moonshot of distilling Claude, the same month MiniMax launched M2.5.

Who is MiniMax? → Big AI Players

All 8 MiniMax entries of the AI News timeline, newest first.

2026

October 2026

2026-10-19

  • MiniMax Launches M3.1 Flash, Reveals Space Bunny
    MiniMax released M3.1 Flash, a next-generation model delivering frontier-level agentic, coding, computer use, and cybersecurity performance with ~3x faster inference and 70% lower costs ($0.15/$0.60 per 1M input/output tokens). The company also revealed that 'Space Bunny,' the anonymous system that topped global leaderboards while processing over 200 trillion tokens, was an early version of M3.1 Flash, as MiniMax confirmed on social media. (minimax.io)

June 2026

2026-06-04

  • MiniMax Releases M3: Open-Weight Model with 1M Context and Native Multimodality
    MiniMax launched M3, the first open-weight model to combine frontier coding and agentic capabilities, a 1M-token context window via its proprietary MiniMax Sparse Attention (MSA) architecture, and native multimodality trained from step zero on 100T+ tokens. The model achieves top-tier scores on coding and agentic benchmarks including BrowseComp (83.5, surpassing Opus 4.7), and is now available via API and OpenCode CLI, with full open-source release planned for HuggingFace and GitHub. (minimax.io)

March 2026

2026-03-28

  • Anthropic loses market share to Chinese AI models
    Anthropic's market share has fallen from 29.1% to 13.3% in a year as six Chinese models now outrank Claude on OpenRouter. The company has spent $10 billion on inference and training against just $5 billion in revenue, while security researchers are abandoning Claude over increasingly aggressive safety guardrails that block legitimate work. (go.theregister.com)
    Anthropic DeepSeek

2026-03-18

  • MiniMax Releases Self-Evolving M2.7 Model
    MiniMax announced the M2.7 model, a Mixture-of-Experts architecture that utilizes a self-evolution training process involving over 100 iterations to achieve significant performance gains. The new model demonstrates top-tier benchmarks in coding and complex reasoning tasks while offering a cost-effective pricing structure. (mp.weixin.qq.com)

February 2026

2026-02-24

  • Anthropic accuses Chinese labs of distilling Claude
    Anthropic accused Chinese AI labs DeepSeek, Moonshot, and MiniMax of conducting industrial-scale distillation campaigns using thousands of fraudulent accounts to extract knowledge from its Claude models, raising national security concerns. An in-depth analysis of the campaigns suggests the impact varies significantly, noting that while MiniMax and Moonshot generated millions of exchanges, DeepSeek's attributed activity was comparatively small. (go.theregister.com)
    Anthropic DeepSeek Moonshot AI

2026-02-19

2025

December 2025

2025-12-23

  • MiniMaxAI Releases VIBE Dataset
    MiniMaxAI launches the VIBE benchmark dataset "Visual & Interactive Benchmark for Execution" for evaluating AI agents on full-stack web development tasks through a dataset of 200 structured coding challenges. (huggingface.co)

  • MiniMax releases M2.1 language model upgrade
    MiniMax launches M2.1 with enhanced multi-language programming support for Rust, Java, and others, optimized for complex real-world tasks and office workflows while improving response efficiency. (minimax.io)