MiniMax M3.1 Flash Preview
MiniMax's new Flash-class coding model, launched Sept 27, 2026 inside MiniMax Code and the Token Plan — 1M context, always-on thinking with five effort levels, no published price or weights.
Why it matters
A tunable-speed reasoning coder (low-to-max thinking effort) for everyday dev work, which developers matched trait-for-trait to the anonymous 'Space Bunny Alpha' that OpenCode's co-founder was recommending.
What you could build with it
A startup building AI coding tools could route routine edits and bug fixes to a fast Flash-class model like this inside its own agent product, reserving expensive flagship models for hard refactors — cutting cost per task on everyday development work.
Does it hold up?
Real organic buzz (OpenCode co-founder Dax suggested trying Space Bunny on the same prompts used for Opus 5.5 tests; multiple writeups praise its frontend/code output), but there are no published benchmarks, no per-token price, no API outside MiniMax Code / Token Plan, and thinking is always on — treat as evaluation-only until MiniMax publishes numbers.
Built with MiniMax M3.1 Flash Preview
- Space Bunny Alpha: The Stealth Model and the MiniMax Cluescellcog.ai · Investigative writeup matching the anonymous OpenRouter model 'Space Bunny Alpha' (1M context, mandatory reasoning, five effort levels, $0 preview) trait-for-trait to M3.1-Flash-Preview.
- MiniMax M3.1-Flash-Preview: A Fast, Long-Context Model for Everyday Codingmedium.com · Developer writeup covering the Sept 27 launch, 1M context window, five reasoning-effort levels, and Token Plan / MiniMax Code availability.
- MiniMax slips a new coding model into its agent tool without a price tagstartupfortune.com · Report on the quiet Sept 27 rollout, the launch post from MiniMax's agent account on X, and developers linking the model to the 'Space Bunny Alpha' sighting on OpenRouter.
Learn more
technical comparison vs MiniMax M3 →
First spotted on article: source.
More AI models
ElevenLabs Eleven v4 + v4 Turbo: new emotive TTS architecture with a 100ms real-time variantElevenLabs released Eleven v4 and Eleven v4 Turbo on September 28: a new text-to-speech architecture with inline…models · JEV 0.72VoiceStudio: open-source, fully-local ElevenLabs alternative with voice cloning, dubbing and dictation in 646 languagesVoiceStudio is an open-source, fully-local voice AI studio: voice cloning from a 3-second sample, voice design, video…models · JEV 0.71DeepSeek V4-Flash official API: public beta with upgraded agent capabilities, Responses API and Codex supportDeepSeek's official V4-Flash is now a public beta API with massively upgraded agent capabilities - benchmark scores…models · JEV 0.69Mistral Forge: enterprise platform for training and continuously improving proprietary modelsMistral AI launched Forge (Sep 28) — a full-lifecycle model training platform (pre-training, SFT, DPO/ODPO, RL…models · JEV 0.69kev: open, trainable Jev-like family of small decision models on Qwen3.5/3.8A family of small decision models (0.8B to 27B) built on Qwen3.5/3.8 that reproduces Jev's prefill-only typed-decision…models · JEV 0.68GPT-Synopsys: OpenAI and Synopsys multi-year deal to build an AI model that operates EDA toolsOpenAI and Synopsys signed a multi-year agreement on Sep 30 to jointly develop GPT-Synopsys, a specialized model that…models · JEV 0.66
Get the week's best AI launches, plus 3 ideas worth building
One email every Saturday. Ranked by traction, not hype. Free.