MiniMax M3.1 Flash unmasked as Space Bunny: #1 on OpenCode for 12 days, open weights soon
MiniMax_AI · x · 2026-10-21
MiniMax officially confirmed that the anonymous "Space Bunny" model is MiniMax M3.1 Flash, with open weights coming (no date yet).
Key specs and benchmarks (per Sophia Yang's summary):
- #1 on OpenCode for 12 consecutive days
- 1M context with text, image, and video input
- 1/3 the cost of M3 at 3x the speed: $0.15/$0.60 per 1M input/output tokens, cache reads $0.015
- Automation-Bench: 62.2%, ahead of Opus 5 (50.3%) and GPT-5.6 Sol (45.8%)
- DeepSWE: 73.7%, roughly tied with Opus 5 (74.0%), behind DeepSeek V4.1 Flash (77.0%)
- ALE: 29.6% for $111 total, close to Opus 5 (32.2%) and GPT-5.6 Sol (30.6%); Opus 5.5 scores 38% but costs over $1,000
- BabyVision: 91.0%, just behind DeepSeek V4.1 Flash (92.5%)
Related event: MiniMax Unveils M3.1 Flash as Mystery "Space Bunny," Open-Weights Thursday(41 posts)→
More from Models
- Ampersand debuts first model '&', built on long-horizon simulation stack — niloofar_mire · 2026-10-21
- Hugging Face quietly added model file size display, a big QoL win — vboykis · 2026-10-21
- Humans& launches Ampersand, first model with privacy built in, plus a funny RL-plot joke — niloofar_mire · 2026-10-21
- A single GLM neuron keeps firing on both sci-fi self-replicating malware and real cyber risk — burny_tech · 2026-10-21
- Gemini 4 Argon tops Isoquant's inference benchmark with 4.38x mean GPU speedup — ycombinator · 2026-10-21
- Gemini 4 Argon Weekly Quota Hits 0% Without a Single Message Sent in Antigravity — koltregaskes · 2026-10-21