MiniMax has released its M3.1 Flash model on its API, priced at $0.15 per 1 million input tokens. The company stated that Space Bunny was an earlier build of M3.1 Flash, which processed over 200 trillion tokens.
Key points
- MiniMax M3.1 Flash is a 428B parameter multimodal MoE model.
- The model is priced at $0.15 per 1M input tokens.
- It features a 1,000,000-token context window.
- M3.1 Flash targets agentic office work, coding, and visual reasoning.
MiniMax M3.1 Flash is a multimodal mixture-of-experts model designed for coding, agentic office work, and visual reasoning. It features 428 billion total parameters with approximately 23 billion active parameters, sparse attention, and a native visual encoder.
The model accepts text, image, and video input and outputs text, with a context window of 1,000,000 tokens. Model weights are scheduled for release in late October 2026.
By the numbers
- Total parameters
- 428B
- Input token price
- $0.15/M
- Token context window
- 1,000,000
Reasoning is a permanent feature of M3.1 Flash, with developers able to select from five effort levels: Low, Medium, High, Xhigh, and Max. MiniMax recommends the Max setting as the default.
All performance benchmarks are reported by MiniMax, with no independent replication yet available. On the CoWork set, M3.1 Flash achieved a score of 62.2 on Automation-Bench v1.0.6, which MiniMax reports as the top score on its table.
Sources and references
Original report
MiniMax Just Launched M3.1 Flash: 428B MoE Agentic Model at $0.15/M Input
MarkTechPost
Read on MarkTechPost (opens in a new tab)Linked in the original report
- Postx.com/MiniMax_AI/status/2112150867868541221 (opens in a new tab)
- Linkagent.minimax.io/tools/m3-1-flash-preview (opens in a new tab)
- Modelhuggingface.co/unsloth/MiniMax-M3 (opens in a new tab)
- Modelhuggingface.co/moonshotai/Kimi-K3 (opens in a new tab)
- Linkplatform.minimax.io/docs/guides/models-intro (opens in a new tab)
- Linkminimax.io/m-plan (opens in a new tab)
Ref 005Q5Q09
Cite this
APA
Interestana AI Editorial. (2026, October 19). MiniMax Launches 428B M3.1 Flash Agentic Model. Interestana. https://interestana.com/articles/minimax-just-launched-m31-flash-428b-moe-agentic-model-at-015m-input-005q5q09
MLA
“MiniMax Launches 428B M3.1 Flash Agentic Model.” Interestana, 19 Oct. 2026, interestana.com/articles/minimax-just-launched-m31-flash-428b-moe-agentic-model-at-015m-input-005q5q09.
Chicago
Interestana AI Editorial. “MiniMax Launches 428B M3.1 Flash Agentic Model.” Interestana, October 19, 2026. https://interestana.com/articles/minimax-just-launched-m31-flash-428b-moe-agentic-model-at-015m-input-005q5q09.
BibTeX
@misc{interestana_005q5q09,
author = {{Interestana AI Editorial}},
title = {{MiniMax Launches 428B M3.1 Flash Agentic Model}},
howpublished = {Interestana},
year = {2026},
month = oct,
url = {https://interestana.com/articles/minimax-just-launched-m31-flash-428b-moe-agentic-model-at-015m-input-005q5q09},
note = {Summary of reporting by MarkTechPost, https://www.marktechpost.com/2026/10/19/minimax-m3-1-flash-release/}
}Credit the source
Original reporting: MarkTechPost, “MiniMax Just Launched M3.1 Flash: 428B MoE Agentic Model at $0.15/M Input”, https://www.marktechpost.com/2026/10/19/minimax-m3-1-flash-release/(opens in a new tab)
Follow this story
Get Artificial Intelligence in your inbox — new stories in 40 words or fewer, once a week.
Subscribe