MiniMax M3.1 Flash
MiniMax M3.1 Flash is MiniMax's October 19, 2026 Flash model for agent work, coding, computer use, and cybersecurity.
Provider
MiniMax
Model family
MiniMax M
Multimodal coding LLM
Cost tier
Flash
Status
Current
Release Oct 19, 2026
Why teams choose it
Pay-as-you-go users can keep calling M2.5 and M2.7 until those models retire on November 18, 2026
Plan traffic to those IDs is already routed to M3.1 Flash.
MiniMax says everyday output can reach 200+ tokens per second, with a median above 170
That is a vendor measurement.
MiniMax says agent usage cost is about 70% lower than the predecessor and that Space
Bunny was an early anonymous version of this model.
Tradeoffs to know
- Weights are not released yet. MiniMax says they will be released later.
- MiniMax M3 stays on the API. M3.1 Flash does not replace that model ID.
- Speed and benchmark figures are MiniMax's reported numbers.
Benchmarks
- Source (vendor-reported)
- https://www.minimax.io/blog/minimax-m31f
- Cybergym (vendor-reported)
- 0.82
- Osworld V2 (vendor-reported)
- 61.7
- Deepswe V1 1 (vendor-reported)
- 0.737
- Bankertoolbench (vendor-reported)
- 76.2
- Agents Last Exam (vendor-reported)
- 29.6
Technical specs
- Inputs
- text, image
- Outputs
- text
- License
- Proprietary API
- API ID
MiniMax-M3.1-Flash- Context tokens
- 1,000,000
- Input Usd Per Million
- 0.15
- Output Usd Per Million
- 0.6
- Cache Read Usd Per Million
- 0.015
- Capabilities
- 1 million token context window
- Tunable thinking depth on the MiniMax API
- Available in the API, M Plan, and MiniMax Code
MiniMax M family lineup
Current models
Related models
MiniMax M3.1 Flash FAQ
What is MiniMax M3.1 Flash?
1 Flash is MiniMax's October 19, 2026 Flash model for agent work, coding, computer use, and cybersecurity. 1-Flash. MiniMax's model list calls it a frontier multimodal coding model with a 1 million token context window and tunable thinking depth. 015 cached input per million tokens. It is available through the MiniMax API, M Plan, and MiniMax Code. MiniMax says model weights will be released later, and that a Pro model will arrive later. 1 Flash. 2.
When does MiniMax M3.1 Flash fit best?
Low-cost coding agents
What should teams watch out for with MiniMax M3.1 Flash?
Weights are not released yet. MiniMax says they will be released later.
Explore next
Models, tools, and comparisons that connect to this reference.