Frontier Agentic Performance at a Fraction of the Cost. Build Fast, Cheap, Long-Running Agents on SambaCloud
MiniMax M3.1 Flash is the next-generation Flash model from MiniMax, built for agentic tasks, coding, computer use, and cybersecurity. It delivers a generational leap in capability over its predecessor while making inference nearly 3x faster and cutting usage costs by 70%. It hits 73.7% on DeepSWE v1.1, 61.7 on OSWorld V2, and 82% on CyberGym, placing it alongside top models on the benchmarks that matter for real-world agents.
MiniMax M3.1 Flash is the latest model from MiniMax and it's now available on SambaCloud at over 200 tokens/s.
Note: MiniMax M3 on SambaCloud is being rerouted to MiniMax M3.1 Flash. If you are calling MiniMax M3, your requests will be served by MiniMax M3.1 Flash. Please update your model string to MiniMax-M3.1-Flash.
Why Use MiniMax M3.1 Flash?
MiniMax M3.1 Flash improves across a broad range of benchmarks, reaching the level of leading models in complex agentic tasks, computer use, software development, and cybersecurity. Key upgrades include:
Complex Tasks and Automation
M3.1 Flash ranks among the top performers on AutomationBench-AA, completing real enterprise workflows across SaaS systems end to end. That means solving practical problems in sales, marketing, operations, customer service, finance, and HR, not just answering questions about them. It scores 76.2 on BankerToolBench, which stresses multi-step planning, consistency across documents and files, financial compliance, and chained tool calls.
Coding and Software Engineering
M3.1 Flash achieves 73.7% on DeepSWE v1.1, placing it alongside top models. DeepSWE focuses on complex software tasks where the model must independently understand a codebase, build features, fix issues, and validate its own work through testing. M3.1 Flash executes reliably across long sequences of steps, supporting full software development workflows at a fraction of the cost.
Computer Use
M3.1 Flash scores 61.7 on OSWorld V2, which measures AI computer operation and cross-application task execution, and 29.6 on Agents' Last Exam (ALE), which evaluates performance on real professional work with economic value. It ranks first among Chinese models on both. The model interacts with real environments, uses tools, and executes long task sequences, moving beyond understanding a task to delivering results.
Cybersecurity
M3.1 Flash achieves 82% on CyberGym, demonstrating leading capability in vulnerability discovery and validation. In MiniMax's internal evaluations by security experts, the model performed strongly in code auditing, vulnerability patching, penetration testing, malware analysis, and reverse engineering.
3x Faster, 70% Cheaper
M3.1 Flash uses more aggressive sparsity than its predecessor, cutting overall usage cost for agentic tasks by 70% and making online inference nearly 3x faster. On SambaCloud, that efficiency compounds with SambaNova's dataflow architecture to deliver over 200 tokens/s.
MiniMax M3.1 Flash pricing on SambaCloud:
Across AutomationBench for complex agentic execution, DeepSWE for real-world coding, and ALE for computer use, M3.1 Flash delivers performance on par with top models at a very low task cost. For teams running agents in production, where every step in a long loop is a model call, that combination of speed and cost sets a new standard for what a Flash-tier model can do.
Get Started Quickly with SambaCloud
With just a few lines of Python, you can call MiniMax M3.1 Flash to plan a multi-step task, call tools, and iterate on its own work, the same agentic loop it uses to drive enterprise workflows and fix real codebases.
Check out SambaCloud and explore MiniMax M3.1 Flash in the playground, or generate an API key and integrate it into your agentic application today.
MiniMax M3.1 Flash brings frontier agentic performance, coding, and computer use to production at a fraction of the cost. SambaNova delivers it at over 200 tokens/s.
