MiniMax M2.7 Running Fastest on SambaCloud

Build Self-Evolving Agents, Use it for Coding or OpenClaw

MiniMax M2.7 (M2.7) is the latest frontier model from MiniMax — and it’s running fastest on SambaCloud. Try it today for coding and OpenClaw on SambaCloud.

It’s the first in the M-series that deeply participates in its own evolution– built for complex agent harnesses, multi-agent collaboration, and high-fidelity productivity tasks. M2.7 hits 56.22% on SWE-Pro , 55.6% on VIBE-Pro for end-to-end project delivery, and 57.0% on Terminal Bench 2. This shows that it is one of the best models for coding and agent harnesses like OpenClaw.

Read the official MiniMax 2.7 announcement here.

MiniMax M2.7 is available today on both the SambaCloud Enterprise and Developer Tier.

Note: If you are using MiniMax M2.5 please migrate over to MiniMax M2.7 immediately.

Why Use MiniMax M2.7?

MiniMax M2.7 marks a significant leap from M2.5, with capability gains driven by the model deeply participating in its own training and harness optimization. Key upgrades include:

• Self-Evolving Agent Harnesses: M2.7 was used internally at MiniMax to autonomously iterate over 100+ rounds of "analyze → plan → modify → evaluate → keep or revert," achieving a 30% performance improvement on internal coding evaluation sets. On MLE Bench Lite (22 ML competitions), it earned 9 gold, 5 silver, and 1 bronze medals — a 66.6% medal rate that ties Gemini 3.1 and trails only Opus 4.6 and GPT-5.4.
• SOTA Software Engineering: Production-grade performance across SWE-Pro (56.22%), SWE Multilingual (76.5%), Multi-SWE-Bench (52.7%), and NL2Repo (39.8%). M2.7 doesn't just write code — it correlates monitoring metrics with deployment timelines, performs causal reasoning on production incidents, and has been used internally to cut live incident recovery times to under three minutes.
• Native Agent Teams: Multi-agent collaboration is internalized as a native capability, not bolted on through prompting. M2.7 stably anchors role identity, performs adversarial reasoning across teammates, and makes autonomous decisions inside complex state machines — making it ideal for orchestrating agent teams in CrewAI, OpenClaw, or custom frameworks.
• Office & Professional Work: Top-tier results on GDPval-AA (ELO 1495) and Toolathon (46.3%). M2.7 handles multi-round, high-fidelity editing across Word, Excel, and PowerPoint, and can execute end-to-end financial workflows — reading annual reports, building revenue models, and producing PPTs and research reports of junior-analyst quality.
• Character Consistency & Emotional Intelligence: Improvements in persona stability and conversational depth open up new product directions beyond pure productivity, including interactive entertainment experiences like MiniMax's open-sourced OpenRoom demo.

MiniMax M2.7 ranks alongside Opus 4.6, GPT-5.4, and Gemini 3.1 on demanding agentic and software engineering tasks — while remaining open-weight and far more cost-efficient. For teams running long-horizon agent workflows in production, M2.7 sits firmly in the most attractive cost-to-intelligence quadrant available today.

Get Started Quickly with SambaCloud

With just a few lines of Python, you can call MiniMax M2.7 to write code, critique its own solution, and ship an improved version. This mirrors the autonomous self-optimization loop M2.7 uses internally to iterate on its own performance.

import os
from sambanova import SambaNova
# Initialize client using environment variable for security
client = SambaNova(
    api_key=os.environ["SAMBANOVA_API_KEY"],
    base_url="https://api.sambanova.ai/v1",
)

system_prompt = "You are a helpful assistant. Your name is MiniMax-M2.7 and is built by MiniMax." 

user_prompt = ( "Write a Python function that finds the longest palindromic substring in a string. " "Then act as a senior reviewer: critique your own solution for correctness, " "edge cases, and time complexity. Finally, produce an improved version that " "addresses every issue you identified, and explain what changed and why." )

response = client.chat.completions.create(
    model="MiniMax-M2.7",
    messages=[ 
{"role": "system", "content": system_prompt}, 
{"role": "user", "content": user_prompt},
    ],
    
    temperature=0.1,
    top_p=0.95
)
print(response.choices[0].message.content)

Check out SambaCloud and explore MiniMax M2.7 in the playground, or generate an API key and integrate it into your agentic application today.

MiniMax M2.7 brings self-evolving agentic intelligence to production. SambaNova delivers it the fastest.