Skip to main content

​
Available models

Model IDParametersContextSpeedBest For
minimax-m2.5—196KMediumAgentic coding, SWE-bench

​
Why MiniMax M2.5?

MiniMax M2.5 achieved 80.2% on SWE-bench, making it one of the top models for real-world software engineering tasks. It excels at:
  • Multi-file code editing — understands project context across files
  • Bug fixing — diagnoses and fixes real GitHub issues
  • Agentic workflows — works well with tool use and multi-step coding tasks
  • Long context — 196K tokens covers large codebases

​
Recommendation

Use minimax-m2.5 for agentic coding — when you need a model that can navigate codebases, edit multiple files, and fix real bugs. Ideal for AI coding assistants like Cline, Roo Code, and similar tools. 
from openai import OpenAI

client = OpenAI(base_url="https://kymaapi.com/v1", api_key="ky-...")

response = client.chat.completions.create(
    model="minimax-m2.5",
    messages=[{"role": "user", "content": "Fix the race condition in this code: ..."}]
)