MiniMax M2

MiniMax M2 is now available on Ollama's cloud, according to an Ollama announcement dated October 28, 2025. The company describes it as a model built for coding and agentic workflows. It can be run with the command ollama run minimax-m2:cloud, or pulled with ollama pull minimax-m2:cloud for use in supported tools.

Ollama states that, according to benchmarks from Artificial Analysis, MiniMax-M2 shows competitive general intelligence across mathematics, science, instruction following, coding, and agentic tool use, and that its composite score ranks first among open-source models globally. Ollama says the model is engineered for end-to-end developer workflows, including multi-file edits, coding-run-fix loops, and test-validated repairs, with strong performance reported on Terminal-Bench and (Multi-)SWE-Bench-style tasks. It is also described as planning and executing long-horizon toolchains across shell, browser, retrieval, and code runners, and as recovering from flaky steps in BrowseComp-style evaluations.

On efficiency, Ollama lists 10 billion activated parameters and 230 billion total parameters, citing lower latency, lower cost, and higher throughput for interactive agents and batched sampling.

Ollama published usage instructions for several tools. For VS Code, users pull the model, open the Copilot chat sidebar, select Manage models, choose Ollama under the provider dropdown, and select the model. For Zed, users pull the model, open the agent panel, choose Configure from the model dropdown, select Ollama as the LLM provider, confirm the host URL http://localhost:11434, connect, and select a model. For Droid, a configuration entry names the model as minimax-m2:cloud with base_url http://localhost:11434/v1 and max_tokens 16384.

Cloud models such as minimax-m2:cloud can also be accessed through ollama.com's cloud API using an API key. Ollama's documentation now includes sections on using Ollama with Codex, Cline, VS Code, Zed, Droid, and Roo code.