Guozhen AIGlobal AI field notes and model intelligence

Realtime AI News

MiniMax Open-Sources MiniMax Code, a Terminal Coding Agent

According to TechNode, MiniMax has open-sourced MiniMax Code, a terminal coding agent that developers can pick up and run directly from the command line. The move signals that MiniMax wants to compete for coding-agent users through open source rather than a closed product, though the repository, licence and support scope still need official confirmation.

Published
MiniMax 开源 MiniMax Code:把终端编程智能体交给开发者
Image source: minimax.io

Another model maker has opened up its coding agent. According to TechNode, MiniMax has open-sourced MiniMax Code, a terminal coding agent that developers can obtain and run directly.

A terminal coding agent lives in the command line. It reads and writes files inside a project, runs commands and works through multi-step coding tasks on its own, which puts it closer to the engineering workflow than an editor plugin: it can edit a repository, run tests, and be scripted into continuous integration pipelines.

What is public so far is deliberately minimal. The company is MiniMax, the product is MiniMax Code, and the action is open-sourcing; repository location, licence terms, supported models and runtimes, and whether a hosted version exists all have to be confirmed from the official release. Those are exactly the details that decide whether developers treat it as a production tool or as something to experiment with first.

The timing matters. Coding agents have been one of the most heavily contested categories over the past year: closed products have won paying users by being tightly coupled to their own models, while open alternatives try to enter enterprises on the strength of self-hosting, auditability and customisation. By open-sourcing a terminal agent, MiniMax shifts the contest from model scores back to engineering usability.

Code stays inside the internal network, the back-end model can be swapped, and the tool can be adapted to internal conventions, which is usually the first set of questions any team asks when evaluating a commercial coding assistant.

Open-sourcing also shifts maintenance responsibility outward. Whether the project keeps shipping, accepts outside contributions and stays in step with new models determines how long it survives in production, and plenty of open coding tools have gone quiet after an initial burst of attention. That risk belongs in any adoption decision.

What to watch next: the full repository and licence, the models and runtimes it supports, whether independent benchmark results appear, and whether MiniMax positions it as the default companion to its own models. Those answers will show whether this is a long-term project or a single announcement.

Why it matters

If MiniMax publishes clear repository, licence and support details, MiniMax Code can become a credible candidate for coding assistants that run behind the corporate firewall. Without those details, it is more likely to remain a developer experiment than an enterprise tool.

MiniMaxCoding AgentOpen Source
Back to AI Daily

Nearby Updates

All

09/21, 14:06

Amazon blocks Meta's Muse AI agent from shopping on its site as agentic commerce fight escalates

Amazon has blocked Meta's new Muse personal AI agent from shopping on Amazon.com, citing security, privacy and transparency concerns, after failing to persuade Meta to exclude the site from the experience. Muse users now see a popup saying continued access by an unauthorized AI agent violates Amazon's Conditions of Use, and Amazon says it never authorized the agent.

09/21, 14:22

Tsinghua and Infinigence Open-Source RPent, an Embodied Agent Stack for Real Robots

RPent, an embodied-agent infrastructure project launched jointly by Tsinghua University, Infinigence AI and Zhengxing Innovation, has been open-sourced with its code published on GitHub. It reports 92.6% task success on the LIBERO-PRO benchmark and more than a 7x speed-up in end-to-end task completion, with real-robot demos including pouring steel balls, two-arm plate wiping and uncovering hidden objects.

09/21, 12:50

Qwen open-sources Qwen-Image-2.1, a 7B visual generator that unifies text-to-image and editing

Qwen released Qwen-Image-2.1 on Hugging Face on September 21, a unified text-to-image generation and image-editing model whose visual generation component carries just 7B parameters across 32 single-stream DiT layers. The release adds native RGBA transparency, support for up to 10 reference images and mask-based local edits, and ships a QwenImage21Pipeline integration in diffusers.

09/21, 15:00

OpenAI Expands Academy With New Learning Paths

OpenAI has announced an expansion of OpenAI Academy with new learning paths aimed at employees, developers, leaders, educators and students. The stated goal is to help those groups build and demonstrate practical AI skills.