Cast AI, the automation platform for cloud-native and AI workloads, today announced that MiniMax M3 is available through Kimchi Coding, its autonomous multi-model coding agent. Kimchi becomes the first coding agent to offer MiniMax M3 access, rolling out now through its Early Access program at kimchi.dev.
The announcement comes as model selection becomes a strategic question for enterprises. The question has shifted from which single AI model to adopt to which model should handle each task. As The Wall Street Journal reported this month, businesses are increasingly routing work across a mix of open-weight and frontier commercial models, reserving the most capable models only for the hardest tasks, an approach that can deliver major cost reductions on some AI-assisted work.
“We built Kimchi to give every developer frontier-quality AI coding without frontier-sized bills or data risk, thanks to our built-in token-optimized model orchestrator,” said Laurent Gil, President and Co-Founder of Cast AI and a Founding Premier Governing Board Member of the Tokenomics Foundation. “MiniMax M3 is the model Kimchi was designed for: open weights, frontier performance, and economics that work at scale.”
Kimchi Coding includes a smart and autonomous multi-model harness obsessed with token optimization. Its model orchestration engine routes each task to the best-fit model based on complexity and cost, reserving frontier models for the hardest evaluations while other open-weight models handle the bulk of the work. The harness scores generated code and runs continuous feedback loops, so output stays accurate while token use is optimized at every step, which is what makes Kimchi both more accurate and more cost-efficient than relying on a single commercial model.
Budget governance with built-in token optimization enforces hard spend caps from individual API keys to entire organizations, with automatic termination of runaway agentic loops. A real-time FinOps dashboard tracks every token, user, and model, with cost attribution per developer, team, and project.MiniMax M3 becomes the new default builder model within Kimchi’s orchestration layer. M3 scored 59% on SWE-bench Pro, the benchmark based on real GitHub issues, placing it ahead of several leading commercial models on the same evaluation. Its MiniMax Sparse Attention architecture reduces per-token compute at one-million-token context to 1/20th of prior levels, with decoding speeds 15x faster. For enterprises running long-horizon coding tasks or large codebase traversals, this directly reduces both cost and time-to-completion.
“Cast AI’s Kimchi Coding is an ideal partner to bring M3 to teams that demand both performance and sovereignty,” said Leanna Ren, VP Global Marketing at MiniMax. “M3 was built to push the boundaries of open-weight models. Kimchi’s model orchestrator and data sovereignty controls make it the natural home for M3 in the enterprise. Together, we’re giving developer teams frontier-quality AI coding inside their own environment, at a fraction of the cost of commercial models alone.”
In continuous shadow-mode evaluations against a commercial-models-only baseline, Kimchi Coding delivers 2.5x lower cost while matching or exceeding quality on spec match and test pass rates.
Kimchi Coding is immediately available as a serverless deployment on Cast AI’s optimized inference clusters or as a fully sovereign deployment in a customer’s AWS, GCP, Azure, or on-prem environment, with air-gap support and compliance readiness for SOC 2, GDPR, ISO 27001, and PCI-DSS.
Access to MiniMax M3 is rolling out in phases. Organizations can join the waitlist at kimchi.dev.
About Kimchi Coding
Kimchi is an open-source, terminal-native coding agent and AI assistant designed to streamline agentic software development and AI-assisted coding workflows. Developed by Cast AI, it bridges the gap between AI models and practical engineering, assigning the best-fit model to each step of a task and drawing on frontier models for the hardest work, while open-weight models handle the rest. Kimchi runs serverless on Cast AI’s inference cloud or deploys inside a customer’s own AWS, GCP, Azure, or on-prem environment, with full air-gap support. Learn more and join Early Access at kimchi.dev.
