Model WarsJuly 8, 2026via The Decoder
Chinese AI startup MiniMax plans to open-source a 2.7 trillion parameter model later this year
Why it matters
MiniMax's trillion-parameter open-source release signals an acceleration in China's AI capability race and could democratize access to frontier-scale models globally, challenging closed-model incumbents.
Key signals
- MiniMax developing 2.7 trillion parameter LLM
- Model planned for open-source release
- Expected release timeline: later in 2026
- Chinese AI developer entering frontier-scale model competition
- Potential competitive pressure on closed US/EU model providers
The hook
2.7 trillion parameters. A Chinese startup just announced it's going open-source—and the West should be paying attention.
Chinese AI developer MiniMax is working on a new large language model with 2.7 trillion parameters. MiniMax plans to release the model as open source.