Chinese AI start-up Moonshot has unveiled the Kimi K3 large language model featuring one million context tokens. The model is means to compete with Anthropic’s Opus 4.8 based on competitive coding benchmarks found in early tests. Benchmark leaks show Kimi K3 performance targets Anthropic’s Claude Fable5 directly. Industry watchers say the release escalates the AI competition between Chinese and Western developers. Moonshot presents the model as a step towards narrowing the performance gap compared to leading global alternatives.