Story perspectives

Chinese Startup Launches Kimi K2: AI Outperforms Competitors

11/10/2025

Summary

Story summary

Moonshot AI, a Chinese startup, released Kimi K2 Thinking on November 6, 2025, an open-source reasoning model that outperforms OpenAI's GPT-5 and Anthropic's Claude Sonnet 4.5.
It scored 44.9% on Humanity’s Last Exam and 60.2% on BrowseComp, showing strong reasoning and coding.
Kimi K2 Thinking can perform 200–300 sequential tool calls autonomously for wide applications.
Development costs are significantly lower than competitors, suggesting open models may disrupt the field.

History

Historical Context

OpenAI launched GPT-2 in February 2019, a significant advancement in natural language processing with 1.5 billion parameters, following the introduction of the transformer architecture in 2017. This innovation spurred competition, leading to the development of models like GPT-3 in 2020, which further expanded AI capabilities and raised concerns about misinformation and ethical use in various applications.

1 of 2

Story summary

Moonshot AI, a Chinese startup, released Kimi K2 Thinking on November 6, 2025, an open-source reasoning model that outperforms OpenAI's GPT-5 and Anthropic's Claude Sonnet 4.5.
It scored 44.9% on Humanity’s Last Exam and 60.2% on BrowseComp, showing strong reasoning and coding.
Kimi K2 Thinking can perform 200–300 sequential tool calls autonomously for wide applications.
Development costs are significantly lower than competitors, suggesting open models may disrupt the field.

1 / 2