Question 1

What is LongCat-2.0?

Accepted Answer

LongCat-2.0 is a 1.6-trillion-parameter open-source large language model from Meituan that uses a sparse mixture-of-experts architecture and offers a 1-million-token context window, designed for AI agent and coding tool applications, according to SiliconANGLE.

Question 2

Why does the domestic chip training matter?

Accepted Answer

Meituan says it trained LongCat-2.0 entirely on Chinese-made ASIC clusters, demonstrating that frontier-scale AI models can be built without Nvidia GPUs, which have been subject to U.S. export restrictions targeting China.

Question 3

How does LongCat-2.0 compare to other AI models?

Accepted Answer

Meituan's published benchmarks place it alongside closed-source models from Google, OpenAI and Anthropic, though independent verification is pending. Its 1-million-token context window exceeds the 128,000-token standard used by models like DeepSeek-R1-0528 and OpenAI's GPT-OSS, per SiliconANGLE.

Meituan

Meituan open-sources 1.6-trillion-parameter LongCat-2.0 AI model trained on Chinese chips

One sharp brief.
Every day.

Meituan

Meituan open-sources 1.6-trillion-parameter LongCat-2.0 AI model trained on Chinese chips

One sharp brief.Every day.

One sharp brief.
Every day.