What is LongCat-2.0?
Meituan's open-weight 1.6T-parameter MoE model with a 1M-token context for coding and agent work.
Who is LongCat-2.0 best for?
LongCat-2.0 is best for developers who want a very large open model with a 1M-token context for long coding and agent tasks.
Long coding sessions
The 1M-token context and long-context training let it read a large codebase and its docs together.
Agent harnesses
Meituan tuned it for Claude Code, OpenClaw and Hermes, and its API speaks both OpenAI and Anthropic formats.
Teams that self-host
The MIT-licensed weights are on Hugging Face and GitHub, with SGLang guides for GPU and NPU serving.
Developers paying by usage
The API has no fixed plan or minimum; you top up a balance and pay only for tokens used.
How much does LongCat-2.0 cost?
LongCat-2.0's weights are free under the MIT license on Hugging Face and GitHub, and Meituan's LongCat API charges pay-as-you-go per million tokens from a prepaid balance in Chinese yuan.
- Download from Hugging Face or GitHub
- Serve it yourself with SGLang
The LongCat API is pay-as-you-go: you top up a prepaid balance with Alipay or WeChat Pay and pay per million uncached input, cached input and output tokens. Its pricing page lists only a limited-time discounted rate, so check that page for the going price. Token Packs sell fixed token quotas that last 30 days. LongCat-2.0 on the API takes up to 1M tokens of context and 128K tokens of output. Full pricing on longcat.chat. We checked these prices on .
How do readers rate LongCat-2.0?
No one has rated LongCat-2.0 yet. Be the first.
Sign in to rate. One rating per account.
Is LongCat-2.0 open source?
Yes. LongCat-2.0's code is public under the MIT licence, so you can run it on your own server.
What are the best LongCat-2.0 alternatives?
LongCat-2.0 is an open source alternative to GPT-5.6 and Claude Sonnet 5. Command A+ and GLM-4.7 do the same job and are open source too.
OpenAI's GPT-5.6 family in Sol, Terra and Luna tiers for professional work.
Anthropic's Sonnet 5 with a 1M context, at $2 and $10 per million tokens.
Cohere's open multimodal reasoning model for enterprise agents.
Z.ai's MIT-licensed model for agentic coding with a 200K context.
Questions about LongCat-2.0
What is LongCat-2.0?
Meituan's open-weight 1.6T-parameter MoE model with a 1M-token context for coding and agent work.
Who is LongCat-2.0 for?
LongCat-2.0 is best for developers who want a very large open model with a 1M-token context for long coding and agent tasks.
Is LongCat-2.0 free?
Yes. LongCat-2.0 has a free Open weights plan: download from Hugging Face or GitHub.
Is LongCat-2.0 open source?
Yes. LongCat-2.0's code is public under the MIT licence, so you can run it on your own server.
What are the best LongCat-2.0 alternatives?
The closest alternatives on Recaply are GPT-5.6, Claude Sonnet 5, Command A+ and GLM-4.7.
Where can I try LongCat-2.0?
On its official website, longcat.chat.
Can I use LongCat-2.0 with Claude Code?
Yes. LongCat's API accepts Anthropic-format requests at its /v1/messages endpoint as well as OpenAI-format chat completions, and Meituan says LongCat-2.0 was built to work with harnesses such as Claude Code, OpenClaw and Hermes.







