Meituan: LongCat 2.0

LongCat 2.0 is a sparse mixture-of-experts language model from Meituan that activates 48 billion parameters out of a 1.6 trillion total, giving it very large world knowledge and tool-use capacity while keeping per-request compute closer to a mid-sized model. A published weights repository is hosted under the meituan-longcat organization on Hugging Face, making it one of the larger openly distributed MoE models of its generation. The design is aimed at coding, repository-level changes, long-horizon problem solving, and agentic workflows, where the model can hold large amounts of code, documentation, and tool state in context at once.

In deployment, LongCat 2.0 is served through an OpenRouter-hosted route with a roughly one-million-token context window, and it can be self-hosted via the SGLang serving stack. At the time of the SGLang documentation, support lived on the main branch and required a nightly wheel or rolling nightly Docker image until a tagged release picked it up, with CUDA 12.9 and CUDA 13 wheel indices available. The combination of a very large MoE with active-parameter efficiency, open weights, and a deep context window makes it a practical fit for code-assistant backends, multi-step agent pipelines, and any application that needs sustained reasoning over long, structured inputs.

Kilo Gatewaymeituan/longcat-2.0longcat

Open provider page Provider docs

Quick Info

Provider: Kilo Gateway
Model key: meituan/longcat-2.0
Release date: Jul 20, 2026
Last updated: Jul 20, 2026
Input modalities
Output modalities
Capabilities

Cost

Input token cost: $0.75
Output token cost: $3.00

Limits

Output tokens: 262,144 tokens
Context window: 1,048,756 tokens

Transparent token rates

Compare Meituan: LongCat 2.0 pricing

Rates are shown per one million tokens. Combined means one million input plus one million output tokens.

Browse this family

Latest news about Meituan: LongCat 2.0

Kilo Gateway

Coverage

Meituan's LongCat 2.0 — a 1.6 trillion-parameter mixture-of-experts language model with 48 billion active parameters — has been released as a fully open-source project under the MIT license, according to a third-party AI news roundup. The release package reportedly includes the model weights, model card, and deployment The model had previously appeared on OpenRouter under the codename "Owl Alpha," where it reportedly reached the third-highest call volume on the platform shortly after launch, according to the same roundup. The open-sourcing of LongCat 2.0 is framed by the aggregator as one of the more notable open-weight releases of t

Kilo Gateway

CoverageAnalysis

A July 3, 2026 analysis from The Tech Society frames Meituan's June 30, 2026 open-sourcing of LongCat-2.0 as a 1.6-trillion-parameter model trained and served entirely on domestic Chinese chips — reportedly 50,000+ accelerators, though the vendor was not confirmed. The piece reports vendor benchmarks of 59.5 on SWE-ben The analysis highlights Meituan's decade of logistics-driven machine-learning infrastructure as the foundation enabling a delivery company to build frontier-scale AI, and positions the release as a direct challenge to the premise of U.S. chip export controls. Pricing is cited at $0.30 per million input tokens versus GP

Kilo Gateway

Coverage

On June 29, 2026, Meituan officially unveiled LongCat-2.0 on GitHub, Hugging Face, and its native platform, revealing the 1.6-trillion-parameter Mixture-of-Experts model as the system behind the anonymous "Owl Alpha" that had been leading OpenRouter developer charts for two months. According to VentureBeat, the release Commercial access launched with a pricing structure in which all context-cache hits are processed free of charge, alongside a pay-as-you-go API listed at $0.75 input / $2.95 output per million tokens. A limited-time promotional discount cut uncached input to $0.30 and output to $1.20 per million tokens, placing LongCat