LongCat

LongCat-2.0 Release: What Developers Need to Know

A source-based overview of the LongCat-2.0 model release, covering what the publisher documents, verified technical details, and how developers can access the model through the documented API.

  • Published: 2026-07-19
  • Tags: LongCat-2.0, release, model-overview

Independent third-party resource. Not affiliated with or endorsed by LongCat or Meituan.

What changed

On 2026-06-30, the LongCat team published a blog post introducing LongCat-2.0, a large-scale mixture-of-experts language model. The model code and weights are available in the meituan-longcat GitHub organization.

The cited publisher blog describes LongCat-2.0 as a model with 1.6 trillion total parameters and approximately 48 billion parameters activated per token. The model weights are released under the MIT License. The GitHub README states that the model was pretrained on more than 35 trillion tokens and trained on hundreds of billions of tokens of 1M-context data.

According to the publisher documentation, the model is accessible through an API platform at api.longcat.chat, which supports both OpenAI-compatible and Anthropic-compatible endpoints. The publisher also documents a Claude Code integration that routes requests through the Anthropic-compatible endpoint.

What the publisher documents

The following details are taken from the cited publisher sources. They describe what the publisher states about the model, not independent findings of this site.

  • Architecture: Mixture-of-experts with LongCat Sparse Attention (LSA) and N-gram Embedding (135B parameters).
  • Context window: Up to 1,000,000 tokens, with a maximum output length of 128,000 tokens (per the publisher API documentation).
  • License: MIT License for model weights (per the LICENSE file in the GitHub repository).
  • API formats: OpenAI-compatible (chat completions) and Anthropic-compatible (messages), both served from https://api.longcat.chat.
  • Integration: The publisher documents configuration steps for Claude Code, OpenClaw, and Hermes.
  • Deployment: The GitHub README links to GPU deployment via SGLang and NPU deployment via SGLang-FluentLLM.

What developers can do next

Developers who want to explore LongCat-2.0 can take the following steps, each documented by the publisher:

  1. Review the model card and license on the GitHub repository to understand the model's documented capabilities and licensing terms.
  2. Create an API key through the LongCat API Platform, following the steps in the publisher quick start.
  3. Send a test request using the OpenAI-compatible or Anthropic-compatible endpoint, as documented in the publisher API overview.
  4. Configure Claude Code by following the publisher Claude Code configuration page or this site's independent setup summary.

This site does not issue, manage, or bill for LongCat API keys. The cited publisher documentation is the source of truth for creating and revoking keys.

Source notes and limitations

This page summarizes publicly available publisher documentation. The following limitations apply:

  • This site has not independently tested LongCat-2.0. All capability descriptions are sourced from the cited publisher documentation.
  • Benchmark scores published by the LongCat team are vendor-reported and have not been independently verified by this site.
  • API pricing is documented on a separate publisher page (not reproduced here, as pricing may change). Review the publisher pricing page for current rates.
  • This page is not a news article, press release, or product announcement. It is an independent third-party resource that summarizes publicly available primary documentation.

Related pages

Sources

  • LongCat-2.0 blog post

    Publisher documentationPublished 2026-06-30Accessed 2026-07-19

    Publisher blog announcing LongCat-2.0. Contains model architecture, training details, benchmark scores, and deployment information.

  • LongCat-2.0 GitHub Repository

    Primary sourceAccessed 2026-07-19

    Primary repository under the meituan-longcat organization. Contains README with model card, LICENSE file (MIT), chat template, and deployment instructions.

  • LongCat API Quick Start

    Publisher documentationAccessed 2026-07-19

    Publisher documentation for API key creation, supported endpoints, rate limits, and integration examples.

  • LongCat API Overview

    Publisher documentationAccessed 2026-07-19

    Publisher documentation for the base URL, authentication, endpoints, error responses, and SDK compatibility.

Independent third-party disclosure

This page is published by an independent third-party site. It is not affiliated with, endorsed by, sponsored by, or operated by Meituan, LongCat, or any of their affiliates. The content summarizes publicly-available primary documentation and does not represent the views of any referenced organization.

Read the disclaimerAbout this site

Last reviewed: 2026-07-19