Ordalin
Submit a tool

Reviewed tool profile

LongCat-2.5

Meituan's multimodal coding model for long-horizon agents, usable in Claude Code and Codex.

Sources checked

Visit website ↗
Pricing
Usage-based API pricing at a limited-time discount: $0.30 per million uncached input tokens, $0.006 per million cached input tokens and $1.20 per million output tokens (¥2, ¥0.04 and ¥8 in China). You pay from a prepaid balance with no fixed plan, or buy 30-day Token Packs in limited daily flash sales. Official source 1 ↗ · Official source 2 ↗ · Official source 3 ↗
Available as
Cloud API on the LongCat API Platform, called with an API key over an OpenAI-compatible endpoint (api.longcat.chat/openai) or an Anthropic-compatible endpoint (api.longcat.chat/anthropic), with the official OpenAI and Anthropic SDKs. · Coding agents and chat clients configured to use the API, with setup guides for Claude Code, Codex, OpenCode, Cline, Kilo Code, Hermes Agent, OpenClaw, Cherry Studio and Chatbox. Official source 1 ↗ · Official source 2 ↗ Official source 1 ↗ · Official source 2 ↗
Primary group
Coding & Development

Also consider

See all 6 similar tools ↓

What it does

LongCat-2.5 is Meituan's LongCat team's multimodal model for agentic coding, made generally available on the LongCat API Platform on 2 November 2026 after a preview from 25 September. It combines image and video understanding with long-horizon coding work across whole codebases, has a 1M-token context window and up to 128K output tokens, and is served through OpenAI- and Anthropic-compatible endpoints so it can run inside Claude Code, Codex, Hermes and other agent harnesses. It is currently the only model on the platform's pay-as-you-go API.

Official source 1 ↗ · Official source 2 ↗

Features

  • Long-horizon agentic coding. LongCat-2.5 is built for multi-step engineering work: it reasons through complex codebases and full-project builds, calls tools and carries a task through to the end. Official source 1 ↗
  • Image and video understanding. The model takes images and videos with text, so it can describe an image, read text from a screenshot, analyze a chart or follow what happens in a video. Up to 50 images (10 MB each) and 10 videos (100 MB, under an hour each) fit in one request; videos must be passed as URLs. Official source 1 ↗
  • Visual coding and 3D tasks. Meituan pairs the visual understanding with agentic coding for visual coding, creative design and 3D tasks. Official source 1 ↗
  • 1M-token context. A 1M-token context window with up to 128K output tokens per response. Official source 1 ↗
  • Drop-in for Claude Code and other harnesses. Because the API speaks both the OpenAI and the Anthropic formats, you point Claude Code, Codex, Hermes and similar tools at the LongCat base URL and set LongCat-2.5 as the model. Official source 1 ↗ · Official source 2 ↗ · Official source 3 ↗

Pricing & free limits

Pay-As-You-Go

$0.30 input / $1.20 output per 1M tokensUsage-based, prepaid balance

Limited-time discounted price; cached input is $0.006 per 1M tokens. No fixed plan or minimum commitment, and the balance never expires. Official source 1 ↗ · Official source 2 ↗

Token Pack

Fixed token quota (price set per sale)One-time purchase, valid 30 days

Sold in limited daily flash sales; cache hits don't use the pack, and unused packs are refunded automatically after 30 days. Official source 1 ↗

Billing details

Each request is billed separately for uncached input, cached input and output tokens; failed requests are not billed. Official source 1 ↗

Top-ups can be paid by Alipay, WeChat Pay, credit or debit card and Google Pay, and unused prepaid balance can be refunded. Official source 1 ↗

The listed prices are a limited-time discount and may change; the platform's own price display is authoritative. Official source 1 ↗

Free access & limits

Free access and its limits were not established by the checked official sources.

Best for & limitations

Running Claude Code, Codex or OpenCode on a lower-cost model for day-to-day coding agents. Official source 1 ↗ · Official source 2 ↗

Building or refactoring whole projects where the agent has to plan, call tools and keep going over many steps. Official source 1 ↗

Turning screenshots, charts, design mocks or screen recordings into code or analysis. Official source 1 ↗ · Official source 2 ↗

  • Discount price may change. The published rates are labelled a limited-time discount, so the regular price is not shown yet. Official source 1 ↗
  • Verification for China users. Users in Mainland China must complete identity or enterprise verification before topping up or buying Token Packs. Official source 1 ↗
  • Per-key rate limits. Rate limits are enforced per API key and return HTTP 429 when exceeded; the docs don't publish the numbers. Official source 1 ↗
  • No open weights yet. Earlier LongCat models were released on Hugging Face, but LongCat-2.5 is offered through the API only. Official source 1 ↗
LongCat-2.5 product screenshot
Website previewFixed desktop viewport · 1440 × 900

Similar tools

Published tools that share its categories.

ColdteaControl layer for AI-written software: map every pull request, test the flows it touches with agents and watch it in production.
Coding & DevelopmentFree planChecked Aug 23
Visit
OpenLLMA routing layer that puts your AI subscriptions, API keys and local models behind one key for Claude Code, Codex and other agents, with fallback chains.
Coding & DevelopmentFree planChecked Oct 31
Visit