Meituan sources
5 official pages from Meituan support 1 model record in the AI model benchmarks. Official sources come from the model providers themselves: they support each model record as a whole, and no benchmark score comes from them.
Each record names the date it was last verified, its dataset version and license where Meituan states one, what it covers and how the data uses it. Missing metadata is not filled or inferred. Every publisher is listed in the source catalog.
Meituan 5
- LongCat API Platform change log
Covers identity · Verified 2026-11-03 · Not versioned · License not stated
Version 2026-11-02 makes LongCat-2.5 available on the LongCat API Platform: native multimodality with agentic coding, long-horizon agents for complex codebases, and compatibility with Claude Code, Codex and other agent harnesses. LongCat-2.5-Preview had been available since 2026-09-25.
- LongCat API model parameters
Covers identity · Verified 2026-11-03 · Not versioned · License not stated
LongCat-2.5: a 1,048,576-token context window counting input and output, max_tokens up to 262,144, and a thinking budget of up to 262,144 tokens.
- LongCat API visual understanding
Covers identity · Verified 2026-11-03 · Not versioned · License not stated
LongCat-2.5 accepts images (up to 100 a request, 10 MB each) and videos by URL (up to 10 a request, 100 MB and under an hour each) alongside text.
- LongCat-2.5 pricing
Covers pricing · Verified 2026-11-03 · Not versioned · License not stated
Pay-as-you-go LongCat-2.5 at a discounted, limited-time price: $0.30 uncached input, $0.006 cached input and $1.20 output per 1M tokens. The page gives no list price and no end date.
- meituan-longcat/LongCat-2.5 model card
Covers identity, benchmark · Verified 2026-11-10 · Not versioned · License MIT
Weights published 2026-11-04 under the MIT License: a native multimodal mixture-of-experts model with 1.6 trillion total parameters and about 48 billion activated per token, with W4A4 quantized experts. Its table marks scores cited from public reports with an asterisk; unmarked scores are Meituan in-house runs it calls indicative only.