TopicLongCat-2.5 Release
Meituan Open-Sources LongCat-2.5 Weights: 1.6T-Parameter Agent Model Trained Without Nvidia Chips
Meituan released the full LongCat-2.5 checkpoint on Hugging Face on Nov 15 under the MIT license, including for commercial use; the 1.6-trillion-parameter model activates about 48B per token with up to 1M tokens of context, one week after its official unveiling.

What happened
The model card for LongCat-2.5 now carries the full checkpoint on Hugging Face, letting developers download, inspect, modify, fine-tune, self-host and redistribute it under the MIT license. Built for long-horizon agent work, the model takes text, image and video input, outputs text, and lets users toggle reasoning on and off.
The release comes about a week after the model's official unveiling. The API supports both the OpenAI and Anthropic formats and plugs into agent tools such as Claude Code and Codex, with a chat interface at longcat.ai. A limited-time launch price is roughly €0.27 per million input tokens and €1.07 per million output tokens.
Key facts
- Scale: A native multimodal mixture-of-experts model with 1.6 trillion total parameters and about 48B activated per token; the API supports up to 1M tokens of context and generations up to 128,000 tokens
- Trained on domestic chips: Meituan says the predecessor LongCat-2.0 was trained in late June on a cluster of more than 50,000 domestic AI chips; the sample LongCat-2.5 deployment runs on 194 Huawei Ascend Atlas A2 accelerators — 2 for image processing, 64 for prompt ingestion and 128 for response generation — with no deployment guidance for Nvidia GPUs
- Terminal-task scores: In Meituan's own evaluation, Terminal Bench v2.1 scores 88.4, against 89.1 for Claude Opus 5, 88.8 for GPT-5.6 Sol and 90.6 for DeepSeek V4.1-Flash; on the harder Terminal Bench v4.0 it scores 38.5, above GPT-5.6 Sol's 37.3 and below Opus 5's 53.9; DeepSWE rises from 9.7 for the predecessor to 64.6
- Quantized deployment: Expert weights and activations use 4-bit quantization, which domestic chips compute natively; against the 8-bit version, online stress tests show 43.3% higher prompt-processing throughput and 42.77% higher decode throughput, with accuracy on par with the unquantized model
Why it matters
a 1.6-trillion-parameter agent model opens its full weights under the MIT license, trained and deployed entirely without Nvidia chips, with a Terminal Bench score of 88.4, close to Claude Opus 5's 89.1.
Follow HIMIAO Horizon
No sign-up · Unsubscribe any timeNotifications
A ping on your phone or computer when the brief is out. No email needed.
Comments
Related
- Nov 2Meituan Launches LongCat-2.5: 1.6T-Parameter Native Multimodal Model for Long-Horizon Agent Tasks
- Nov 15SemiAnalysis Names Nvidia's Seven Biggest GPU Customers; Six Build Their Own Chips
- Nov 10Anthropic Sets IPO Price at $70–$80, Seeking Up to $69.3 Billion
- Nov 7NVIDIA Releases Four Cosmos 3 Open Weights: 16B World Models for Driving Data and Robot Simulation
- Oct 22DeepSeek Open-Sources DeepInfer, the Inference Engine Behind Its Production Services
Today
Nov 17 Tuesday- Brief
- GitHubGitHub Weekly Vol.6: 7 Projects Worth a Look This Week
Tested or read — weeklyAll →
- Quiz
- Call
Latest news
All →- Nov 17OpenAI Launches Science Blog as GPT-6 Astra Leads Proof of Candidate Lattice Standard Model Definition
- Nov 17Apple Smart Glasses Said to Debut at WWDC 2027 With No Display, Leaning on Siri and iPhone
- Nov 17AI Labs Race to Hire Counterintelligence Staff: Anthropic 33 This Year, OpenAI 22
- Nov 17Trump Backs Huang's Return to Nvidia: 'He Learned a Big Lesson'
- Nov 16Kuo: OpenAI and Anthropic to Buy About 26M Google TPUs in 2028–29
- Nov 16Huawei Cloud Launches Ascend-Native Embodied World Model CloudRobo-WM1.2
™