LongCat-2.5 model released with native multimodality for agentic tasks
Original title: LongCat-2.5: Native Multimodality for Agentic Tasks
LongCat-2.5 features 1.6 trillion parameters and native visual understanding for long-horizon agentic tasks.
LongCat-2.5 is a new multimodal model with 1.6 trillion total parameters and 48 billion activated per token. It builds on LongCat-2.0 with stronger visual understanding and agentic coding via long-horizon RL data synthesis and visual tokenization. Showcases include software development and professional office scenarios.
- 1.6 trillion parameters
- 48 billion activated per token
- Native visual tokenization via RVQ
- Strong agentic RL pipeline
- End-to-end demos in coding and office tasks
- AI models
- LongCat-2.5
From the original article
12 min read at longcat.aiOverview Metrics Showcases Visual Modeling Agentic RL MOPD Long-Context Training Supernode Parallelism FSDP & Systems RL Infrastructure Consistency Rollout W4A4 Evaluations We introduce LongCat-2. 5, a native multimodal model for long-horizon agentic tasks. With 1.
By pretext · longcat.ai
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.