Model Detail
LongCat-Next
—LongCat-Next is a code generation model with 37.1B parameters released by meituan-longcat. The model is registered under the any-to-any pipeline tag on Hugging Face, distributed under the permissive mit license.
LongCat-Next ships with 37.1B parameters. Total weight footprint is approximately 74.3 GB, which is the relevant figure when planning local-inference VRAM. The mit license is permissive, allowing commercial deployment and derivative work without per-seat fees, though attribution requirements still apply.
LongCat-Next is best fit for code completion, repository-scale Q&A, and pair-programming integrations. It is a less obvious choice for one-shot generation of security-critical code without review. Treat this as a starting matrix rather than a benchmark verdict — the right deployment usually depends on the specific evaluation suite that mirrors your workload.
LongCat-Next: Lexicalizing Modalities as Discrete Tokens
arXiv:2603.27538v1 Announce Type: cross Abstract: The prevailing Next-Token Prediction (NTP) paradigm has driven the success of large language models through discrete autoregressive modeling. However, contemporary multimodal systems remain language-centric, often treating non-linguistic modalities a
LongCat-DeepResearch Technical Report
arXiv:2609.36071v1 Announce Type: new Abstract: We present LongCat-DeepResearch, a deep research system that combines an enhanced LongCat model with a multi-agent workflow for producing comprehensive, evidence-grounded reports. The workflow separates global planning from detailed investigation and c
