LongCat 2.0

LongCat 2.0 is a text chat model released under the LongCat name. It sits in our catalog as a conversational, instruction-following model intended for the same general workloads as other chat LLMs: question answering, drafting, summarization, and code assistance. Our database entry for LongCat 2.0 is deliberately narrow — we list the creator, the model type, and independently measured performance figures, and we do not currently track a confirmed context window, modality list, or tool-calling specification for it. The performance data we carry comes from Artificial Analysis, which measured LongCat 2.0 at roughly 69.44 output tokens per second with a time to first token of about 1,760 milliseconds. That combination describes a model that streams at a moderate pace with a noticeable initial delay before the first token appears — relevant if you are building interactive chat interfaces where perceived responsiveness matters, and less relevant for batch or background jobs where total completion time dominates. Serving speed also varies by provider and by the hardware and routing behind a given endpoint, so treat these numbers as a reference point rather than a guarantee for any particular API. Because published details on LongCat 2.0 are limited in our records, the most practical way to evaluate it is empirically: send your own prompts, compare output quality against models you already use, and check the pricing table on this page to see which providers currently serve it and at what rate. Readers who need a documented context window, confirmed vision input, or a formal function-calling schema should verify those directly with the serving provider before committing to the model.