meituan
Meituan: LongCat 2.0
meituan/longcat-2.0
Our take
AI-assisted analysis based on the sources below. Suitability is an editorial assessment; test it on your own workload.
Meituan's LongCat 2.0 is a sparse mixture-of-experts text model with 1.6 trillion total parameters and roughly 48 billion active per token, according to its model card. The card reports training on more than 35 trillion tokens, a custom sparse-attention scheme, and about one million tokens of context, which the OpenRouter catalog also lists. Meituan positions it for coding, repository-level edits, and agentic workflows, and publishes mostly in-house benchmark comparisons against leading proprietary models. Weights are released under the MIT license, with GPU and NPU deployment paths documented.[1][2]
Strengths
- Very long context: the model card describes training on 1M-token data, and the catalog lists a context window of about one million tokens.[1][2]
- Vendor-reported strength on coding and agentic tasks, with published (largely self-measured) benchmark results and stated integration with tools such as Claude Code, OpenClaw, and Hermes.[2]
- Open weights under the MIT license, function/tool calling support, and an optional thinking mode for token efficiency.[1][2]
Limitations
- Text-only input and output; knowledge cutoff is not stated in the available sources.[1]
- Most evaluation figures come from Meituan's own harness, with external comparison scores partly cited from other vendors' reports, so independent verification is limited.[2]
- Meituan notes the model was not comprehensively evaluated for every downstream use and advises assessing accuracy, safety, and fairness before sensitive deployments; real workload validation is still required.[2]
Where to use it
- A candidate for agentic coding assistants and repository-level editing, matching the vendor's stated focus and tool integrations; worth piloting before adoption.[2]
- Plausible fit for long-document or long-context analysis given the roughly 1M-token window, though real-world retrieval quality at that length is untested in these sources.[1][2]
- Self-hosted deployments where MIT-licensed weights and documented GPU/NPU serving paths matter.[2]
Sources
- [1] OpenRouter catalog · Meituan: LongCat 2.0Sources checked: Nov 13, 2026
- [2] Model documentation · Meituan: LongCat 2.0Sources checked: Nov 13, 2026
About this model
Description below is the original text supplied by OpenRouter, shown in the source language.
LongCat 2.0 is a sparse mixture-of-experts language model from Meituan, with 48B active parameters out of 1.6T total. It is suited for coding, repository-level changes, long-horizon problem solving, and agentic...
Specifications
- Model ID
- meituan/longcat-2.0
- Canonical ID
- meituan/longcat-2.0-20260720
- Creator
- meituan
- Added to OpenRouter
- 20.07.2026
- Context length
- 1,048,756
- Modality
- text->text
- Input modalities
- text
- Output modalities
- text
- Token pricing
- USD per 1M tokens: in $0.300 / out $1.2
- Knowledge cutoff
- —
- Expiration date
- —
Tasks and capabilities
A grade appears only where the source supplies a matching measurement for that task, the model is active and the cohort holds at least 10 comparable models. The capability column lists facts the source declares — supported formats, reasoning and tool parameters — which are not measured quality.
| Task | Grade | Source-declared capability |
|---|---|---|
| General quality | No measurement | reasoning: optional · parameters: include_reasoning, reasoning |
| Coding | BB — next 30% of the cohort | — |
| Agents | BB — next 30% of the cohort | parameters: tool_choice, tools |
| Writing | No measurement | output: text |
| Reasoning | No measurement | reasoning: optional · parameters: include_reasoning, reasoning |
| Document analysis | No measurement | — |
| Vision | No measurement | — |
| Image generation | No measurement | — |
| Audio | No measurement | — |
| Video | No measurement | — |
| Embeddings | No measurement | — |
Evaluations
| Source | Benchmark | Value | Grade | Data retrieved |
|---|---|---|---|---|
| Artificial Analysis | Coding index | 45.3 | BB — next 30% of the cohort | 13.11.2026 |
| Artificial Analysis | Agentic index | 14 | BB — next 30% of the cohort | 13.11.2026 |
How grades work
Each letter grade compares a model only against other models scored by the same source and metric in the current snapshot. S means the top 10%, A the next 20%, B the next 30%, C the rest. Models without a published measurement are shown as not rated — a supported format alone never implies measured quality.
- S — S — top 10% of the cohort
- A — A — next 20% of the cohort
- B — B — next 30% of the cohort
- C — C — below the B band
Providers
Optional live provider statistics from OpenRouter. Missing values are unknown, not zero.
| Provider | Context length | Price in /1M | Price out /1M | Latency p50 (30m, s) | Throughput p50 (30m, tok/s) | Uptime (30m) |
|---|---|---|---|---|---|---|
| AtlasCloudfp8 | 1,048,756 | $0.300 | $1.2 | — | — | 100% |
Provider data retrieved: 13.11.2026
Related articles
No published articles identify this model yet.