Back to Models
Meituan: LongCat 2.0
Meituantext8.8 / 10 Overall Rating
LongCat 2.0 is a sparse Mixture-of-Experts language model from Meituan utilizing 48B active parameters out of 1.6T total. Designed for repository-level software development and long-horizon agentic workflows, it features a 1,048,576 token context window and a 262,144 token max output. At $0.30/$1.20 per million tokens, it offers low-cost long-context processing relative to major benchmarks.
Context Window
1049K
Knowledge Cutoff
2025
Max Output
16K
License
Proprietary
Where it Excels (Pros)
- 1M token input context window
- 262K max output token capacity
- Significantly cheaper than top baselines
Limitations (Cons)
- Unverified performance on Western benchmarks
- Potential latency during peak routing
Benchmark Breakdown
Reasoning (MMLU)82%
Coding (HumanEval)78%
Mathematics (MATH)76%
Long-Context Retrieval95%
Pricing Matrix
Input cost / 1M tokens$0.30
Output cost / 1M tokens$1.20
Compare Meituan: LongCat 2.0 with other frontier models: