name = "LongCat-2.0"
description = "Meituan LongCat-2.0, a reasoning model with tool calling and a 1M-token context window"
family = "longcat"
attachment = false
reasoning = true
temperature = true
tool_call = true
release_date = "2026-06-30"
last_updated = "2026-06-30"
open_weights = false

[limit]
context = 1_000_000
output = 131_072

[modalities]
input = ["text"]
output = ["text"]

[[benchmarks]]
name = "SWE-Bench Pro"
score = 59.5
metric = "resolve rate"
source = "https://github.com/meituan-longcat/longcat-2.0"
date = "2026-06-30"

[[benchmarks]]
name = "SWE-Bench Multilingual"
score = 77.3
metric = "resolve rate"
source = "https://github.com/meituan-longcat/longcat-2.0"
date = "2026-06-30"

[[benchmarks]]
name = "Terminal-Bench"
score = 70.8
metric = "success rate"
version = "2.1"
source = "https://github.com/meituan-longcat/longcat-2.0"
date = "2026-06-30"

[[benchmarks]]
name = "GPQA Diamond"
score = 88.9
metric = "accuracy"
source = "https://github.com/meituan-longcat/longcat-2.0"
date = "2026-06-30"

[[benchmarks]]
name = "BrowseComp"
score = 79.9
metric = "accuracy"
source = "https://github.com/meituan-longcat/longcat-2.0"
date = "2026-06-30"

[[benchmarks]]
name = "IFEval"
score = 90.0
metric = "accuracy"
source = "https://github.com/meituan-longcat/longcat-2.0"
date = "2026-06-30"

[[benchmarks]]
name = "FORTE"
score = 73.2
metric = "success rate"
source = "https://github.com/meituan-longcat/longcat-2.0"
date = "2026-06-30"