whatmodel.sh / LongCat-Flash-Chat (560B-A27B)

LongCat-Flash-Chat (560B-A27B)

+ contribute

LongCat-Flash-Chat (560B-A27B) is an open-weights language model from Meituan, released August 2025. Its published size is 560B parameters (27B active per token) and a 128k-token context window. Licence: mit.

Catalogue identity notes

Meituan LongCat team, meituan-longcat/LongCat-Flash-Chat @f0b8e8d (created 2025-08-29, MIT). Comparison-only for now: weights are public (about 1.12 TB BF16, 75 safetensors files) but the card points to an external SGLang/vLLM deployment guide rather than giving a command, so no run option is recorded. Its own columns sit in the card runs that already hold six catalogued comparators. Arena entry 'longcat-flash-chat' (MIT). Non-thinking chat model. Not LongCat 2.0 or the 2602-exp chat preview (separate entries).

Model information sources

Release date ↗ · Weights licence ↗

Card: '560 billion total parameters ... activates 18.6B∼31.3B parameters (averaging∼27B) based on contextual demands'; safetensors metadata 561,862,880,256.

65.5±1.0#138 of 719 LLMsfrom 23 benchmark results across 2 sources

$ how it compares

The LLMs ranked either side of LongCat-Flash-Chat (560B-A27B), out of 828 with a score, then more from Meituan.

#	model	score	where	RAM	released
155	Nemotron Cascade 2 30B-A3B NVIDIA	66±1	local	~20GB	2026-03
156	Laguna XS 2.1 (33B-A3B) Poolside	66±2	local	~21GB	2026-06
157	LongCat-Flash-Chat (560B-A27B) Meituan	66±1	local	~333GB	2025-08
158	o3-mini OpenAI	65±1	api $1.93/M		2025-01
159	Claude 3.7 Sonnet Anthropic	65±1	comparison only		2025-02
	more from Meituan
63	LongCat 2.0 (1.6T-A48B) Meituan	73±2	api $1.30/M +local	~2155GB	2026-06
114	LongCat-Flash-Thinking (560B-A27B) Meituan	69±1	comparison only		2025-09

see all LLMs · find the ones that fit your machine

$ what it's good at

agent tasks▁▁▃▆█▆▅▅▅▄▃▁ #102 of 300 · 21% of its score · 5 tests

Beats Grok 4.20 (0309 reasoning), Grok 4.3, o3-pro here, though they rank higher overall. Trails SmolLM3 3B, Nova 2 Pro (Preview) here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
SWE-bench Verified lab	60.4% ✎	▁▁▁▁▁▁▂▄█▆▁▁ beats 55/131	hard · avg 18%	▇▇▇ 7.1%	sharp · at its level
τ²-bench Retail lab	71.3% ✎	▅···█·▃····▃ beats 4/6	medium · avg 41%	▇▇ 4.7%	at its level
τ²-bench Telecom lab	73.7% ✎	██··███···██ beats 6/6	very hard · avg 9.6%	▇▇ 4.3%	sharp · at its level · noisy
τ²-bench Airline lab	58.0% ✎	▄·▄··▄···█▄▄ beats 5/6	hard · avg 26%	▇ 3.0%	at its level
Terminal-Bench (pre-2.0 release) lab	39.5% ✎	▄·▄···▄▄···█ beats 4/5	very hard · avg 12%	▇ 2.1%	at its level

coding▁▁▁▃▆██▇▆▄▃▁ #159 of 458 · 22% of its score · 4 tests

Beats Gemini 2.5 Pro, Qwen3-Coder-Next (80B-A3B), Grok 4.1 Fast (Reasoning) here, though they rank higher overall. Trails QwQ-32B-Preview, Ministral 3 8B Reasoning 2512, Phi-4-reasoning 14B here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
SWE-bench Verified lab	60.4% ✎	▁▁▁▁▁▁▂▄█▆▁▁ beats 55/131	hard · avg 18%	▇▇▇ 7.1%	sharp · at its level
LiveCodeBench (version unspecified) lab	48.0% ✎	█·······▄▄▄▄ beats 5/5	hard · avg 21%	▇▇▇ 6.7%	at its level
MBPP+ lab	79.6% ✎	▃·····▃█▃··▃ beats 2/6	medium · avg 61%	▇▇ 4.3%	typical
HumanEval+ lab	88.4% ✎	▄▄▄·····▄▄·█ beats 2/6	easy · avg 67%	▇▇ 4.0%	sharp · maxed out

chat & writing▁▁▁▃▅▄▅█▆▆▇▃ #107 of 393 * · 11% of its score · 2 tests

Beats LongCat 2.0 (1.6T-A48B), GPT-5.4 nano, Step 3.5 Flash (196B-A11B) here, though they rank higher overall. Trails Gemini 2.5 Flash, Mistral Large 3 (675B-A41B), GPT-4.1 here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
LMArena (text)	1401.2 ✎	▁▁▁▂▃▃▃▆▆▇█▃ beats 245/379	avg 1268	▇▇▇ 8.2%	blunt
IFEval (aggregation unspecified) lab	89.6% ✎	█··█·█··████ beats 6/6	easy · avg 75%	▇ 3.3%	typical

math▁▂▃▅▆▅▄█▄▁▁▁ #129 of 398 · 7.7% of its score · 3 tests

Beats Claude Sonnet 4.6, MiniMax-M3, MiMo-V2.5 (310B-A15B) here, though they rank higher overall. Trails Olmo 3.1 32B Think, Qwen3-VL 30B-A3B Thinking, Granite 4.2 8B here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
AIME 2025 lab	61.3% ✎	▄▄···▄▄··▄·█ beats 4/6	hard · avg 19%	▇▇ 3.8%	sharp · at its level · noisy
AIME 2024 lab	70.4% ✎	▁▁▁▂▁▃▁▅███▃ beats 38/85	hard · avg 32%	▇ 2.1%	sharp · noisy
MATH-500 lab	96.4% ✎	▄·▄··█··▄·▄▄ beats 3/6	very easy · avg 86%	▇ 1.7%	maxed out · noisy

other tests10 not in a task · 44% of its score

test	result	vs the others in its run	difficulty	counts	why
MMLU lab	89.7% ✎	▄·····█▄▄▄·▄ beats 2/6	medium · avg 71%	▇▇▇ 8.5%	typical
GPQA Diamond lab	73.2% ✎	█·█·█··██··█ beats 2/5	hard · avg 42%	▇▇▇ 8.1%	sharp · at its level
MMLU-Pro lab	82.7% ✎	█▄·▄···▄··▄▄ beats 3/6	medium · avg 59%	▇▇▇ 6.9%	typical
CMMLU lab	84.3% ✎	█····▄··▄▄▄▄ beats 2/6	medium · avg 67%	▇▇ 4.3%	typical
C-Eval lab	90.4% ✎	█····▄·▄·▄▄▄ beats 4/6	medium · avg 68%	▇▇ 3.9%	typical
VitaBench lab	24.3% ✎	█·······▄█·█ beats 6/6	very hard · avg 6.4%	▇ 3.7%	typical
COLLIE (constrained text generation) lab	57.1% ✎	▄···▄█▄····█ beats 6/6	hard · avg 30%	▇ 3.4%	at its level
DROP lab	79.1% ✎	▄···▄·▄█·▄·▄ beats 4/6	medium · avg 41%	▇ 3.0%	noisy

2 more tests

test	result	vs the others in its run	difficulty	counts	why
ZebraLogic lab	89.3% ✎	█···█··███·█ beats 4/5	medium · avg 53%	▇ 1.4%	noisy
BeyondAIME lab	43.0% ✎	█····█·▄▄··▄ beats 4/6	hard · avg 25%	▇ 1.2%	at its level

$ where to run it

LM Studio · recommended

Install LM Studio · Open the app once before using the lms command.
download it (macOS / Windows / Linux) ↗
Run LongCat-Flash-Chat (560B-A27B) · needs ~315.4 GB, Mac only · from huggingface.co/mlx-community
open in LM Studio ↗
Or download it in the terminal
lms get https://huggingface.co/mlx-community/LongCat-Flash-Chat-4bit -y
Then run lms ls to see its name, and lms chat followed by that name

MLX (Mac)

Install MLX (Mac) · Apple silicon Macs only, text only.
with uv
uv tool install mlx-lm
or
pip install mlx-lm
Run LongCat-Flash-Chat (560B-A27B) · needs ~315.4 GB, Mac only · from huggingface.co/mlx-community
chat in the terminal
mlx_lm.chat --model mlx-community/LongCat-Flash-Chat-4bit
Or start an OpenAI-compatible server at http://localhost:8080/v1
mlx_lm.server --model mlx-community/LongCat-Flash-Chat-4bit

Weights and documentation: huggingface.co/meituan-longcat ↗

$ how the score was worked out · Meituan LongCat-Flash-Chat mo… 91%, LMArena leaderboard dataset 8%

Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works

source	results	share of the score	licence
Meituan LongCat-Flash-Chat model card (evaluation table) · lab-reported	22	▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 91%	MIT
LMArena leaderboard dataset	1	▇▇ 8%	CC BY 4.0
prior · pull toward the middle		▇ 0%
weighted average		= 65.5

$ improve this page

Update model information Add a place to run it Add a test result Propose a benchmark Correct a benchmark score Edit data on GitHub

Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.

data updated 2026-10-11 02:20 UTC · sources and licences · json