whatmodel.sh / Qwen3-Max (2025-09-23)

Qwen3-Max (2025-09-23)

+ contribute

Qwen3-Max (2025-09-23) is a closed-weights language model from Alibaba, released September 2025. Licence: proprietary.

Catalogue identity notes

Comparison model from the Epoch AI internal-evaluation CSVs (Epoch release date 2025-09-24). Dated API snapshot qwen3-max-2025-09-23 (Epoch display 'Qwen3-Max-Instruct'); distinct from qwen3-max-2026-01-23 and qwen3-max-preview. Listed as a qwen3-max snapshot on Model Studio (observed 2026-09-28).

65.9±1.0#132 of 719 LLMsfrom 12 benchmark results across 4 sources

$ how it compares

The LLMs ranked either side of Qwen3-Max (2025-09-23), out of 828 with a score, then more from Alibaba.

#	model	score	where	RAM	released
149	o1 (2024-12-17) OpenAI	66±1	api $26.25/M		2024-12
150	MiniMax-M2.5 MiniMax	66±1	api $0.53/M +local	~75GB	2026-02
151	Qwen3-Max (2025-09-23) Alibaba	66±1	api $2.40/M		2025-09
152	DeepSeek V3.1 DeepSeek	66±1	api $0.42/M +local	~110GB	2025-08
153	Spark-X2.5 4B iFLYTEK (SparkLLM)	66±1	local	~3.9GB	2026-08
	more from Alibaba
18	Qwen3.8 Max (2.4T-A95B) Alibaba	78±1	api $3.00/M +local	~380GB	2026-08
34	Qwen3.8 Max (0902) Alibaba	76±1	api $3.00/M		2026-09
37	Qwen3.8-Flash-Next Alibaba	75±1	local	~59GB	2026-08

see all LLMs · find the ones that fit your machine

$ what it's good at

agent tasks▁▁▃▆█▆▅▅▅▄▃▁ #114 of 300 · 13% of its score · 3 tests

Beats o3-pro, o3, GPT-5 mini here, though they rank higher overall. Trails SmolLM3 3B, Jamba 1.5 Large, Nova 2 Pro (Preview) here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
τ²-bench Telecom lab	84.2% ✎	▁·▅█▆▅▂▃▃▂▃▁ beats 62/89	hard · avg 34%	▇▇▇ 6.4%	sharp · small run
τ²-bench Retail lab	72.2% ✎	▁····▂▁▃▆█▃▁ beats 33/44	medium · avg 39%	▇ 3.6%	at its level · small run
τ²-bench Airline lab	59.5% ✎	▁▁·▂▁▃▂▁█▅▂▃ beats 30/42	hard · avg 30%	▇ 2.8%	at its level · small run

chat & writing▁▁▁▃▅▄▅█▆▆▇▃ #86 of 393 * · 9.4% of its score · 1 test

Beats LongCat 2.0 (1.6T-A48B), Inkling Small (266B-A12B), Qwen3.5 122B-A10B here, though they rank higher overall. Trails Gemma 4 12B, Gemma 4 26B-A4B, Mistral Medium 3.5 (128B) here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
LMArena (text)	1422.9 ✎	▁▁▁▂▃▃▃▆▆▇█▃ beats 274/379	avg 1268	▇▇▇▇ 9.4%	blunt

math▁▂▃▅▆▅▄█▄▁▁▁ #114 of 398 · 47% 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, Gemini 2.5 Flash, Qwen3-VL 32B Thinking here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
FrontierMath tiers 1–3 v2	18.9% ✎	▂··▁▁▃▄▄█▅▃▃ beats 12/82	very hard · avg 1.2%	▇▇▇▇▇▇▇▇▇▇▇▇▇ 32%	sharp
OTIS mock AIME 2024–2025	73.3% ✎	▃▃▄▃▃▃▇▇█▃▂▅ beats 107/193	very hard · avg 8.5%	▇▇▇ 7.6%	sharp · at its level · noisy
MATH level 5	97.1% ✎	▂▂▅▆▄█▆▃▅▂▃▄ beats 87/92	hard · avg 32%	▇▇▇ 7.2%	sharp

other tests5 not in a task · 30% of its score

test	result	vs the others in its run	difficulty	counts	why
GPQA Diamond	72.6% ✎	▁▁▁▁▂▄▆▄▅▇█▅ beats 110/208	hard · avg 40%	▇▇▇▇▇▇ 15%	sharp · at its level
KINA (Knowledge Index of Noah's Ark) accuracy, 899 ten-option items	35.9% ✎	▁▁▂▁▃▄█▆▄▅▂▃ beats 29/39	very hard · avg 22%	▇▇ 6.2%	blunt · at its level
Chess Puzzles (Epoch)	4.0% ✎	█▁▁▁▁▂▁▄▂▂▁▁ beats 1/135	very hard · avg 5.9%	▇ 3.2%	sharp · noisy
Mystery Game Puzzles (Epoch)	5.0% ✎	█··▂▃▇▄█▂▃▁▁ beats 1/74	very hard · avg 9.4%	▇ 3.1%	sharp · too hard for it · noisy
SimpleQA Verified	48.7% ✎	▁▁▃▂▃▃▆▆█▃▂▅ beats 51/78	very hard · avg 8.0%	▇ 3.0%	at its level

$ where to run it

Through an API

api · Alibaba Cloud Model Studio · $2.40/M · $1.2 in / $6 out per 1M tokens · 262k context · www.alibabacloud.com ↗ · price sourceUSD per 1M tokens, Singapore (International), dated snapshot qwen3-max-2025-09-23 (non-thinking mode only), 0<Token<=32K input tier (32K-128K and 128K-256K tiers are higher): input $1.2, output $6. Batch and context-cache discounts, higher tiers and free quota excluded. Observed 2026-09-28.

$ how the score was worked out · Epoch AI — internal evaluatio… 71%, τ²-bench leaderboard: Qwen (A… 13%, LMArena leaderboard dataset 9% and 1 more

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
Epoch AI — internal evaluations	7	▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 71%	CC BY 4.0
τ²-bench leaderboard: Qwen (Alibaba) own submissions · lab-reported	3	▇▇▇▇ 13%	MIT
LMArena leaderboard dataset	1	▇▇▇ 9%	CC BY 4.0
Knowledge Index of Noah's Ark (arXiv:2606.05104v2)	1	▇▇ 6%	CC BY 4.0
prior · pull toward the middle		▇ 0%
weighted average		= 65.9