Qwen3 4B Thinking 2507

Qwen·Pending·Released 2025-08-06text
Context262.144k
Max output-
Input / 1M-
Output / 1M-
Cached in-
Benchmarks15

Benchmarks

Cited public sources · rank is exact-benchmark, apples-to-apples

AIME25#6/1281.3% ↗
Arena-Hard v2#9/934.9% ↗
BFCL-v3#4/1071.2% ↗
GPQA#7/1265.8% ↗
HMMT25#6/1055.5% ↗
IFEval#18/2487.4% ↗
LiveBench 20241125#5/771.8% ↗
LiveCodeBench v6 (25.02-25.05)#5/855.2% ↗
MMLU-Pro#42/4974% ↗
MMLU-Redux#12/1386.1% ↗
OJBench#4/417.9% ↗
SuperGPQA#16/1747.8% ↗
TAU2-Airline#2/558% ↗
TAU2-Retail#7/853.5% ↗
TAU2-Telecom#2/327.2% ↗

Qwen3 4B Thinking 2507 FAQ

What is Qwen3 4B Thinking 2507?
Qwen3 4B Thinking 2507 is a large language model from Qwen, released on August 6, 2025. It accepts text input.
When was Qwen3 4B Thinking 2507 released?
Qwen released Qwen3 4B Thinking 2507 on August 6, 2025.
What is Qwen3 4B Thinking 2507's context window?
Qwen3 4B Thinking 2507 has a 262K-token context window.
How does Qwen3 4B Thinking 2507 score on benchmarks?
Qwen3 4B Thinking 2507 has 15 cited benchmark scores, including AIME25 81.3%, Arena-Hard v2 34.9% and BFCL-v3 71.2%. Each score links to its source.

Access Qwen3 4B Thinking 2507 and every other model through one endpoint with automatic failover: Respan gateway.