Artificial Analysis lists Qwen3.5 4B benchmarks
Artificial Analysis has published a benchmark page for Qwen3.5 4B, a two-model release whose non-reasoning variant returns its first answer token in 1.42 seconds, though the page excerpt includes no intelligence scores.
Independent benchmarking firm Artificial Analysis has added a release page for Qwen3.5 4B. The page covers intelligence, performance and price for the release.
According to the page, the release includes two models, each with its own intelligence, performance and pricing profile. Only one is named in the material: Qwen3.5 4B (Non-reasoning), which has the lower time to first answer token of the two at 1.42 seconds.
Artificial Analysis says it compares models against others in the same class. Non-reasoning models are compared only with other non-reasoning models. Reasoning models are compared with both reasoning and non-reasoning models. Open-weights models are grouped by size, with "Tiny" covering models of 4B parameters or fewer and "Small" covering 4B to 40B. Proprietary models are grouped by blended price per million tokens, using a 3:1 input-to-output ratio.
The page's capability indexes draw on several of the firm's evaluations. These include AA-Omniscience v1.1, AA-Briefcase v1.1, HLE-Diamond, AutomationBench-AA, GDP.pdf, MLCR-AA, CritPt v2 (Hard Subset), Terminal-Bench 4.0, Terminal-Bench-Science 0.1 and FrontierCode-AA 2.0.
The excerpt does not include the models' Intelligence Index scores, cost per task, throughput, pricing or licensing. It also does not name the second model or give a release date. It is not yet possible to say how Qwen3.5 4B ranks against other small models.