Qwen3 VL 235B A22B Instruct vs Qwen3.5 397B A17B

Qwen3 VL 235B A22B Instruct
Wins 0 of 5 benchmarks
Qwen3.5 397B A17B
$0.47 / task
Wins 3 of 5 benchmarks
Different efforts: non-thinking · thinking
Humanity's Last Exam Humanity's Last Exam: Very hard expert questions across many fields, written to stump today’s models
Qwen3 VL 235B A22B Instruct6.6
Qwen3.5 397B A17B29
Terminal-Bench 4.0 Terminal-Bench 4.0 · source setting 2: Agents use a terminal to complete executable coding, system administration and research tasks in the 4.0 suite
Qwen3 VL 235B A22B Instruct—
Qwen3.5 397B A17B0
AutomationBench-AA AutomationBench-AA: Agents orchestrate SaaS workflows through APIs; objectives receive credit only when the task has no guardrail violation
Qwen3 VL 235B A22B Instruct—
Qwen3.5 397B A17B6.9
Arena Text · creative writing Arena Text · creative writing: Human pairwise preferences in the stated text vote pool
Qwen3 VL 235B A22B Instruct1,357
Qwen3.5 397B A17B1,406
MMMU-Pro (vision) MMMU-Pro (vision): College-level questions that require understanding images: charts, diagrams, photos
Qwen3 VL 235B A22B Instruct67.6
Qwen3.5 397B A17B77.3

Specs

SpecQwen3 VL 235B A22B InstructQwen3.5 397B A17B
Input, per 1M$0.20$0.60
Output, per 1M$0.88$3.60
Cost per task?—$0.47
Speed, tok/s3855
Context262K262K
WeightsOpenOpen
ReleasedSep 2025Feb 2026
Memory (4-bit)141 GB242 GB
LicenceApache-2.0Apache-2.0
Commercial useAllowedAllowed
See images✓✓
Read PDFs——
Hear audio——
Watch video—✓
Call tools✓✓
Force a tool callPartialPartial
Return strict JSON✓✓
Search the web??
Run code??
Use a computer??

What breaks when you switch

Qwen3 VL 235B A22B Instruct → Qwen3.5 397B A17B
Memory (4-bit): 141 GB → 242 GB
Qwen3.5 397B A17B → Qwen3 VL 235B A22B Instruct
Nothing known.