Compare / Qwen3 VL 235B A22B Thinking vs GPT-4.1
Qwen3 VL 235B A22B Thinking vs GPT-4.1
Which is cheaper, which is more capable, and which does better at each kind of work, from 16 results on 3 sources that measured both models.
Qwen3 VL 235B A22B Thinking or GPT-4.1? The short answer
- Qwen3 VL 235B A22B Thinking is cheaper: $1.30 against $3.50 per million tokens, 2.7 times less.
- They score about the same on the Artificial Analysis Intelligence Index (13.4 and 12.7).
- On our business benchmarks, Qwen3 VL 235B A22B Thinking ranks higher for reasoning and knowledge.
- On our business benchmarks, GPT-4.1 ranks higher for writing.
At a glance
| Qwen3 VL 235B A22B Thinking | GPT-4.1 | |
|---|---|---|
| Price per million tokens | $0.40 in · $4.00 out | $2.00 in · $8.00 out |
| Blended (3 in : 1 out) | $1.30 | $3.50 |
| Cheapest host, blended | $1.30 (Alibaba) | $3.50 (Azure) |
| Intelligence Index | 13.4 | 12.7 |
| Spring Prompt overall | – | – |
| Output speed, tokens a second | – | – |
| Context window | 131,072 tokens | 1,047,576 tokens |
| Weights | Open | Closed |
| Developer based in | China | the US |
Prices are the developer's list price where we have read it, otherwise the typical host on OpenRouter; see every model's price. Intelligence and speed from Artificial Analysis.
Which is better for what
Each model's rank among models available today, for each kind of work, from our own benchmarks and licensed sources. The better rank is in green.
| Use case | Qwen3 VL 235B A22B Thinking | GPT-4.1 |
|---|---|---|
| Agents and tool useMulti-step tasks with tools: support desks, coding agents, function calls | not measured | 40 of 161 |
| Reasoning and knowledgeHard questions across science, maths and general knowledge | 118 of 195 | 148 of 195 |
| WritingWhat people prefer in blind comparisons, and judged writing quality | 125 of 187 | 54 of 187 |
| Sticking to the factsSummarising without inventing things, and factual answers | not measured | 92 of 153 |
Every shared result
| Benchmark | Metric | Qwen3 VL 235B A22B Thinking | GPT-4.1 |
|---|---|---|---|
| Arena (formerly LMArena)reported | Business, management and financeArena rating | 1,392 | 1,415 |
| Arena (formerly LMArena)reported | Creative writingArena rating | 1,339 | 1,402 |
| Arena (formerly LMArena)reported | Expert promptsArena rating | 1,439 | 1,405 |
| Arena (formerly LMArena)reported | Instruction followingArena rating | 1,383 | 1,403 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,395 | 1,415 |
| Arena (formerly LMArena)reported | Writing, literature and languageArena rating | 1,365 | 1,405 |
| Artificial Analysisreported | Artificial Analysis Intelligence Indexindex score | 13.4 | 12.7 |
| Artificial Analysisreported | GPQA Diamond% of questions | 77.2% | 66.6% |
| Artificial Analysisreported | Humanity's Last Exam% of questions | 11.9% | 4.2% |
| Artificial Analysisreported | IFBench% of instructions | 56.5% | 43.0% |
| Artificial Analysisreported | Long-context reasoning (AA-LCR)% of questions | 63.7% | 68.3% |
| Artificial Analysisreported | Terminal-Bench Hard% of tasks | 11.4% | 13.6% |
| Artificial Analysisreported | Τ²-bench telecom% of tasks | 54.1% | 47.1% |
| UGI Leaderboardreported | Requested-length error% off the requested word count | 30.0% | 6.0% |
| UGI Leaderboardreported | Style adherencescore from 0 to 1 | 0.34 | 0.33 |
| UGI Leaderboardreported | Writing scorescore out of 100 | 32.3 | 67.5 |
Bold green marks the better value on that metric. Where a source reports ranges that overlap, the difference may not be meaningful; see the benchmark page for ranges.
Quick answers
Is Qwen3 VL 235B A22B Thinking better than GPT-4.1?
It depends on the task. Among models available today, on our benchmarks and licensed sources, Qwen3 VL 235B A22B Thinking ranks higher for reasoning and knowledge; GPT-4.1 ranks higher for writing.
Which is cheaper, Qwen3 VL 235B A22B Thinking or GPT-4.1?
Qwen3 VL 235B A22B Thinking costs $1.30 per million tokens (three input to one output), against $3.50 for GPT-4.1.
Run Qwen3 VL 235B A22B Thinking and GPT-4.1 on your own prompt
Benchmarks aren't your data. Try both side by side in the playground, with the cost of every answer.