Compare · ModelsLive · 2 picked · head to head
GPT-5.1 vs Qwen3 235B A22B
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
GPT-5.1 wins on 6/6 benchmarks
GPT-5.1 wins 6 of 6 shared benchmarks. Leads in arena · general · knowledge.
Category leads
arena·GPT-5.1general·GPT-5.1knowledge·GPT-5.1reasoning·GPT-5.1coding·GPT-5.1
Hype vs Reality
Attention vs performance
GPT-5.1
#139 by perf·#10 by attention
Qwen3 235B A22B
#96 by perf·#2 by attention
Best value
Qwen3 235B A22B
5.5x better value than GPT-5.1
GPT-5.1
8.6 pts/$
$5.63/M
Qwen3 235B A22B
47.2 pts/$
$1.14/M
Vendor risk
Who is behind the model
OpenAI
$840.0B·Tier 1
Alibaba (Qwen)
$293.0B·Tier 1
Head to head
6 benchmarks · 2 models
GPT-5.1Qwen3 235B A22B
Chatbot Arena Elo · Overall
GPT-5.1 leads by +63.3
GPT-5.1
1438.5
Qwen3 235B A22B
1375.2
Dtbench
GPT-5.1 leads by +24.0
GPT-5.1
83.5
Qwen3 235B A22B
59.5
GPQA diamond
GPT-5.1 leads by +22.6
Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs.
GPT-5.1
83.5
Qwen3 235B A22B
60.9
Lmca
GPT-5.1 leads by +22.2
GPT-5.1
51.6
Qwen3 235B A22B
29.4
SimpleBench
GPT-5.1 leads by +26.6
SimpleBench · tests fundamental reasoning capabilities with straightforward problems designed to expose gaps in basic logical and spatial thinking.
GPT-5.1
43.8
Qwen3 235B A22B
17.2
WeirdML
GPT-5.1 leads by +23.5
WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns.
GPT-5.1
60.8
Qwen3 235B A22B
37.3
Full benchmark table
| Benchmark | GPT-5.1 | Qwen3 235B A22B |
|---|---|---|
Chatbot Arena Elo · Overall | 1438.5 | 1375.2 |
Dtbench | 83.5 | 59.5 |
GPQA diamond Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs. | 83.5 | 60.9 |
Lmca | 51.6 | 29.4 |
SimpleBench SimpleBench · tests fundamental reasoning capabilities with straightforward problems designed to expose gaps in basic logical and spatial thinking. | 43.8 | 17.2 |
WeirdML WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns. | 60.8 | 37.3 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
| Model | Input | Output | Context | Projected $/mo |
|---|---|---|---|---|
| $1.25 | $10.00 | 400K tokens (~200 books) | $34.38 | |
| $0.46 | $1.82 | 131K tokens (~66 books) | $7.96 |
People also compared