Compare · ModelsLive · 2 picked · head to head
Qwen2-72B vs Qwen3 235B A22B
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
Qwen3 235B A22B wins on 3/3 benchmarks
Qwen3 235B A22B wins 3 of 3 shared benchmarks. Leads in knowledge · math · coding.
Category leads
knowledge·Qwen3 235B A22Bmath·Qwen3 235B A22Bcoding·Qwen3 235B A22B
Head to head · from the data · data as of 2026-11-22
- Benchmarks · Qwen2-72B leads on 0 of 3 shared benchmarks, Qwen3 235B A22B on 3 (GPQA diamond, MATH level 5, WeirdML)
- Open weights · Qwen2-72B and Qwen3 235B A22B both have open weights
- Price history · Qwen3 235B A22B OpenRouter input list price had no move of 5% or more in the 90 days to 2026-11-22
Sources · scores from public leaderboards, list prices and provider offers from OpenRouter, price moves from the BenchGecko changelog (5% or more, held in the next daily check), Gecko tests run by BenchGecko.
Hype vs Reality
Attention vs performance
Qwen2-72B
#221 by perf·#1 by attention
Qwen3 235B A22B
#101 by perf·#1 by attention
Vendor risk
Who is behind the model
Alibaba (Qwen)
$293.0B·Tier 1
Alibaba (Qwen)
$293.0B·Tier 1
Head to head
3 benchmarks · 2 models
Qwen2-72BQwen3 235B A22B
GPQA diamond
Qwen3 235B A22B leads by +39.0
Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs.
Qwen2-72B
21.9
Qwen3 235B A22B
60.9
MATH level 5
Qwen3 235B A22B leads by +29.8
MATH Level 5 · the hardest tier of the MATH benchmark, featuring competition-level problems from AMC, AIME, and Olympiad-style mathematics.
Qwen2-72B
39.1
Qwen3 235B A22B
68.9
WeirdML
Qwen3 235B A22B leads by +26.0
WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns.
Qwen2-72B
11.3
Qwen3 235B A22B
37.3
Full benchmark table
| Benchmark | Qwen2-72B | Qwen3 235B A22B |
|---|---|---|
GPQA diamond Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs. | 21.9 | 60.9 |
MATH level 5 MATH Level 5 · the hardest tier of the MATH benchmark, featuring competition-level problems from AMC, AIME, and Olympiad-style mathematics. | 39.1 | 68.9 |
WeirdML WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns. | 11.3 | 37.3 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
| Model | Input | Output | Context | Projected $/mo |
|---|---|---|---|---|
| n/a | n/a | n/a | n/a | |
| $0.46 | $1.82 | 131K tokens (~66 books) | $7.96 |
Qwen2-72B vs Qwen3 235B A22B · questions answered from the data
Qwen2-72B leads on 0 of 3 shared benchmarks, Qwen3 235B A22B on 3 (GPQA diamond, MATH level 5, WeirdML)
People also compared