Compare · ModelsLive · 2 picked · head to head
GPT-5 Mini vs Qwen3 235B A22B Thinking 2507
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
GPT-5 Mini wins on 14/20 benchmarks
GPT-5 Mini wins 14 of 20 shared benchmarks. Leads in knowledge · general · math.
Category leads
knowledge·GPT-5 Minigeneral·GPT-5 Minimath·GPT-5 Minicoding·GPT-5 Minireasoning·Qwen3 235B A22B Thinking 2507language·GPT-5 Mini
Hype vs Reality
Attention vs performance
GPT-5 Mini
#116 by perf·#8 by attention
Qwen3 235B A22B Thinking 2507
#104 by perf·#2 by attention
Best value
GPT-5 Mini
1.1x better value than Qwen3 235B A22B Thinking 2507
GPT-5 Mini
45.7 pts/$
$1.13/M
Qwen3 235B A22B Thinking 2507
41.9 pts/$
$1.26/M
Vendor risk
Who is behind the model
OpenAI
$840.0B·Tier 1
Alibaba (Qwen)
$293.0B·Tier 1
Head to head
20 benchmarks · 2 models
GPT-5 MiniQwen3 235B A22B Thinking 2507
Chess Puzzles
GPT-5 Mini leads by +12.6
Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities.
GPT-5 Mini
20.0
Qwen3 235B A22B Thinking 2507
7.4
Dtbench
GPT-5 Mini leads by +0.4
GPT-5 Mini
67.5
Qwen3 235B A22B Thinking 2507
67.1
Fiction.LiveBench
Qwen3 235B A22B Thinking 2507 leads by +5.6
Fiction.LiveBench · a continuously updated benchmark using recently published fiction to test reading comprehension and reasoning, preventing data contamination.
GPT-5 Mini
69.4
Qwen3 235B A22B Thinking 2507
75.0
FrontierMath-2025-02-28-Private
GPT-5 Mini leads by +12.4
FrontierMath (Feb 2025) · original research-level math problems created by mathematicians, testing capabilities at the boundary of current AI mathematical reasoning.
GPT-5 Mini
27.2
Qwen3 235B A22B Thinking 2507
14.9
FrontierMath-Tier-4-2025-07-01-Private
GPT-5 Mini leads by +6.3
FrontierMath Tier 4 (Jul 2025) · the most challenging tier of frontier mathematics, containing problems that push the absolute limits of AI mathematical reasoning.
GPT-5 Mini
6.3
Qwen3 235B A22B Thinking 2507
0.0
GPQA diamond
Qwen3 235B A22B Thinking 2507 leads by +6.7
Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs.
GPT-5 Mini
66.7
Qwen3 235B A22B Thinking 2507
73.4
Lech Mazur Writing
GPT-5 Mini leads by +0.7
Lech Mazur Writing · evaluates creative writing ability, assessing prose quality, narrative coherence, and stylistic sophistication.
GPT-5 Mini
83.1
Qwen3 235B A22B Thinking 2507
82.4
LiveBench · Agentic Coding
GPT-5 Mini leads by +28.3
GPT-5 Mini
35.0
Qwen3 235B A22B Thinking 2507
6.7
LiveBench · Coding
GPT-5 Mini leads by +7.1
GPT-5 Mini
76.1
Qwen3 235B A22B Thinking 2507
69.0
LiveBench · Data Analysis
Qwen3 235B A22B Thinking 2507 leads by +2.6
GPT-5 Mini
49.6
Qwen3 235B A22B Thinking 2507
52.2
LiveBench · If
GPT-5 Mini leads by +23.6
GPT-5 Mini
64.2
Qwen3 235B A22B Thinking 2507
40.6
LiveBench · Language
Qwen3 235B A22B Thinking 2507 leads by +0.4
GPT-5 Mini
69.2
Qwen3 235B A22B Thinking 2507
69.5
LiveBench · Mathematics
GPT-5 Mini leads by +1.0
GPT-5 Mini
74.4
Qwen3 235B A22B Thinking 2507
73.4
LiveBench · Overall
GPT-5 Mini leads by +8.0
GPT-5 Mini
61.0
Qwen3 235B A22B Thinking 2507
53.0
LiveBench · Reasoning
Qwen3 235B A22B Thinking 2507 leads by +0.8
GPT-5 Mini
58.6
Qwen3 235B A22B Thinking 2507
59.4
Lmca
GPT-5 Mini leads by +5.8
GPT-5 Mini
40.2
Qwen3 235B A22B Thinking 2507
34.5
Mystery Game Puzzles
GPT-5 Mini leads by +0.9
GPT-5 Mini
0.9
Qwen3 235B A22B Thinking 2507
0.0
OTIS Mock AIME 2024-2025
OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills.
GPT-5 Mini
86.7
Qwen3 235B A22B Thinking 2507
86.7
SimpleQA Verified
Qwen3 235B A22B Thinking 2507 leads by +18.8
SimpleQA Verified · short factual questions with verified answers, measuring factual accuracy and the tendency to hallucinate or provide incorrect information.
GPT-5 Mini
21.6
Qwen3 235B A22B Thinking 2507
40.4
WeirdML
GPT-5 Mini leads by +11.6
WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns.
GPT-5 Mini
52.7
Qwen3 235B A22B Thinking 2507
41.0
Full benchmark table
| Benchmark | GPT-5 Mini | Qwen3 235B A22B Thinking 2507 |
|---|---|---|
Chess Puzzles Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities. | 20.0 | 7.4 |
Dtbench | 67.5 | 67.1 |
Fiction.LiveBench Fiction.LiveBench · a continuously updated benchmark using recently published fiction to test reading comprehension and reasoning, preventing data contamination. | 69.4 | 75.0 |
FrontierMath-2025-02-28-Private FrontierMath (Feb 2025) · original research-level math problems created by mathematicians, testing capabilities at the boundary of current AI mathematical reasoning. | 27.2 | 14.9 |
FrontierMath-Tier-4-2025-07-01-Private FrontierMath Tier 4 (Jul 2025) · the most challenging tier of frontier mathematics, containing problems that push the absolute limits of AI mathematical reasoning. | 6.3 | 0.0 |
GPQA diamond Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs. | 66.7 | 73.4 |
Lech Mazur Writing Lech Mazur Writing · evaluates creative writing ability, assessing prose quality, narrative coherence, and stylistic sophistication. | 83.1 | 82.4 |
LiveBench · Agentic Coding | 35.0 | 6.7 |
LiveBench · Coding | 76.1 | 69.0 |
LiveBench · Data Analysis | 49.6 | 52.2 |
LiveBench · If | 64.2 | 40.6 |
LiveBench · Language | 69.2 | 69.5 |
LiveBench · Mathematics | 74.4 | 73.4 |
LiveBench · Overall | 61.0 | 53.0 |
LiveBench · Reasoning | 58.6 | 59.4 |
Lmca | 40.2 | 34.5 |
Mystery Game Puzzles | 0.9 | 0.0 |
OTIS Mock AIME 2024-2025 OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills. | 86.7 | 86.7 |
SimpleQA Verified SimpleQA Verified · short factual questions with verified answers, measuring factual accuracy and the tendency to hallucinate or provide incorrect information. | 21.6 | 40.4 |
WeirdML WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns. | 52.7 | 41.0 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
| Model | Input | Output | Context | Projected $/mo |
|---|---|---|---|---|
| $0.25 | $2.00 | 400K tokens (~200 books) | $6.88 | |
| $0.23 | $2.30 | 128K tokens (~64 books) | $7.47 |