Compare · ModelsLive · 2 picked · head to head
GPT-4 Turbo (batch) vs Qwen2.5 72B Instruct
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
Qwen2.5 72B Instruct wins on 8/13 benchmarks
Qwen2.5 72B Instruct wins 8 of 13 shared benchmarks. Leads in reasoning · knowledge · math.
Category leads
reasoning·Qwen2.5 72B Instructknowledge·Qwen2.5 72B Instructgeneral·GPT-4 Turbo (batch)math·Qwen2.5 72B Instructcoding·Qwen2.5 72B Instruct
Hype vs Reality
Attention vs performance
GPT-4 Turbo (batch)
#188 by perf·no signal
Qwen2.5 72B Instruct
#141 by perf·#1 by attention
Best value
Qwen2.5 72B Instruct
31.0x better value than GPT-4 Turbo (batch)
GPT-4 Turbo (batch)
4.2 pts/$
$10.00/M
Qwen2.5 72B Instruct
129.7 pts/$
$0.38/M
Vendor risk
Who is behind the model
OpenAI
$840.0B·Tier 1
Alibaba (Qwen)
$293.0B·Tier 1
Head to head
13 benchmarks · 2 models
GPT-4 Turbo (batch)Qwen2.5 72B Instruct
BBH
Qwen2.5 72B Instruct leads by +6.2
BIG-Bench Hard · a curated subset of 23 challenging tasks from BIG-Bench where language models previously failed to outperform average humans.
GPT-4 Turbo (batch)
66.8
Qwen2.5 72B Instruct
73.1
CMMLU
Qwen2.5 72B Instruct leads by +14.7
GPT-4 Turbo (batch)
71.0
Qwen2.5 72B Instruct
85.7
Dtbench
Qwen2.5 72B Instruct leads by +0.4
GPT-4 Turbo (batch)
37.8
Qwen2.5 72B Instruct
38.2
GPQA diamond
Qwen2.5 72B Instruct leads by +25.7
Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs.
GPT-4 Turbo (batch)
7.8
Qwen2.5 72B Instruct
33.5
HellaSwag
GPT-4 Turbo (batch) leads by +14.0
HellaSwag · tests commonsense reasoning by asking models to predict the most plausible continuation of everyday scenarios.
GPT-4 Turbo (batch)
93.7
Qwen2.5 72B Instruct
79.7
Lmca
GPT-4 Turbo (batch) leads by +4.3
GPT-4 Turbo (batch)
20.1
Qwen2.5 72B Instruct
15.8
MATH level 5
Qwen2.5 72B Instruct leads by +40.2
MATH Level 5 · the hardest tier of the MATH benchmark, featuring competition-level problems from AMC, AIME, and Olympiad-style mathematics.
GPT-4 Turbo (batch)
23.0
Qwen2.5 72B Instruct
63.2
Metr Time Horizons
GPT-4 Turbo (batch) leads by +0.3
GPT-4 Turbo (batch)
36.1
Qwen2.5 72B Instruct
35.8
MMLU
Qwen2.5 72B Instruct leads by +3.9
Massive Multitask Language Understanding · 57 subjects spanning STEM, humanities, social sciences, and more. The standard benchmark for broad knowledge.
GPT-4 Turbo (batch)
76.5
Qwen2.5 72B Instruct
80.4
OTIS Mock AIME 2024-2025
Qwen2.5 72B Instruct leads by +7.0
OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills.
GPT-4 Turbo (batch)
1.0
Qwen2.5 72B Instruct
8.0
TriviaQA
GPT-4 Turbo (batch) leads by +12.9
TriviaQA · reading comprehension benchmark with trivia questions, requiring models to find and reason over evidence from provided documents.
GPT-4 Turbo (batch)
84.8
Qwen2.5 72B Instruct
71.9
WeirdML
Qwen2.5 72B Instruct leads by +3.6
WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns.
GPT-4 Turbo (batch)
12.4
Qwen2.5 72B Instruct
16.0
Winogrande
GPT-4 Turbo (batch) leads by +10.4
WinoGrande · large-scale commonsense reasoning benchmark where models must resolve ambiguous pronouns in carefully constructed sentence pairs.
GPT-4 Turbo (batch)
75.0
Qwen2.5 72B Instruct
64.6
Full benchmark table
| Benchmark | GPT-4 Turbo (batch) | Qwen2.5 72B Instruct |
|---|---|---|
BBH BIG-Bench Hard · a curated subset of 23 challenging tasks from BIG-Bench where language models previously failed to outperform average humans. | 66.8 | 73.1 |
CMMLU | 71.0 | 85.7 |
Dtbench | 37.8 | 38.2 |
GPQA diamond Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs. | 7.8 | 33.5 |
HellaSwag HellaSwag · tests commonsense reasoning by asking models to predict the most plausible continuation of everyday scenarios. | 93.7 | 79.7 |
Lmca | 20.1 | 15.8 |
MATH level 5 MATH Level 5 · the hardest tier of the MATH benchmark, featuring competition-level problems from AMC, AIME, and Olympiad-style mathematics. | 23.0 | 63.2 |
Metr Time Horizons | 36.1 | 35.8 |
MMLU Massive Multitask Language Understanding · 57 subjects spanning STEM, humanities, social sciences, and more. The standard benchmark for broad knowledge. | 76.5 | 80.4 |
OTIS Mock AIME 2024-2025 OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills. | 1.0 | 8.0 |
TriviaQA TriviaQA · reading comprehension benchmark with trivia questions, requiring models to find and reason over evidence from provided documents. | 84.8 | 71.9 |
WeirdML WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns. | 12.4 | 16.0 |
Winogrande WinoGrande · large-scale commonsense reasoning benchmark where models must resolve ambiguous pronouns in carefully constructed sentence pairs. | 75.0 | 64.6 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
| Model | Input | Output | Context | Projected $/mo |
|---|---|---|---|---|
| $5.00 | $15.00 | 128K tokens (~64 books) | $75.00 | |
| $0.36 | $0.40 | 33K tokens (~16 books) | $3.70 |