Noometry

Model comparison

GPT-4o mini vs Qwen2.5 72B Instruct

Qwen2.5 72B Instruct is the stronger model overall, scoring 31.5 to 25.4 on the Noometry Index. GPT-4o mini costs 9.3× less per token, which makes it the better buy when Qwen2.5 72B Instruct's lead doesn't matter for your workload.

Last verified . 35 shared benchmarks.

GPT-4o mini OpenAI

25.4

Rank #337 Confirmed

Qwen2.5 72B Instruct Alibaba (Qwen)

31.5

Rank #259 Confirmed

Summary

  • They share 35 benchmarks with published results for both. GPT-4o mini scores higher in 3 categories and Qwen2.5 72B Instruct in 6 categories; 8 gaps are clear of the uncertainty.
  • The widest gap is in reasoning, where Qwen2.5 72B Instruct leads 21.4 to 8.7.
  • The biggest single-benchmark swing is Confabulations: 37.2% for GPT-4o mini and 19.1% for Qwen2.5 72B Instruct.
  • GPT-4o mini is cheaper at $0.15 / $0.60 per million input/output tokens, against $1.40 / $5.60 for Qwen2.5 72B Instruct.
  • Qwen2.5 72B Instruct accepts more context: 131K tokens versus 128K.
  • Qwen2.5 72B Instruct has downloadable open weights; the other is API-only.

Side by side

GPT-4o miniQwen2.5 72B Instruct
ProviderOpenAIAlibaba (Qwen)
Noometry Index25.431.5
Released2024-07-182024-09
WeightsProprietaryOpen
Context window128K131K
Max output16K8K
Input $ / M tokens$0.15$1.40
Output $ / M tokens$0.60$5.60
Results tracked6043

Sponsored placements are available on pages like this one. Advertise on Noometry

Category by category

Coding Qwen2.5 72B Instruct leads

GPT-4o mini: 21.6 (#330), Qwen2.5 72B Instruct: 32.8 (#256)

BenchmarkGPT-4o miniQwen2.5 72B Instruct
WeirdML11.8%16%
BigCodeBench Instruct46.1%45.8%
LMArena Coding12901293
BigCodeBench Complete57.4%55.9%
Aider Polyglot3.6%—
LiveBench Coding43.1%—
HumanEval+83.5%—
MBPP+72.2%—

Agentic & Tool Use GPT-4o mini leads

GPT-4o mini: 27.2 (#95), Qwen2.5 72B Instruct: 22.0 (#124)

BenchmarkGPT-4o miniQwen2.5 72B Instruct
BALROG17.4%16.2%
TheAgentCompany—5.7%
METR Time Horizons—35.8%

Reasoning Qwen2.5 72B Instruct leads

GPT-4o mini: 8.7 (#343), Qwen2.5 72B Instruct: 21.4 (#197)

BenchmarkGPT-4o miniQwen2.5 72B Instruct
LMArena Hard Prompts12671271
DTBench54.4%62.9%
LMCA10.4%13.4%
Epoch Capabilities Index126.56129.01
PIQA88.7%82.6%
ARC-AGI-20%—
SimpleBench10.7%—
Kagi LLM Benchmark28.8%—
Chess Puzzles0%—
LiveBench Reasoning32.8%—
Mystery Game Puzzles12%—
LiveBench Data Analysis50%—
BIG-Bench Hard—79.8%
ForecastBench—57.5
HellaSwag—84.8%
LiveBench41.3%—
WinoGrande—82.3%

Math Qwen2.5 72B Instruct leads

GPT-4o mini: 9.9 (#307), Qwen2.5 72B Instruct: 18.1 (#281)

BenchmarkGPT-4o miniQwen2.5 72B Instruct
OTIS Mock AIME 2024-20256.9%8.1%
Omni-MATH28%33%
LMArena Math12671283
MATH Level 552.6%63.2%
FrontierMath (Tiers 1-3)0.7%—
LiveBench Math36.3%—
GSM8K91.3%—

Knowledge Qwen2.5 72B Instruct leads

GPT-4o mini: 17.7 (#279), Qwen2.5 72B Instruct: 26.9 (#247)

BenchmarkGPT-4o miniQwen2.5 72B Instruct
GPQA Diamond37.7%49.1%
MMLU-Pro60.3%63.1%
Confabulations37.2%19.1%
GPQA (HELM)36.8%42.6%
LMArena Expert12351245
MMLU81.8%85.3%
SimpleQA Verified8.3%—
ARC (AI2) Challenge—94.5%
BoolQ88.7%—
TriviaQA—71.9%

Multimodal Not comparable

GPT-4o mini: 25.3 (#117), Qwen2.5 72B Instruct: —

BenchmarkGPT-4o miniQwen2.5 72B Instruct
LMArena Vision1066—
Video-MME64.8%—
GeoBench64%—
VPCT34%—

Multilingual GPT-4o mini leads

GPT-4o mini: 42.2 (#194), Qwen2.5 72B Instruct: 41.2 (#208)

BenchmarkGPT-4o miniQwen2.5 72B Instruct
LMArena Non-English12661252
LMArena Chinese12641272
LMArena French12961279
LMArena German12721234
LMArena Japanese12161180
LMArena Korean11951188
LMArena Russian12751264
LMArena Spanish12761255

Instruction Following Qwen2.5 72B Instruct leads

GPT-4o mini: 62.1 (#234), Qwen2.5 72B Instruct: 65.7 (#216)

BenchmarkGPT-4o miniQwen2.5 72B Instruct
IFEval78.2%80.6%
LMArena Instruction Following12591254
LiveBench Instruction Following56.8%—

Long Context Too close to call

GPT-4o mini: 39.1 (#181), Qwen2.5 72B Instruct: 38.9 (#183)

BenchmarkGPT-4o miniQwen2.5 72B Instruct
LMArena Longer Query12891282

Writing & Preference Qwen2.5 72B Instruct leads

GPT-4o mini: 39.7 (#243), Qwen2.5 72B Instruct: 46.9 (#210)

BenchmarkGPT-4o miniQwen2.5 72B Instruct
LMArena Text12861269
LMArena Creative Writing12681221
WildBench79.1%80.2%
LMArena Multi-Turn12851272
Short-Story Creative Writing67.2%—
EQ-Bench Creative Writing873—
LiveBench Language28.6%—

Frequently asked questions

Is GPT-4o mini better than Qwen2.5 72B Instruct?

Qwen2.5 72B Instruct is the stronger model overall, scoring 31.5 to 25.4 on the Noometry Index. GPT-4o mini costs 9.3× less per token, which makes it the better buy when Qwen2.5 72B Instruct's lead doesn't matter for your workload.

Which is cheaper, GPT-4o mini or Qwen2.5 72B Instruct?

GPT-4o mini is cheaper. It lists at $0.15 per million input tokens and $0.60 per million output tokens; Qwen2.5 72B Instruct lists at $1.40 and $5.60.

Is GPT-4o mini or Qwen2.5 72B Instruct better for coding?

Qwen2.5 72B Instruct scores higher on coding benchmarks: 32.8 versus 21.6 in the Noometry coding category.

Which has the bigger context window?

Qwen2.5 72B Instruct does, with 131K tokens against 128K.

How many benchmarks do GPT-4o mini and Qwen2.5 72B Instruct share?

35 benchmarks have published results for both models. GPT-4o mini has 60 scored results on Noometry and Qwen2.5 72B Instruct has 43.

Related comparisons

Go deeper