Noometry

Model comparison

GPT-5.4 vs Qwen2.5 72B Instruct

GPT-5.4 is the stronger model overall, scoring 59.4 to 31.9 on the Noometry Index. Qwen2.5 72B Instruct costs 2.3× less per token, which makes it the better buy when GPT-5.4's lead doesn't matter for your workload.

Last verified . 25 shared benchmarks.

GPT-5.4 OpenAI

59.4

Rank #16 Confirmed

Qwen2.5 72B Instruct Alibaba (Qwen)

31.9

Rank #267 Confirmed

Summary

  • They share 25 benchmarks with published results for both. GPT-5.4 scores higher in 9 categories and Qwen2.5 72B Instruct in 0 categories; 9 gaps are clear of the uncertainty.
  • The widest gap is in math, where GPT-5.4 leads 73.5 to 19.3.
  • The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 97.8% for GPT-5.4 and 8.1% for Qwen2.5 72B Instruct.
  • Qwen2.5 72B Instruct is cheaper at $1.40 / $5.60 per million input/output tokens, against $2.50 / $15 for GPT-5.4.
  • GPT-5.4 accepts more context: 1.05M tokens versus 131K.
  • Qwen2.5 72B Instruct has downloadable open weights; the other is API-only.

Side by side

GPT-5.4Qwen2.5 72B Instruct
ProviderOpenAIAlibaba (Qwen)
Noometry Index59.431.9
Released2026-03-052024-09
WeightsProprietaryOpen
Context window1.05M131K
Max output128K8K
Input $ / M tokens$2.50$1.40
Output $ / M tokens$15$5.60
Results tracked6843

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

Category by category

Coding GPT-5.4 leads

GPT-5.4: 52.6 (#33), Qwen2.5 72B Instruct: 33.2 (#260)

BenchmarkGPT-5.4Qwen2.5 72B Instruct
WeirdML77.7%16%
LMArena Coding14971292
SWE-bench Verified76.9%—
DeepSWE51.8%—
LMArena WebDev1465—
SciCode56.6%—
GSO31.4%—
BigCodeBench Instruct—45.8%
MirrorCode15.6%—
BigCodeBench Complete—55.9%
ALE-Bench1,607—
AlgoTune1.85—

Agentic & Tool Use GPT-5.4 leads

GPT-5.4: 46.5 (#13), Qwen2.5 72B Instruct: 22.1 (#133)

BenchmarkGPT-5.4Qwen2.5 72B Instruct
METR Time Horizons74.3%35.8%
Terminal-Bench81.8%—
APEX-Agents52.4%—
TheAgentCompany—5.7%
τ²-bench Banking39.4%—
DeepResearch Bench35.1%—
PostTrainBench19%—
BALROG—16.2%
GBAEval45.1%—
LMArena Search1197—
Vending-Bench 26,144—

Reasoning GPT-5.4 leads

GPT-5.4: 61.8 (#19), Qwen2.5 72B Instruct: 22.3 (#199)

BenchmarkGPT-5.4Qwen2.5 72B Instruct
LMArena Hard Prompts14851271
DTBench94.4%62.9%
LMCA52%13.4%
Epoch Capabilities Index156.81129
ForecastBench59.557.5
ARC-AGI-274%—
Kagi LLM Benchmark63.8%—
NYT Connections (extended)91.3%—
ARC-AGI-193.7%—
CritPt23.4%—
Chess Puzzles44%—
EnigmaEval16%—
Thematic Generalization80%—
EBR-Bench25.4%—
Mystery Game Puzzles37%—
BIG-Bench Hard—79.8%
HellaSwag—84.8%
PIQA—82.6%
WinoGrande—82.3%

Math GPT-5.4 leads

GPT-5.4: 73.5 (#19), Qwen2.5 72B Instruct: 19.3 (#287)

BenchmarkGPT-5.4Qwen2.5 72B Instruct
OTIS Mock AIME 2024-202597.8%8.1%
LMArena Math14881283
FrontierMath (Tiers 1-3)78.6%—
FrontierMath Tier 449%—
MathArena Final-Answer Competitions83.1%—
ProofBench56%—
Omni-MATH—33%
MATH Level 5—63.2%
FrontierMath (Feb 2025 set)47.6%—
FrontierMath Tier 4 (v1)27.1%—

Knowledge GPT-5.4 leads

GPT-5.4: 65.3 (#14), Qwen2.5 72B Instruct: 27.0 (#253)

BenchmarkGPT-5.4Qwen2.5 72B Instruct
GPQA Diamond93.3%49.1%
LMArena Expert15071245
Humanity's Last Exam36.2%—
SimpleQA Verified45.1%—
MMLU-Pro—63.1%
Confabulations—19.1%
Vectara Hallucination Rate7%—
GPQA (HELM)—42.6%
ARC (AI2) Challenge—94.5%
MMLU—85.3%
TriviaQA—71.9%

Multimodal Not comparable

GPT-5.4: 43.7 (#20), Qwen2.5 72B Instruct: —

BenchmarkGPT-5.4Qwen2.5 72B Instruct
LMArena Vision1303—
Blueprint-Bench 227.1%—
Furniture Assembly37.5%—
LMArena Document1471—

Multilingual GPT-5.4 leads

GPT-5.4: 56.2 (#23), Qwen2.5 72B Instruct: 41.0 (#213)

BenchmarkGPT-5.4Qwen2.5 72B Instruct
LMArena Non-English14651252
LMArena Chinese15191272
LMArena French14931280
LMArena German14721234
LMArena Japanese14851180
LMArena Korean14481188
LMArena Russian14801264
LMArena Spanish14541256

Instruction Following GPT-5.4 leads

GPT-5.4: 77.1 (#27), Qwen2.5 72B Instruct: 65.5 (#221)

BenchmarkGPT-5.4Qwen2.5 72B Instruct
LMArena Instruction Following14691254
IFEval—80.6%

Long Context GPT-5.4 leads

GPT-5.4: 50.3 (#8), Qwen2.5 72B Instruct: 38.9 (#188)

BenchmarkGPT-5.4Qwen2.5 72B Instruct
LMArena Longer Query14731282
CL-bench27.9%—
CL-bench Life21.7%—

Writing & Preference GPT-5.4 leads

GPT-5.4: 71.9 (#17), Qwen2.5 72B Instruct: 46.7 (#215)

BenchmarkGPT-5.4Qwen2.5 72B Instruct
LMArena Text14691269
LMArena Creative Writing14391221
LMArena Multi-Turn14821272
EQ-Bench Creative Writing1840—
WildBench—80.2%
EQ-Bench 41272—

Frequently asked questions

Is GPT-5.4 better than Qwen2.5 72B Instruct?

GPT-5.4 is the stronger model overall, scoring 59.4 to 31.9 on the Noometry Index. Qwen2.5 72B Instruct costs 2.3× less per token, which makes it the better buy when GPT-5.4's lead doesn't matter for your workload.

Which is cheaper, GPT-5.4 or Qwen2.5 72B Instruct?

Qwen2.5 72B Instruct is cheaper. It lists at $1.40 per million input tokens and $5.60 per million output tokens; GPT-5.4 lists at $2.50 and $15.

Is GPT-5.4 or Qwen2.5 72B Instruct better for coding?

GPT-5.4 scores higher on coding benchmarks: 52.6 versus 33.2 in the Noometry coding category.

Which has the bigger context window?

GPT-5.4 does, with 1.05M tokens against 131K.

How many benchmarks do GPT-5.4 and Qwen2.5 72B Instruct share?

25 benchmarks have published results for both models. GPT-5.4 has 68 scored results on Noometry and Qwen2.5 72B Instruct has 43.

Related comparisons

Go deeper