Noometry

Model comparison

GPT-4 Turbo vs Phi 3 Mini 128k Instruct

GPT-4 Turbo is the stronger model overall, scoring 30.3 to 29.1 on the Noometry Index.

Last verified . 19 shared benchmarks.

GPT-4 Turbo OpenAI

30.3

Rank #281 Confirmed

Phi 3 Mini 128k Instruct Microsoft

29.1

Rank #303 Confirmed

Summary

  • They share 19 benchmarks with published results for both. GPT-4 Turbo scores higher in 5 categories and Phi 3 Mini 128k Instruct in 3 categories; 8 gaps are clear of the uncertainty.
  • The widest gap is in writing & preference, where GPT-4 Turbo leads 47.9 to 27.3.
  • The biggest single-benchmark swing is BigCodeBench Instruct: 48.2% for GPT-4 Turbo and 29.6% for Phi 3 Mini 128k Instruct.
  • Phi 3 Mini 128k Instruct has downloadable open weights; the other is API-only.

Side by side

GPT-4 TurboPhi 3 Mini 128k Instruct
ProviderOpenAIMicrosoft
Noometry Index30.329.1
Released2023-11-062024-04-23
WeightsProprietaryOpen
Context window128K—
Max output4K—
Input $ / M tokens$10—
Output $ / M tokens$30—
Results tracked3619

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

Category by category

Coding GPT-4 Turbo leads

GPT-4 Turbo: 33.4 (#247), Phi 3 Mini 128k Instruct: 28.3 (#307)

BenchmarkGPT-4 TurboPhi 3 Mini 128k Instruct
BigCodeBench Instruct48.2%29.6%
LMArena Coding12681040
BigCodeBench Complete58.2%40.6%
WeirdML18%—
HumanEval+86.6%—
MBPP+73.3%—

Agentic & Tool Use Not comparable

GPT-4 Turbo: —, Phi 3 Mini 128k Instruct: —

BenchmarkGPT-4 TurboPhi 3 Mini 128k Instruct
METR Time Horizons36.7%—

Reasoning Phi 3 Mini 128k Instruct leads

GPT-4 Turbo: 14.9 (#311), Phi 3 Mini 128k Instruct: 18.7 (#254)

BenchmarkGPT-4 TurboPhi 3 Mini 128k Instruct
LMArena Hard Prompts12511028
SimpleBench25.1%—
Chess Puzzles6%—
DTBench61.6%—
LMCA9.8%—
Epoch Capabilities Index127.25—
ForecastBench59.4—

Math Phi 3 Mini 128k Instruct leads

GPT-4 Turbo: 8.6 (#316), Phi 3 Mini 128k Instruct: 29.2 (#220)

BenchmarkGPT-4 TurboPhi 3 Mini 128k Instruct
LMArena Math12721089
FrontierMath (Tiers 1-3)0.7%—
OTIS Mock AIME 2024-20256.7%—
MATH Level 546.7%—

Knowledge Phi 3 Mini 128k Instruct leads

GPT-4 Turbo: 24.3 (#263), Phi 3 Mini 128k Instruct: 26.3 (#251)

BenchmarkGPT-4 TurboPhi 3 Mini 128k Instruct
LMArena Expert1223984
GPQA Diamond46.6%—
Confabulations28.4%—
MMLU81.3%—

Multimodal Not comparable

GPT-4 Turbo: 30.0 (#104), Phi 3 Mini 128k Instruct: —

BenchmarkGPT-4 TurboPhi 3 Mini 128k Instruct
LMArena Vision1090—

Multilingual GPT-4 Turbo leads

GPT-4 Turbo: 40.7 (#211), Phi 3 Mini 128k Instruct: 25.4 (#280)

BenchmarkGPT-4 TurboPhi 3 Mini 128k Instruct
LMArena Non-English12451000
LMArena Chinese12421016
LMArena French12751038
LMArena German12591006
LMArena Japanese1194899
LMArena Korean1187856
LMArena Russian12591004
LMArena Spanish12601058

Instruction Following GPT-4 Turbo leads

GPT-4 Turbo: 66.0 (#211), Phi 3 Mini 128k Instruct: 52.1 (#289)

BenchmarkGPT-4 TurboPhi 3 Mini 128k Instruct
LMArena Instruction Following12501023

Long Context GPT-4 Turbo leads

GPT-4 Turbo: 38.0 (#200), Phi 3 Mini 128k Instruct: 30.3 (#284)

BenchmarkGPT-4 TurboPhi 3 Mini 128k Instruct
LMArena Longer Query1254996

Writing & Preference GPT-4 Turbo leads

GPT-4 Turbo: 47.9 (#201), Phi 3 Mini 128k Instruct: 27.3 (#296)

BenchmarkGPT-4 TurboPhi 3 Mini 128k Instruct
LMArena Text12721050
LMArena Creative Writing12691024
LMArena Multi-Turn1267989

Frequently asked questions

Is GPT-4 Turbo better than Phi 3 Mini 128k Instruct?

GPT-4 Turbo is the stronger model overall, scoring 30.3 to 29.1 on the Noometry Index.

Is GPT-4 Turbo or Phi 3 Mini 128k Instruct better for coding?

GPT-4 Turbo scores higher on coding benchmarks: 33.4 versus 28.3 in the Noometry coding category.

How many benchmarks do GPT-4 Turbo and Phi 3 Mini 128k Instruct share?

19 benchmarks have published results for both models. GPT-4 Turbo has 36 scored results on Noometry and Phi 3 Mini 128k Instruct has 19.

Related comparisons

Go deeper