Noometry

Model comparison

GPT-4 vs Granite 3.0 8b Instruct

Granite 3.0 8b Instruct is the stronger model overall, scoring 31.1 to 28.9 on the Noometry Index.

Last verified . 14 shared benchmarks.

GPT-4 OpenAI

28.9

Rank #308 Confirmed

Granite 3.0 8b Instruct IBM

31.1

Rank #266 Confirmed

Summary

  • They share 14 benchmarks with published results for both. GPT-4 scores higher in 5 categories and Granite 3.0 8b Instruct in 3 categories; 8 gaps are clear of the uncertainty.
  • The widest gap is in math, where Granite 3.0 8b Instruct leads 30.3 to 10.1.
  • The biggest single-benchmark swing is BigCodeBench Complete: 57.2% for GPT-4 and 35.4% for Granite 3.0 8b Instruct.
  • Granite 3.0 8b Instruct has downloadable open weights; the other is API-only.

Side by side

GPT-4Granite 3.0 8b Instruct
ProviderOpenAIIBM
Noometry Index28.931.1
Released2023-03-14—
WeightsProprietaryOpen
Context window8K—
Max output8K—
Input $ / M tokens$30—
Output $ / M tokens$60—
Results tracked3814

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

Category by category

Coding GPT-4 leads

GPT-4: 31.2 (#277), Granite 3.0 8b Instruct: 29.1 (#297)

BenchmarkGPT-4Granite 3.0 8b Instruct
BigCodeBench Instruct46%29.3%
LMArena Coding12541113
BigCodeBench Complete57.2%35.4%
WeirdML12.4%—
HumanEval+79.3%—

Agentic & Tool Use Not comparable

GPT-4: —, Granite 3.0 8b Instruct: —

BenchmarkGPT-4Granite 3.0 8b Instruct
METR Time Horizons36.1%—

Reasoning Granite 3.0 8b Instruct leads

GPT-4: 17.3 (#285), Granite 3.0 8b Instruct: 20.0 (#227)

BenchmarkGPT-4Granite 3.0 8b Instruct
LMArena Hard Prompts12411092
Chess Puzzles4%—
Mystery Game Puzzles12%—
DTBench62.7%—
LMCA17.1%—
BIG-Bench Hard75.1%—
Epoch Capabilities Index125.88—
ForecastBench57.8—
HellaSwag95.3%—
WinoGrande87.5%—

Math Granite 3.0 8b Instruct leads

GPT-4: 10.1 (#304), Granite 3.0 8b Instruct: 30.3 (#211)

BenchmarkGPT-4Granite 3.0 8b Instruct
LMArena Math12691143
OTIS Mock AIME 2024-20251.1%—
MATH Level 523%—
GSM8K92%—

Knowledge Granite 3.0 8b Instruct leads

GPT-4: 18.4 (#277), Granite 3.0 8b Instruct: 29.1 (#233)

BenchmarkGPT-4Granite 3.0 8b Instruct
LMArena Expert12111087
GPQA Diamond35.7%—
MMLU86.4%—
TriviaQA84.8%—

Multilingual GPT-4 leads

GPT-4: 40.8 (#210), Granite 3.0 8b Instruct: 27.4 (#271)

BenchmarkGPT-4Granite 3.0 8b Instruct
LMArena Non-English12461037
LMArena Chinese12421063
LMArena Russian12511060
LMArena French1282—
LMArena German1251—
LMArena Japanese1209—
LMArena Korean1184—
LMArena Spanish1261—

Instruction Following GPT-4 leads

GPT-4: 65.5 (#217), Granite 3.0 8b Instruct: 56.2 (#271)

BenchmarkGPT-4Granite 3.0 8b Instruct
LMArena Instruction Following12411088

Long Context GPT-4 leads

GPT-4: 37.7 (#207), Granite 3.0 8b Instruct: 34.0 (#247)

BenchmarkGPT-4Granite 3.0 8b Instruct
LMArena Longer Query12441122

Writing & Preference GPT-4 leads

GPT-4: 35.0 (#263), Granite 3.0 8b Instruct: 31.2 (#280)

BenchmarkGPT-4Granite 3.0 8b Instruct
LMArena Text12631096
LMArena Creative Writing12441071
LMArena Multi-Turn12571063
EQ-Bench Creative Writing752—

Frequently asked questions

Is GPT-4 better than Granite 3.0 8b Instruct?

Granite 3.0 8b Instruct is the stronger model overall, scoring 31.1 to 28.9 on the Noometry Index.

Is GPT-4 or Granite 3.0 8b Instruct better for coding?

GPT-4 scores higher on coding benchmarks: 31.2 versus 29.1 in the Noometry coding category.

How many benchmarks do GPT-4 and Granite 3.0 8b Instruct share?

14 benchmarks have published results for both models. GPT-4 has 38 scored results on Noometry and Granite 3.0 8b Instruct has 14.

Related comparisons

Go deeper