Noometry

Model comparison

GPT-3.5-turbo vs StarCoder 2 15B

GPT-3.5-turbo has enough public results to be ranked (#350); StarCoder 2 15B does not yet, so treat this comparison as directional.

Last verified . 7 shared benchmarks.

GPT-3.5-turbo OpenAI

23.2

Rank #350 Confirmed

StarCoder 2 15B NVIDIA

38.1

Unranked Sparse

Summary

  • They share 7 benchmarks with published results for both. GPT-3.5-turbo scores higher in 0 categories and StarCoder 2 15B in 1 category; one gap is clear of the uncertainty.
  • The widest gap is in coding, where StarCoder 2 15B leads 36.6 to 23.9.
  • The biggest single-benchmark swing is BigCodeBench Complete: 50.6% for GPT-3.5-turbo and 38.4% for StarCoder 2 15B.
  • StarCoder 2 15B has downloadable open weights; the other is API-only.

Side by side

GPT-3.5-turboStarCoder 2 15B
ProviderOpenAINVIDIA
Noometry Index23.238.1
Released2023-03-012024-02-20
WeightsProprietaryOpen
Context window16K—
Max output4K—
Input $ / M tokens$0.50—
Output $ / M tokens$1.50—
Results tracked447

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

Category by category

Coding StarCoder 2 15B leads

GPT-3.5-turbo: 23.9 (#331), StarCoder 2 15B: 36.6

BenchmarkGPT-3.5-turboStarCoder 2 15B
BigCodeBench Complete50.6%38.4%
HumanEval+70.7%37.8%
WeirdML3.5%—
BigCodeBench Instruct39.1%—
LMArena Coding1136—
MBPP+69.7%—

Agentic & Tool Use Not comparable

GPT-3.5-turbo: —, StarCoder 2 15B: —

BenchmarkGPT-3.5-turboStarCoder 2 15B
METR Time Horizons21.5%—

Reasoning Not comparable

GPT-3.5-turbo: 13.8 (#332), StarCoder 2 15B: —

BenchmarkGPT-3.5-turboStarCoder 2 15B
Epoch Capabilities Index118.55105
WinoGrande81.6%64.3%
Chess Puzzles0%—
LMArena Hard Prompts1108—
Mystery Game Puzzles3%—
DTBench48.5%—
LMCA9.7%—
Adversarial NLI58.1%—
BIG-Bench Hard61.6%—
CommonsenseQA 2.057%—
ForecastBench50.4—

Math Not comparable

GPT-3.5-turbo: 6.3 (#327), StarCoder 2 15B: —

BenchmarkGPT-3.5-turboStarCoder 2 15B
GSM8K57.8%57.7%
FrontierMath (Tiers 1-3)0%—
OTIS Mock AIME 2024-20252.2%—
LMArena Math1142—
MATH Level 515.9%—

Knowledge Not comparable

GPT-3.5-turbo: 10.0 (#303), StarCoder 2 15B: —

BenchmarkGPT-3.5-turboStarCoder 2 15B
ARC (AI2) Challenge87.4%47.2%
MMLU71.4%64.1%
GPQA Diamond28%—
LMArena Expert1070—
BoolQ87%—
OpenBookQA86%—
TriviaQA85.8%—

Multilingual Not comparable

GPT-3.5-turbo: 31.5 (#258), StarCoder 2 15B: —

BenchmarkGPT-3.5-turboStarCoder 2 15B
LMArena Non-English1108—
LMArena Chinese1075—
LMArena French1118—
LMArena German1090—
LMArena Japanese1043—
LMArena Korean1019—
LMArena Russian1123—
LMArena Spanish1121—

Instruction Following Not comparable

GPT-3.5-turbo: 57.9 (#262), StarCoder 2 15B: —

BenchmarkGPT-3.5-turboStarCoder 2 15B
LMArena Instruction Following1119—

Long Context Not comparable

GPT-3.5-turbo: 34.0 (#254), StarCoder 2 15B: —

BenchmarkGPT-3.5-turboStarCoder 2 15B
LMArena Longer Query1121—

Writing & Preference Not comparable

GPT-3.5-turbo: 25.3 (#305), StarCoder 2 15B: —

BenchmarkGPT-3.5-turboStarCoder 2 15B
LMArena Text1125—
LMArena Creative Writing1092—
EQ-Bench Creative Writing451—
LMArena Multi-Turn1117—

Frequently asked questions

Is GPT-3.5-turbo better than StarCoder 2 15B?

GPT-3.5-turbo has enough public results to be ranked (#350); StarCoder 2 15B does not yet, so treat this comparison as directional.

Is GPT-3.5-turbo or StarCoder 2 15B better for coding?

StarCoder 2 15B scores higher on coding benchmarks: 36.6 versus 23.9 in the Noometry coding category.

How many benchmarks do GPT-3.5-turbo and StarCoder 2 15B share?

7 benchmarks have published results for both models. GPT-3.5-turbo has 44 scored results on Noometry and StarCoder 2 15B has 7.

Related comparisons

Go deeper