Noometry

Model comparison

GPT-3.5-turbo vs Phi-4

Phi-4 is the stronger model overall, scoring 31.2 to 23.2 on the Noometry Index.

Last verified . 25 shared benchmarks.

GPT-3.5-turbo OpenAI

23.2

Rank #350 Confirmed

Phi-4 Microsoft

31.2

Rank #279 Confirmed

Summary

  • They share 25 benchmarks with published results for both. GPT-3.5-turbo scores higher in 0 categories and Phi-4 in 8 categories; 8 gaps are clear of the uncertainty.
  • The widest gap is in knowledge, where Phi-4 leads 32.6 to 10.0.
  • The biggest single-benchmark swing is MATH Level 5: 15.9% for GPT-3.5-turbo and 64.9% for Phi-4.
  • Phi-4 is cheaper at $0.07 / $0.14 per million input/output tokens, against $0.50 / $1.50 for GPT-3.5-turbo.
  • Phi-4 accepts more context: 128K tokens versus 16K.
  • Phi-4 has downloadable open weights; the other is API-only.

Side by side

GPT-3.5-turboPhi-4
ProviderOpenAIMicrosoft
Noometry Index23.231.2
Released2023-03-012024-12-11
WeightsProprietaryOpen
Context window16K128K
Max output4K4K
Input $ / M tokens$0.50$0.07
Output $ / M tokens$1.50$0.14
Results tracked4437

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

Category by category

Coding Phi-4 leads

GPT-3.5-turbo: 23.9 (#331), Phi-4: 34.4 (#239)

BenchmarkGPT-3.5-turboPhi-4
BigCodeBench Instruct39.1%45.5%
LMArena Coding11361231
BigCodeBench Complete50.6%55.4%
WeirdML3.5%—
LiveBench Coding—30.7%
HumanEval+70.7%—
MBPP+69.7%—

Agentic & Tool Use Not comparable

GPT-3.5-turbo: —, Phi-4: 22.8 (#128)

BenchmarkGPT-3.5-turboPhi-4
Berkeley Function Calling Leaderboard—28.8%
BALROG—11.6%
METR Time Horizons21.5%—

Reasoning Phi-4 leads

GPT-3.5-turbo: 13.8 (#332), Phi-4: 17.7 (#291)

BenchmarkGPT-3.5-turboPhi-4
Chess Puzzles0%1%
LMArena Hard Prompts11081220
Epoch Capabilities Index118.55130.42
LiveBench Reasoning—47.8%
Mystery Game Puzzles3%—
DTBench48.5%—
LiveBench Data Analysis—45.2%
LMCA9.7%—
Adversarial NLI58.1%—
BIG-Bench Hard61.6%—
CommonsenseQA 2.057%—
ForecastBench50.4—
LiveBench—41.6%
WinoGrande81.6%—

Math Phi-4 leads

GPT-3.5-turbo: 6.3 (#327), Phi-4: 20.8 (#285)

BenchmarkGPT-3.5-turboPhi-4
OTIS Mock AIME 2024-20252.2%13.8%
LMArena Math11421246
MATH Level 515.9%64.9%
FrontierMath (Tiers 1-3)0%—
LiveBench Math—42%
GSM8K57.8%—

Knowledge Phi-4 leads

GPT-3.5-turbo: 10.0 (#303), Phi-4: 32.6 (#209)

BenchmarkGPT-3.5-turboPhi-4
GPQA Diamond28%56.1%
LMArena Expert10701203
MMLU71.4%84.8%
Confabulations—29.4%
Vectara Hallucination Rate—3.7%
ARC (AI2) Challenge87.4%—
BoolQ87%—
OpenBookQA86%—
TriviaQA85.8%—

Multilingual Phi-4 leads

GPT-3.5-turbo: 31.5 (#258), Phi-4: 37.2 (#237)

BenchmarkGPT-3.5-turboPhi-4
LMArena Non-English11081197
LMArena Chinese10751212
LMArena French11181224
LMArena German10901222
LMArena Japanese10431158
LMArena Korean10191151
LMArena Russian11231209
LMArena Spanish11211234

Instruction Following Phi-4 leads

GPT-3.5-turbo: 57.9 (#262), Phi-4: 60.4 (#251)

BenchmarkGPT-3.5-turboPhi-4
LMArena Instruction Following11191201
LiveBench Instruction Following—58.4%

Long Context Phi-4 leads

GPT-3.5-turbo: 34.0 (#254), Phi-4: 36.9 (#226)

BenchmarkGPT-3.5-turboPhi-4
LMArena Longer Query11211217

Writing & Preference Phi-4 leads

GPT-3.5-turbo: 25.3 (#305), Phi-4: 40.5 (#244)

BenchmarkGPT-3.5-turboPhi-4
LMArena Text11251217
LMArena Creative Writing10921182
LMArena Multi-Turn11171206
Short-Story Creative Writing—62.6%
EQ-Bench Creative Writing451—
LiveBench Language—25.6%

Frequently asked questions

Is GPT-3.5-turbo better than Phi-4?

Phi-4 is the stronger model overall, scoring 31.2 to 23.2 on the Noometry Index.

Which is cheaper, GPT-3.5-turbo or Phi-4?

Phi-4 is cheaper. It lists at $0.07 per million input tokens and $0.14 per million output tokens; GPT-3.5-turbo lists at $0.50 and $1.50.

Is GPT-3.5-turbo or Phi-4 better for coding?

Phi-4 scores higher on coding benchmarks: 34.4 versus 23.9 in the Noometry coding category.

Which has the bigger context window?

Phi-4 does, with 128K tokens against 16K.

How many benchmarks do GPT-3.5-turbo and Phi-4 share?

25 benchmarks have published results for both models. GPT-3.5-turbo has 44 scored results on Noometry and Phi-4 has 37.

Related comparisons

Go deeper