Noometry

Model comparison

GPT-4 vs Tulu 3 (Tülu 3) 70B

Tulu 3 (Tülu 3) 70B is the stronger model overall, scoring 33.0 to 29.1 on the Noometry Index.

Last verified . 14 shared benchmarks.

GPT-4 OpenAI

29.1

Rank #316 Confirmed

Tulu 3 (Tülu 3) 70B Allen Institute for AI (Ai2)

33.0

Rank #251 Confirmed

Summary

  • They share 14 benchmarks with published results for both. GPT-4 scores higher in 3 categories and Tulu 3 (Tülu 3) 70B in 5 categories; 5 gaps are clear of the uncertainty.
  • The widest gap is in writing & preference, where Tulu 3 (Tülu 3) 70B leads 45.6 to 34.9.
  • The biggest single-benchmark swing is MATH Level 5: 23% for GPT-4 and 42.7% for Tulu 3 (Tülu 3) 70B.
  • Tulu 3 (Tülu 3) 70B has downloadable open weights; the other is API-only.

Side by side

GPT-4Tulu 3 (Tülu 3) 70B
ProviderOpenAIAllen Institute for AI (Ai2)
Noometry Index29.133.0
Released2023-03-142024-11-21
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 Tulu 3 (Tülu 3) 70B leads

GPT-4: 31.6 (#283), Tulu 3 (Tülu 3) 70B: 36.0 (#214)

BenchmarkGPT-4Tulu 3 (Tülu 3) 70B
LMArena Coding12541235
WeirdML12.4%—
BigCodeBench Instruct46%—
BigCodeBench Complete57.2%—
HumanEval+79.3%—

Agentic & Tool Use Not comparable

GPT-4: —, Tulu 3 (Tülu 3) 70B: —

BenchmarkGPT-4Tulu 3 (Tülu 3) 70B
METR Time Horizons36.1%—

Reasoning Tulu 3 (Tülu 3) 70B leads

GPT-4: 17.8 (#289), Tulu 3 (Tülu 3) 70B: 23.9 (#169)

BenchmarkGPT-4Tulu 3 (Tülu 3) 70B
LMArena Hard Prompts12411220
Chess Puzzles4%—
Mystery Game Puzzles12%—
DTBench62.7%—
LMCA17.1%—
BIG-Bench Hard75.1%—
Epoch Capabilities Index125.89—
ForecastBench57.8—
HellaSwag95.3%—
WinoGrande87.5%—

Math Tulu 3 (Tülu 3) 70B leads

GPT-4: 10.8 (#309), Tulu 3 (Tülu 3) 70B: 14.2 (#303)

BenchmarkGPT-4Tulu 3 (Tülu 3) 70B
OTIS Mock AIME 2024-20251.1%4.4%
LMArena Math12691242
MATH Level 523%42.7%
GSM8K92%—

Knowledge Tulu 3 (Tülu 3) 70B leads

GPT-4: 18.4 (#282), Tulu 3 (Tülu 3) 70B: 25.0 (#264)

BenchmarkGPT-4Tulu 3 (Tülu 3) 70B
GPQA Diamond35.7%46.3%
LMArena Expert1211—
MMLU86.4%—
TriviaQA84.8%—

Multilingual Too close to call

GPT-4: 40.6 (#215), Tulu 3 (Tülu 3) 70B: 39.9 (#222)

BenchmarkGPT-4Tulu 3 (Tülu 3) 70B
LMArena Non-English12461236
LMArena Chinese12421249
LMArena Russian12511246
LMArena French1283—
LMArena German1251—
LMArena Japanese1209—
LMArena Korean1184—
LMArena Spanish1261—

Instruction Following Too close to call

GPT-4: 65.3 (#222), Tulu 3 (Tülu 3) 70B: 64.8 (#227)

BenchmarkGPT-4Tulu 3 (Tülu 3) 70B
LMArena Instruction Following12411233

Long Context Too close to call

GPT-4: 37.7 (#212), Tulu 3 (Tülu 3) 70B: 37.1 (#222)

BenchmarkGPT-4Tulu 3 (Tülu 3) 70B
LMArena Longer Query12441224

Writing & Preference Tulu 3 (Tülu 3) 70B leads

GPT-4: 34.9 (#268), Tulu 3 (Tülu 3) 70B: 45.6 (#223)

BenchmarkGPT-4Tulu 3 (Tülu 3) 70B
LMArena Text12631256
LMArena Creative Writing12441231
LMArena Multi-Turn12571252
EQ-Bench Creative Writing752—

Frequently asked questions

Is GPT-4 better than Tulu 3 (Tülu 3) 70B?

Tulu 3 (Tülu 3) 70B is the stronger model overall, scoring 33.0 to 29.1 on the Noometry Index.

Is GPT-4 or Tulu 3 (Tülu 3) 70B better for coding?

Tulu 3 (Tülu 3) 70B scores higher on coding benchmarks: 36.0 versus 31.6 in the Noometry coding category.

How many benchmarks do GPT-4 and Tulu 3 (Tülu 3) 70B share?

14 benchmarks have published results for both models. GPT-4 has 38 scored results on Noometry and Tulu 3 (Tülu 3) 70B has 14.

Related comparisons

Go deeper