Noometry

Model comparison

GPT-4 vs gpt-oss-20b

gpt-oss-20b is the stronger model overall, scoring 32.1 to 28.9 on the Noometry Index.

Last verified . 24 shared benchmarks.

GPT-4 OpenAI

28.9

Rank #308 Confirmed

gpt-oss-20b OpenAI

32.1

Rank #249 Confirmed

Summary

  • They share 24 benchmarks with published results for both. GPT-4 scores higher in 1 category and gpt-oss-20b in 7 categories; 6 gaps are clear of the uncertainty.
  • The widest gap is in math, where gpt-oss-20b leads 37.6 to 10.1.
  • The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 1.1% for GPT-4 and 65.3% for gpt-oss-20b.
  • gpt-oss-20b is cheaper at $0.018 / $0.09 per million input/output tokens, against $30 / $60 for GPT-4.
  • gpt-oss-20b accepts more context: 131K tokens versus 8K.
  • gpt-oss-20b has downloadable open weights; the other is API-only.

Side by side

GPT-4gpt-oss-20b
ProviderOpenAIOpenAI
Noometry Index28.932.1
Released2023-03-142025-08-05
WeightsProprietaryOpen
Context window8K131K
Max output8K16K
Input $ / M tokens$30$0.018
Output $ / M tokens$60$0.09
Results tracked3834

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

Category by category

Coding gpt-oss-20b leads

GPT-4: 31.2 (#277), gpt-oss-20b: 37.1 (#185)

BenchmarkGPT-4gpt-oss-20b
WeirdML12.4%40.9%
LMArena Coding12541307
SciCode—34.4%
BigCodeBench Instruct46%—
BigCodeBench Complete57.2%—
ALE-Bench—566.05
HumanEval+79.3%—

Agentic & Tool Use Not comparable

GPT-4: —, gpt-oss-20b: 9.2 (#146)

BenchmarkGPT-4gpt-oss-20b
Terminal-Bench—3.4%
METR Time Horizons36.1%—

Reasoning gpt-oss-20b leads

GPT-4: 17.3 (#285), gpt-oss-20b: 18.8 (#251)

BenchmarkGPT-4gpt-oss-20b
Chess Puzzles4%4%
LMArena Hard Prompts12411274
DTBench62.7%68%
LMCA17.1%14.5%
Epoch Capabilities Index125.88137.81
Kagi LLM Benchmark—53.2%
CritPt—1.4%
Mystery Game Puzzles12%—
BIG-Bench Hard75.1%—
ForecastBench57.8—
HellaSwag95.3%—
WinoGrande87.5%—

Math gpt-oss-20b leads

GPT-4: 10.1 (#304), gpt-oss-20b: 37.6 (#93)

BenchmarkGPT-4gpt-oss-20b
OTIS Mock AIME 2024-20251.1%65.3%
LMArena Math12691317
Omni-MATH—56.5%
MATH Level 523%—
GSM8K92%—

Knowledge gpt-oss-20b leads

GPT-4: 18.4 (#277), gpt-oss-20b: 34.6 (#186)

BenchmarkGPT-4gpt-oss-20b
GPQA Diamond35.7%60.8%
LMArena Expert12111258
MMLU-Pro—74%
GPQA (HELM)—59.4%
MMLU86.4%—
TriviaQA84.8%—

Multilingual gpt-oss-20b leads

GPT-4: 40.8 (#210), gpt-oss-20b: 42.4 (#192)

BenchmarkGPT-4gpt-oss-20b
LMArena Non-English12461269
LMArena Chinese12421311
LMArena German12511249
LMArena Japanese12091245
LMArena Korean11841239
LMArena Russian12511279
LMArena Spanish12611267
LMArena French1282—

Instruction Following GPT-4 leads

GPT-4: 65.5 (#217), gpt-oss-20b: 62.0 (#235)

BenchmarkGPT-4gpt-oss-20b
LMArena Instruction Following12411236
IFEval—73.2%

Long Context Too close to call

GPT-4: 37.7 (#207), gpt-oss-20b: 37.9 (#203)

BenchmarkGPT-4gpt-oss-20b
LMArena Longer Query12441251

Writing & Preference Too close to call

GPT-4: 35.0 (#263), gpt-oss-20b: 35.8 (#260)

BenchmarkGPT-4gpt-oss-20b
LMArena Text12631288
LMArena Creative Writing12441201
EQ-Bench Creative Writing752666
LMArena Multi-Turn12571268
WildBench—73.7%

Frequently asked questions

Is GPT-4 better than gpt-oss-20b?

gpt-oss-20b is the stronger model overall, scoring 32.1 to 28.9 on the Noometry Index.

Which is cheaper, GPT-4 or gpt-oss-20b?

gpt-oss-20b is cheaper. It lists at $0.018 per million input tokens and $0.09 per million output tokens; GPT-4 lists at $30 and $60.

Is GPT-4 or gpt-oss-20b better for coding?

gpt-oss-20b scores higher on coding benchmarks: 37.1 versus 31.2 in the Noometry coding category.

Which has the bigger context window?

gpt-oss-20b does, with 131K tokens against 8K.

How many benchmarks do GPT-4 and gpt-oss-20b share?

24 benchmarks have published results for both models. GPT-4 has 38 scored results on Noometry and gpt-oss-20b has 34.

Related comparisons

Go deeper