priorsBuilder

OpenAI · released 2025-08-05

gpt-oss-20b

Measured on 48 benchmarks.

General frontier

Not ranked in this mix — too little measured data for its benchmarks.

Your mix’s benchmarks

2 of 8 measured

reasoning

Humanity's Last Exam

587 results
  1. #295high of 58711.0%
  2. #425low of 5875.3%

Leader: Opus 5.5 · 61.4%

knowledge

AA-Omniscience

526 results
  1. #431low of 526-58.5
  2. #463high of 526-63.0

Leader: Opus 5.5 · 46.4

All 48 benchmarks

show 46 more

agentic

AA-Briefcase v1.1

224 results
  1. #215high of 2240

Leader: Sonnet 5.5 · 1823

agentic

APEX-Agents-AA

33 results
  1. #33high of 330.7%

Leader: Gemini 3.5 Flash · 47.1%

agentic

AutomationBench-AA

219 results
  1. #214high of 2190.2%

Leader: Gemini 4 · 77.5%

agentic

Terminal-Bench 2.0

40 results
  1. #39high of 4017.0%

Leader: GPT-5.5 · 84.7%

agentic

τ²-Bench Telecom

393 results
  1. #183high of 39360.2%
  2. #199low of 39350.3%

Leader: GLM 5.2 · 99.1%

agentic

τ³-Banking

206 results
  1. #165high of 2067.0%

Leader: Qwen 3.8 Max · 51.3%

coding

LiveCodeBench

282 results
  1. #34high of 28277.7%
  2. #86low of 28265.2%

Leader: Gemini 3 Pro · 91.7%

coding

LiveCodeBench (Vals)

127 results
  1. #70high of 12780.4%

Leader: Fable 5.1 · 90.5%

coding

SciCode

226 results
  1. #178high of 22638.9%

Leader: Opus 5.5 · 66.9%

coding

Terminal-Bench Hard-AA

387 results
  1. #223high of 38710.6%
  2. #268low of 3874.5%

Leader: GPT-5.6 Sol · 65.9%

coding

Terminal-Bench v2.1-AA

238 results
  1. #189high of 23813.9%

Leader: Fable 5.1 · 91.4%

coding

Terminal-Bench v4-AA

216 results
  1. #197high of 2160.0%

Leader: Sonnet 5.5 · 63.6%

coding

WeirdML v2

127 results
  1. #86high of 12740.9%

Leader: GPT-6 Astra Pro · 93.6%

composite

AA Intelligence Index v4.3.2

610 results
  1. #373low of 61010.0
  2. #403high of 6109.0

Leader: Opus 5.5 · 57.6

composite

Arena Text

220 results
  1. #176high of 2201317

Leader: Gemini 4 · 1525

composite

Arena Text — Multi-Turn

220 results
  1. #185high of 2201289

Leader: Gemini 4 · 1553

cost

AA Intelligence Index Cost per Task

177 results
  1. #8high of 177$0.01

Leader: GPT-6 Luna · $0.00

cost

Cost to Run AA Intelligence Index

177 results
  1. #11high of 177$31.41

Leader: GPT-6 Luna · $10.63

economics

CorpFin (Vals)

117 results
  1. #91high of 11753.1%

Leader: Opus 5 · 73.2%

economics

GDP.pdf

210 results
  1. #184high of 2102.0%

Leader: Muse Wonder Full (Max) · 32.6%

economics

GDPval-AA v2.1

288 results
  1. #233high of 288339

Leader: Opus 5.5 · 1866

economics

TaxEval (Vals)

125 results
  1. #107high of 12563.7%

Leader: Muse Spark 1.2 · 80.4%

instruction

IFBench

399 results
  1. #103high of 39965.1%
  2. #133low of 39957.8%

Leader: Grok 4.3 · 83.3%

knowledge

AA-Omniscience Accuracy

526 results
  1. #380high of 52616.0%
  2. #402low of 52615.1%

Leader: Fable 5.1 · 67.2%

knowledge

AA-Omniscience Non-hallucination

526 results
  1. #354low of 52613.3%
  2. #486high of 5265.9%

Leader: MiniCPM5-1B (Non-reasoning) · 99.1%

knowledge

CaseLaw (Vals)

53 results
  1. #53high of 5343.8%

Leader: Grok 4.3 · 79.3%

knowledge

Global-MMLU-Lite

118 results
  1. #90high of 11875.2%
  2. #96low of 11872.8%

Leader: Gemini 3.1 Pro · 93.2%

knowledge

GPQA Diamond

539 results
  1. #299high of 53968.8%
  2. #353low of 53961.1%

Leader: GPT-6 Astra · 96.3%

knowledge

GPQA Diamond (Vals)

122 results
  1. #92high of 12268.9%

Leader: Gemini 3.1 Pro · 95.5%

knowledge

MedQA (Vals)

80 results
  1. #55high of 8082.9%

Leader: o1 · 96.5%

knowledge

MMLU Pro (Vals)

122 results
  1. #105high of 12271.6%

Leader: Fable 5.1 · 92.4%

knowledge

MMLU-Pro

285 results
  1. #150high of 28574.8%
  2. #169low of 28571.8%

Leader: Gemini 3 Pro · 89.8%

long-context

AA-LCR v1.1

523 results
  1. #363high of 52334.7%
  2. #376low of 52331.0%

Leader: Kimi K3 · 88.7%

long-context

MLCR-AA

94 results
  1. #90high of 940.6%

Leader: Sonnet 5.5 · 75.0%

math

AIME (Vals)

82 results
  1. #34high of 8286.0%

Leader: Gemini 3.1 Pro · 98.1%

math

AIME 2025

228 results
  1. #30high of 22889.3%
  2. #98low of 22862.3%

Leader: GPT-5.2 · 99.0%

math

MATH 500 (Vals)

48 results
  1. #12high of 4894.2%

Leader: Gemini 3 Pro · 96.4%

math

MGSM (Vals)

62 results
  1. #43high of 6289.0%

Leader: Opus 4.5 · 95.2%

reasoning

CritPt

529 results
  1. #214high of 5291.4%
  2. #384low of 5290.0%

Leader: Muse Wonder Full (Max) · 32.9%

reasoning

LegalBench

135 results
  1. #111high of 13570.8%

Leader: Fable 5 · 88.6%

reliability

Tool-call error rate

171 results
  1. #135high of 1714.6%

Leader: GPT-5.4 Pro · 0.0%

reliability

Uptime (7d)

181 results
  1. #126high of 18198.7%

Leader: GPT-5 Pro · 100.0%

speed

Latency (Time to First Token)

351 results
  1. #25low of 3510.7s
  2. #27high of 3510.8s

Leader: Gemini 2.5 Flash-Lite · 0.3s

speed

Output Speed (tokens/s)

351 results
  1. #13low of 351302.8
  2. #28high of 351223.4

Leader: Celeris-1 · 1522.5

speed

Provider latency

177 results
  1. #36high of 1770.7s

Leader: Qwen 2.5 7B · 0.2s

speed

Provider throughput

176 results
  1. #38high of 17696.0

Leader: Muse Spark 1.1 · 204.3