in the loop

Gemini 3 Pro
#43 of 274 scored.

GoogleNov 18, 2025Closed

Capabilities Index

Range 150.9–155.1

152.9

Context

Tokens

—

Input

Per 1M tokens

—

Output

Per 1M tokens

—

Benchmarks

How it scores.

  • GPQA Diamond90.1%

    22nd of 186

  • SWE-bench Verified72.9%

    21st of 32

  • ARC-AGI-231.1%

    36th of 79

  • Humanity's Last Exam34.4%

    7th of 40

  • Mock AIME91.4%

    47th of 176

  • Terminal-Bench69.4%

    7th of 35

Everything else Epoch has for it

  • VPCT86.5%
  • GeoBench84.0%
  • ARC-AGI75.0%
  • SimpleBench71.7%
  • METR Time Horizons71.0%
  • WeirdML69.9%
  • FrontierMath-2025-02-28-Private66.0%
  • Balrog58.1%
  • DeepResearch Bench46.3%
  • GDPval40.3%
  • FrontierMath-Tier-4-2025-07-01-Private31.3%
  • Chess Puzzles27.4%
  • ProofBench20.0%
  • GSO-Bench18.6%
  • CL-bench15.8%
  • Remote Labor Index1.3%

On the index

Around it.

  1. #40Qwen3.7-MaxAlibaba · $1.48 in153.7
  2. #41Grok 4.7xAI · $2.00 in153.5
  3. #42GPT-5.2OpenAI · $1.75 in153.4
  4. #43Gemini 3 ProGoogle152.9
  5. #44Claude Sonnet 4.6Anthropic · $3.00 in152.2
  6. #45Muse SparkMeta152.0
  7. #46Grok 4.20xAI · $1.25 in152.0

Sources