in the loop

GPT-5.4 Mini
#62 of 274 scored.

OpenAIMar 17, 2026Closed

Capabilities Index

Range 147.1–150.3

148.8

Context

Takes file, image, text

400K

Input

Per 1M tokens

$0.75

Output

Per 1M tokens

$4.50

Benchmarks

How it scores.

  • GPQA Diamond82.5%

    57th of 186

  • FrontierMath51.2%

    45th of 81

  • ARC-AGI-218.9%

    38th of 79

  • SimpleQA Verified29.4%

    61st of 77

  • Mock AIME88.9%

    54th of 176

Everything else Epoch has for it

  • DTBench66.7%
  • ARC-AGI63.7%
  • WeirdML60.3%
  • LMCA48.0%
  • DeepResearch Bench36.3%
  • FrontierCode27.0%
  • ProofBench21.0%
  • Chess Puzzles20.0%
  • FrontierMath-Tier-4-v2-Private9.8%
  • Mystery Game Puzzles2.0%

On the index

Around it.

  1. #59Qwen 3.6 Max (Preview)Alibaba149.2
  2. #60Grok 4.3 BetaxAI149.2
  3. #61DeepSeek-V4-ProDeepSeek · $0.21 in149.1
  4. #62GPT-5.4 MiniOpenAI · $0.75 in148.8
  5. #63InklingThinking Machines · $1.00 in148.5
  6. #64Kimi K2.5Moonshot · $0.45 in148.0
  7. #65Qwen 3.6 PlusAlibaba · $0.33 in147.7

Sources