Skip to content
aiqbenchAI & Quantum Intelligence
NewsInsightsBenchmarksSkillsAsk AILearnAbout
...Loading your session
NewsInsightsBenchmarksSkillsAsk AILearnAbout
Back to list
aiqbench takeAI Research

aiqbench analysis

This article analyzes the longevity of GPT-4's dominance on the Epoch Capabilities Index, contrasting its one-year lead with shorter tenures by subsequent models like o1 and various Gemini or Claude iterations. The data highlights a significant shift in model capability trajectories, noting that no model since GPT-4 has maintained a leading position for as extended a period.
1
Source coveringEpoch AI

GPT-4 led in ECI far longer than any other model

Epoch AI·July 2, 2026·30 min read

From Epoch AI

Excerpt

OpenAI’s GPT-4 topped the Epoch Capabilities Index (ECI), our aggregate measure of language model capability, for roughly a year after its release in March 2023. No model since has led for as long. The second-longest lead, by OpenAI’s o1, lasted a little over three months, less than a third of GPT-4’s.

gpt-5.4-pro-2026-03-05_xhigh

GPT-5.4 Pro (xhigh)

gpt-5.2-pro-2025-12-11_xhigh

GPT-5.2 Pro (xhigh)

gpt-5.5-pro_xhigh

GPT-5.5 Pro (xhigh)

A smaller version of OpenAI's latest flagship frontier model, benchmarked using high reasoning effort.

gpt-5.4-nano-2026-03-17_high

GPT-5.4 nano (high)

The smallest version of OpenAI's latest flagship frontier model, benchmarked using high reasoning effort.

gpt-5.4-mini-2026-03-17_xhigh

GPT-5.4 mini (xhigh)

Gemini 2.5 Pro (Jun 2025)

Google’s general-purpose frontier reasoning model.

gemini-3.1-pro-preview

Gemini 3.1 Pro Preview

Anthropic’s largest flagship model, benchmarked with up to 32,000 reasoning tokens.

Claude Opus 4.1 (32k thinking)

Claude Sonnet 4.5

Claude Sonnet 4.5 (32k thinking)

Anthropic's latest mid-size model, part of the Claude 4.5 series, evaluated with up to 32,000 reasoning tokens.

gpt-5.4-2026-03-05_xhigh

gpt-5.2-2025-12-11_xhigh

The latest smaller size o-series reasoning model from OpenAI, evaluated with high reasoning effort.

A smaller version of OpenAI's o3 reasoning model, evaluated with high reasoning effort.

Claude Opus 4.5 (32k thinking)

OpenAI's latest flagship frontier model, benchmarked using high reasoning effort.

Excerpt shown under the publisher's RSS terms · 199 words · full text and figures are on the publisher's site

Read at Epoch AI ↗Ask about this article

Related coverage

AI Safety

Q3.5 Timelines Update: Looks Like the Intelligence Explosion Will Happen in 2027 After All

AI Futures Project · Nov 10, 2026

Links #3: 2026/06 Part 1

LessWrong · Jun 15, 2026

The Case for MAPLE

LessWrong · Nov 15, 2026

2026.10: You are here 📍

Home Assistant · Oct 7, 2026

Article info

Source

Epoch AI

Published

July 2, 2026

Read time

30 min

Category

AI Research
aiqbench
AI & Quantum Intelligence© aiqbench. All rights reserved.

Feeds:RSS·JSON Feed

LanguageEnglish·Românăbeta
NewsAI & TechQuantumInsights
ExploreBenchmarksLearn
CompanyAboutTrustLegal noticeTerms of ServicePrivacy PolicyCopyright noticesData erasureReport content

The aiqbench take, the insights, the daily digest and Ask answers are written by AI models from the sources shown, without editorial review before publication. They may contain errors — check the source. How the model is used →

  • News
  • Insights
  • Ask AI
  • Benchmarks
  • Account