Frontier
The AI frontier over time
The single best-scoring model at every release across 2023 – 2026 — by the Artificial Analysis Intelligence Index — and how the labs leapfrog each other for the lead. The current best is Claude Fable 5 (Anthropic, 59.9).
OpenAI led 11×
Anthropic led 10×
Google led 1×
Every time the frontier moved
Newest first. Each model set a new all-time-high Intelligence Index at its release.
- Claude Fable 5 Anthropic 59.9 Jun 2026
- Claude Opus 4.8 Anthropic · took the lead 55.7 May 2026
- GPT-5.5 OpenAI · took the lead 54.8 Apr 2026
- Claude Opus 4.7 Anthropic · took the lead 53.5 Apr 2026
- GPT-5.4 OpenAI · took the lead 51.4 Mar 2026
- Claude Sonnet 4.6 Anthropic · took the lead 47.2 Feb 2026
- GPT-5.3-codex OpenAI · took the lead 44.3 Feb 2026
- Claude Opus 4.6 Anthropic · took the lead 43.7 Feb 2026
- GPT-5.2 OpenAI · took the lead 42.2 Dec 2025
- Claude Opus 4.5 Anthropic · took the lead 40.8 Nov 2025
- Gemini 3 Pro Preview Google · took the lead 39.6 Nov 2025
- GPT-5.1 OpenAI 38.9 Nov 2025
- GPT-5 OpenAI · took the lead 36.1 Aug 2025
- Claude Opus 4.1 Anthropic · took the lead 33.7 Aug 2025
- o3-pro OpenAI 32.5 Jun 2025
- o3 OpenAI · took the lead 30.4 Apr 2025
- Claude 3.7 Sonnet Anthropic · took the lead 27.1 Feb 2025
- o1 OpenAI · took the lead 23.4 Dec 2024
- Claude 3.5 Haiku Anthropic 12.3 Nov 2024
- Claude 3 Opus Anthropic · took the lead 11.8 Mar 2024
- GPT-4 OpenAI 7.0 Jun 2023
- GPT-3.5-turbo OpenAI · took the lead 3.6 Feb 2023
Know the moment the frontier moves.
Set up alerts
Get alerted on every new frontier model — free, double opt-in, on email or Telegram.