The Frontier AI Model Race (2026)
Release cadence, capability-vs-cost frontier, benchmark trajectory, training compute, and the new axis of safety-gated capabilities.
Each point is a frontier model: y = benchmark score, x = $/M tokens.
View data & sources →Benchmark trajectory by lab
Composite benchmark score over time, leading labs.
Estimated training compute per frontier release (log scale, FLOPs).
View data & sources →For each model × domain: allowed (green) / safety-gated (amber) / blocked (red).
View data & sources →Data table
| meta | openai | series | quarter | anthropic | source_ref | value_basis | model | benchmark_score | cost_per_m_tokens | year | flops_estimate | bio | chem | cyber | legal | medical | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | 0 | 1 | releases | 2024 Q2 | 1 | anthropic-news | Claude 3.5 Sonnet (Jun 2024); GPT-4o (May 2024) — Anthropic + OpenAI release notes | ||||||||||
| 1 | 0 | 1 | releases | 2024 Q3 | 0 | openai-research | OpenAI o1-preview (Sep 2024); Meta Llama 3.1 405B (Jul 2024) | ||||||||||
| 0 | 1 | 1 | releases | 2024 Q4 | 0 | google-deepmind | Google Gemini 2.0 (Dec 2024); OpenAI o1/o3 (Dec 2024) | ||||||||||
| 0 | 0 | 0 | releases | 2025 Q1 | 1 | anthropic-news | Claude 3.7 Sonnet (Feb 2025) | ||||||||||
| 1 | 0 | 0 | releases | 2025 Q2 | 1 | meta-ai | Meta Llama 4 (Apr 2025); Anthropic Claude 4 / Opus 4 (May 2025) |
16 more rows + CSV download
The full 21-row dataset, one-click CSV export, and the AI-ready context file are free with an account. Prefer to verify it yourself? The full methodology and sources are published below.
Methodology & sources
Last updated: Jul 17, 2026Methodology
Source-backed values are seeded for four of the five charts: the release cadence by lab (2024 → 2026, from each lab’s release notes), the capability-vs-cost scatter (Artificial Analysis Intelligence Index vs output $/M tokens, June 2026 snapshot), training-compute growth by year (Epoch AI, corroborated by Stanford HAI), and a safety-gated capabilities matrix. Every numeric point carries a sources[].ref and a value_basis. Sources are each lab’s own model cards/release notes, Artificial Analysis, Epoch AI, and Stanford HAI — cross-checked against public release timelines. EDITORIAL ENCODING: the safety-gated matrix scores each model × domain as 3 = allowed / 2 = gated / 1 = blocked. This is an interpretation of each model’s published safety policy, not a measured benchmark. PLACEHOLDER: the per-lab benchmark-trajectory chart is left unseeded — a consistent historical per-quarter, per-lab benchmark series was not sourceable without mixing incompatible benchmarks. Re-verified 2026-06-15.
Sources
- Anthropic Model Cards & Release Notes ↗ Publicly available
- OpenAI Research Blog ↗ Publicly available
- Google DeepMind Research ↗ Publicly available
- Meta AI — Llama release notes ↗ Publicly available
- LMArena Leaderboard ↗ Open access
- Epoch AI — compute & training trends ↗ Open access
- Stanford HAI — AI Index Report ↗ CC BY-ND 4.0 — attribution required
Comparisons are informative, not definitive. See each source for definitions and limits.