The leading edge of AI capability, and the models and laboratories at it. Chapter 1.
From The AI Compute Race. Read the book
Where the book uses it
- Chapter 1 — The Stakes
What the race is for: the belief that a lead in AI can compound into a lasting advantage, how long such leads have actually lasted, and what a trillion dollars a year is being bet against.
Where the record uses it
- Campus Stargate Shackelford / Frontier
- Scorecard indicator Post-2024 METR horizon doubling time
- Scorecard indicator China-to-US frontier lag
- Dated test Chapter 7, the frontier’s deployment, due 31 December 2027
- Dated test Chapters 1 and 12, the first route, due 31 August 2028
- Dated test Chapter 2, compute moved rather than stopped, due 31 March 2029
Sources that use it
- Google DeepMind, Frontier Safety Framework 3.0, 22 September 2025 (“We recommend Security level 4 for this capability threshold [machine-learning R&D automation level 1], but emphasize that this must be taken on by the frontier AI field as a whole”; a 3.1 revision followed; “In practice, our overall security posture may commonly exceed the baseline levels recommended here”). In the Sources
- Google DeepMind, Frontier Safety Framework 3.1, published 17 April 2026 (“We recommend Security level 4 for this capability threshold, but emphasize that this must be taken on by the frontier AI field as a whole”, p. 15, for ML R&D automation level 1; “In practice, our overall security posture may commonly exceed the baseline levels recommended here”, p. 11; the version’s change log, p. 17, lists CBRN thresholds, a combined ML R&D and Misalignment risk domain and Security Level 2+ for the CBRN, cyber and harmful-manipulation levels, and does not list the level-4 recommendation among its changes). In the Sources
- Moonshot AI, “Kimi K3 Tech Blog: Open Frontier Intelligence,” Kimi blog, July 2026 (“Pricing is $0.30/MTok for cache-hit input, $3.00/MTok for cache-miss input, and $15.00/MTok for output”, that is, $3 and $15 per million tokens of input and output; the post prints no date of its own, gives its benchmark scores as “current as of July 16, 2026” and promises the full weights “by July 27, 2026”; Moonshot’s API price card, “Model Inference Pricing Explanation,” Kimi API Platform, listed the same $0.30, $3.00 and $15.00 for kimi-k3 on 20 September 2026, with cache-write prices of $3.00 and $6.00 that the post does not give; read live 20 September 2026). In the Sources
- NVIDIA, “NVIDIA Vera Rubin Opens Agentic AI Frontier,” keynote press release, 16 March 2026 (it does not carry the live blog’s figure — its three uses of “trillion” are all “trillion-parameter”). In the Sources
- OpenAI, “GPT-5.6: Frontier intelligence that scales with your ambition,” 9 July 2026, with the post’s update lines of 30 July and 21 August 2026 (“Sol is $5 input / $30 output”; “Update on August 21, 2026: OpenAI dropped the API and credit pricing of GPT-5.6 Sol by over 20% for the next 3 months”; the launch page’s multi-agent chart, SEC-Bench Pro at the max reasoning setting, read from its embedded spec on 10 October 2026: 71.4% at $15.47 a task with one agent, 74.3% at $32.41 with four, 76.2% at $65.86 with sixteen). In the Sources
And 19 more: search the Sources.
Related terms
From the glossary of The AI Compute Race.
How to cite this page
Dryburgh, Lee S. “Frontier.” In The AI Compute Race. DRYBURGH. https://dryburgh.com/vol1/glossary/frontier (accessed 2026-11-01).
BibTeX
@misc{dryburgh-frontier-2026,
author = {Dryburgh, Lee S.},
title = {{Frontier}},
howpublished = {In \emph{The AI Compute Race}, DRYBURGH},
year = {2026},
url = {https://dryburgh.com/vol1/glossary/frontier},
note = {Accessed 2026-11-01}
}