Topic · Concept
Frontier models
"Frontier models" are the most capable general-purpose AI models available at a given time, such as the top tiers from Anthropic, OpenAI and Google. The term is used in both industry and regulation, and its meaning moves as the state of the art advances.
Fact file
Frontier Model Forum501(c)(6) nonprofit, established 2023
FMF definitionGeneral-purpose; outperforms models in wide use ≥12 months on benchmarks or high-risk capability assessments
Example: Anthropic top tierClaude Fable 5.1 ($10/$50 per MTok), Claude Opus 5.5 ($4/$20)
Latest on Frontier models
Oct 9Open source · NewsDeepSeek-V4.1-Flash: an MIT-licensed open-weights MoE that DeepSeek says beats its own flagshipOct 9Hardware · NewsNvidia's Vera Rubin reaches customer production as AMD buys an AI labOct 9Models · NewsGoogle's Gemini 4 Argon is here, but only cyber defenders can use it so farOct 9Models · NewsGPT-6 reaches every ChatGPT user: what OpenAI shipped, what it costs, and where each model runsOct 9Research · AnalysisOpenAI's Navier–Stokes claim, explained for builders: the verifier is the storyOct 9Models · AnalysisWhat the Claude Opus 5.5 system card actually saysOct 9Business · AnalysisToken pricing isn't the story: what the official price pages actually tell youOct 9Hardware · AnalysisThe gigawatt datacenter race: what's announced, what's running, and what's holding it backOct 9Models · BenchmarksLong-context recall: what independent benchmarks say "1 million tokens" really buys youOct 9Business · CompareFrontier APIs vs open models: what they really cost
Nothing in this format on this page — try another filter or the next page.
About this hub
"Frontier models" are the most capable general-purpose AI models available at a given time, such as the top tiers…