The introduction of Fable has significantly accelerated the progression of AI capabilities, with task length doubling in efficiency from every approximately seven months to every four months, thanks to Mythos-class models. Recent data highlights this shift, showing that while GPT-2 took around three seconds for tasks in 2019, the latest iterations, like Fable 5, are projected to take about 30 hours by June 2026. This trend demonstrates a remarkable leap in the handling of complex tasks by AI, following a series of impactful releases from leading AI labs that have consistently pushed the boundaries of what is possible.

Fable: Fable is the AI project developing Mythos-class models that have accelerated progress in handling longer and more complex tasks. It released Fable 5 in June 2026, marking a notable step beyond prior frontier systems in the field. The project is highlighted as driving the faster doubling rate for AI capabilities described in the news.
GPT-2: GPT-2 is an early OpenAI language model from 2019 that serves as a historical benchmark for AI task handling in the progression outlined. It is referenced to illustrate the starting point of the multi-year advancement in model performance.
GPT-4: GPT-4 is a 2023 OpenAI model positioned as a key milestone in the timeline of increasing AI task lengths. Its capabilities are contrasted with later systems to show the pace of development up to recent years.
Fable 5: Fable 5 is the June 2026 release from the Fable project, representing the latest Mythos-class model with substantially extended task length performance. It exemplifies the accelerated AI progression highlighted in the news.
Claude Opus 4.5: Claude Opus 4.5 is a December 2025 Anthropic model that further extended AI task length capabilities in the reported progression. It is cited as part of the accelerating trend leading to 2026 systems.
Claude Opus 4.6: Claude Opus 4.6 is a February 2026 Anthropic model that continued the rapid improvement in handling extended tasks. It is included in the comparison showing the shift to shorter doubling intervals.
Claude Sonnet 3.7: Claude Sonnet 3.7 is an early 2025 Anthropic model noted for extending task duration benchmarks in the AI evolution discussed. It represents an intermediate stage before more advanced Claude releases.

AI Capability Growth: Mythos-class models have shortened the doubling interval for AI task lengths compared to earlier generations.
Frontier Model Releases: Recent releases from multiple leading AI labs have pushed boundaries in complex task handling beyond previous standards.