Anthropic, Google, Meta, and OpenAI have been sequentially announcing their latest AI models, which serve as their flagship models.
Model ReleasesAnthropicOpenAIClaude Fable 5.1GPT-6 Astra
Anthropic and OpenAI Release Next-Generation Models in Succession; OpenAI Mentions Slowing Development Pace
This article is a translation. Read the Japanese original
Anthropic has released "Claude Fable 5.1", which features significantly improved performance. Following this, Google has drastically refreshed its main model, announcing "Gemini 3.8 Flash" with performance approaching Opus 5. Additionally, Meta has released "Muse Spark 1.3", and OpenAI has released "GPT-6 Astra".
These models have recorded high scores on benchmarks and have demonstrated applications in advanced tasks such as protein design.
On the other hand, in contrast to the progress in development, shifts in the pace of development have also been suggested. OpenAI has expressed the intention to slow down its future development pace. Anthropic also argues for a slowdown across the industry, stating that as the social influence of AI expands, time is required for ensuring safety and performing "alignment"—the process of ensuring AI behavior aligns with human values.
Furthermore, in the United States, regulatory movements are intensifying, with Senator Bernie Sanders and others submitting bills calling for a temporary halt to the development of powerful AI and a ban on Artificial Superintelligence (ASI).
Sources
- 3日間で各社最上位AIモデルが続々登場 "驚けない”凄さと開発の「減速」 - Impress Watch （Google News: Anthropic、2026-09-05）