Breaking

OpenAI, Anthropic, Google and Meta Release New AI Models in Three-Day Wave

September 3, 2026 at 20:14 EDT

Foundation Models
AI Agents
Security

OpenAI, Anthropic, Google and Meta released new AI models between September 1 and September 3, intensifying competition around long-running agents, coding and cyber defense. The wave includes GPT-6 Astra, Claude Fable 5.1 and Claude Mythos 5.1, Gemini 3.8 Flash and Gemini 3.8 Flash Cyber, and Muse Spark 1.3.

September 1-3, 2026 · Frontier model wave

Four labs, four new bets on AI agents

OpenAI, Anthropic, Google and Meta released new models for coding, long-running workflows and cyber defense in three days.

Input price per 1M tokens

The introductory Gemini rate is about 13× below the $10 rate of Astra and Fable.

$0.75 Gemini 3.8 Flash

$10 GPT-6 Astra

$10 Claude Fable 5.1

1.05M

Astra context tokens

128K

Astra maximum output

~20%

Fewer tool calls for Muse Spark 1.3

What users value

Longer unattended tasks, cleaner coding loops and fewer unnecessary turns are becoming practical differentiators.

What remains difficult

Safety gates, possible timeouts, higher token use and non-equivalent benchmarks complicate direct comparisons.

The decision point

The next contest is cost per completed task, sustained quality and control in long-running agents.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.