Breaking

Anthropic Opens Claude Fable 5.1 as OpenAI Begins GPT-6 Astra Rollout

September 3, 2026 at 21:06 EDT

Foundation Models
AI Agents
Software Dev & Coding

Anthropic has put Claude Fable 5.1 into broad consumer and enterprise availability, while OpenAI has started the staged rollout of GPT-6 Astra. The back-to-back launches make long-running agents, computer use, and deployment safeguards the central fault line in the frontier-model market. Availability details Launch details

September 2026 · Anthropic + OpenAI

Two frontier models target long-running work

Anthropic released Claude Fable 5.1 on September 1; OpenAI began staged access to GPT-6 Astra on September 3. Both emphasize coding, computer use, research, and enterprise workflows.

The release in four figures

1.05M

Astra context tokens

128K

Maximum output tokens

47%

Less reported OSWorld time

$10 / $50

Input / output per 1M tokens

Reported OSWorld 2.0 task time

Approximate time per task in the lab simulation

Astra · 40 min

GPT-5.6 Sol · 75 min

Astra reached the higher score in roughly 47% less simulated task time.

Capability meets deployment friction

Promising in practice

Long-horizon coding and research

Fewer iterations and faster verification

Tool-enabled workflows can run unattended

Limits remain

Sensitive Fable requests can route to Opus

Astra safeguards may pause legitimate work

Access and billing vary by plan

The deployment boundary matters

Astra reaches OpenAI’s Critical cybersecurity threshold. Fable 5.1 is broadly accessible, but sensitive tasks remain behind safeguards.

Broad access

Fable 5.1 is available to paid users and developers.

Restricted capability

Mythos 5.1 and advanced Astra cyber workflows remain controlled.

Continue reading

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