Anthropic suspended global access to its newly launched large language models Claude Fable 5 and Claude Mythos 5 on June 12, 2026, after a US government export control directive. According to reports, the company was given roughly 90 minutes' notice.
June 9–13, 2026 · Anthropic
Launched, then pulled in 90 minutes: Claude Fable 5 & Mythos 5 go dark
Anthropic's most powerful general-release model shipped on June 9. Four days later a US government export-control directive forced the company to disable both models globally — unable to filter foreign-national access in real time.
90 min
Time given to act after the government's call at ~1 p.m. ET
2
Models disabled worldwide; other Claude models unaffected
4 days
From June 9 launch to the June 12–13 shutdown
The 24-hour timeline
JUN 9
Fable 5 released — first Mythos-class model safe for general use
→
JUN 13 · 1 PM ET
Government demands a block on foreign-national access — no threat detail
→
JUN 12–13
Real-time nationality filtering infeasible → both models disabled globally
Same foundation, different guardrails
Both models share one underlying model.
Fable 5 — general release
Safety classifiers fall back to Opus 4.8 in cyber / bio / chem domains
Mythos 5 — restricted
Guardrails selectively removed; trusted partners only (Project Glasswing)
Specs & pricing (pre-shutdown)
Context 1M in / 128K out
Input price $10 / M tokens
Output price $50 / M tokens
Vs Opus 4.8 ~2× the cost
Prompt caching up to −90% input
SWE-Bench Pro — claimed gain over Opus 4.8
Anthropic claims roughly +10 points on the agentic software-engineering benchmark.
Praise
Handled real tasks "with ease" over a 5.5-h Continue reading The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in ✓ Signed in — this article isn’t included in your current plan.Unlocking the full article…