June 30 was a two-headline day for Anthropic: the US cleared Claude Fable 5 to return, and Sonnet 5 shipped the same afternoon. One is the frontier you keep in your back pocket; the other is the near-Opus workhorse you'll actually run all day. Here's how they stack up.
Three weeks ago, Anthropic's most capable model was a ghost. Claude Fable 5 launched on June 9, went dark on June 12 under a US export-control directive, and spent two weeks returning one cleared organization at a time. Then came June 30, 2026 — a two-headline day. The government cleared Fable 5 to come back for everyone, and Anthropic used the same afternoon to ship a brand-new workhorse: Claude Sonnet 5.
So this is really two stories that landed together: the flagship's comeback, and the quieter model that's going to do most of your actual work.
The Comeback Is Official
On June 30, Anthropic announced that the US Department of Commerce had lifted the export controls that pulled Fable 5 and its unguarded sibling Mythos 5 offline in mid-June. The company began restoring access on July 1, redeploying Fable 5 globally for general use across the Claude API and every subscription plan.
If you're just tuning in: Fable 5 is a Mythos-class model — Anthropic's top tier — made safe for general use. It claims state-of-the-art results on nearly all tested benchmarks, from software engineering to scientific research, and can work autonomously for longer stretches than any Claude before it. What keeps it “safe” is a set of classifiers that quietly hand the riskiest requests (offensive cybersecurity, dangerous bio/chem, and capability distillation) off to Claude Opus 4.8 instead. Those fallbacks trigger in fewer than 5% of sessions, so more than 95% of the time you're talking to Fable directly.
The catch is the price of the frontier: Fable 5 runs $10 per million input tokens and $50 per million output tokens. It's the model you reach for when the problem genuinely warrants it — a gnarly migration, a hard research question, a long autonomous agent run.
Same Day, a New Workhorse: Sonnet 5
Here's the part that got a little buried under the comeback headlines. On that same June 30, Anthropic shipped Claude Sonnet 5, calling it “the most agentic Sonnet model yet” — built to make plans, use tools like browsers and terminals, and run autonomously through long, multi-step workflows.
The headline claim is the one that matters: Sonnet 5's performance lands close to Opus 4.8, Anthropic's flagship, at a fraction of the price. Compared to February's Sonnet 4.6, it's a real step up in reasoning, tool use, coding, and knowledge work — and, notably, in follow-through. Early partners said it “finishes complex tasks where previous Sonnet models would stop short,” and it self-checks its own output without being told to. It also hallucinates and sucks up to you less, refuses malicious requests more reliably, and resists prompt-injection hijacks better than its predecessor.
Then there's the price, which is the whole point of a Sonnet:
- Introductory pricing of $2 per million input tokens and $10 per million output tokens, through August 31, 2026.
- After that, it settles at $3 / $15 — still a fraction of Fable 5's $10 / $50, and roughly half the cost of running Opus 4.8 at medium effort.
Sonnet 5 is now the default model for Free and Pro plans, available to Max, Team, and Enterprise users, and wired into Claude Code and the API (call it as claude-sonnet-5). It ships with cybersecurity safeguards on by default — the same posture as Opus 4.7/4.8, and deliberately less permissive than Fable 5.
So Which One Do You Reach For?
The honest answer: mostly Sonnet, occasionally Fable, Opus when accuracy is everything.
- Reach for Sonnet 5 for the daily grind — feature work, code review, long agent loops, tool use, drafting. It's fast, it's near-Opus in quality, and at $2/$10 right now it's almost hard to justify anything pricier.
- Reach for Fable 5 when the task is genuinely at the frontier and the token bill is worth it — a sprawling migration, deep research, or an autonomous run where you want the strongest reasoning available and don't want to babysit it.
- Reach for Opus 4.8 when you need the highest accuracy Anthropic offers and cost is a distant second concern.
That's the real shape of Anthropic's lineup after June 30: a frontier model that's finally back in your back pocket, and a new workhorse that quietly closes most of the gap to it for a fifth of the price. The exciting news is that Fable 5 is available again. The useful news is that, for most of what we actually do all day, Sonnet 5 is now more than enough.
Sources: Anthropic — Introducing Claude Sonnet 5 · Anthropic — Claude Fable 5 and Claude Mythos 5
