Fable, Opus, Mythos. Anthropic’s model names, explained in plain English.

Updated 15 October 2026 against Anthropic's own release pages. The June export-control suspension was lifted on 30 June 2026.

A year ago, knowing your Claude from your Copilot was enough. Now the model names matter commercially: which one your subscription includes, which one is gated behind a trusted-access programme, and which one your team should build workflows on are different questions with different answers. Here is the whole naming system on one page, in plain English, with the prices in pounds.

Two naming quirks worth knowing. First, the version numbers do not line up across tiers: Opus is on 4.8 while the Mythos class launched at 5. Second, Fable and Mythos are the same model; Anthropic chose two names precisely because the difference is the safeguards, not the intelligence. Fable comes from the Latin fabula, “that which is told”, a sibling of the Greek mythos.

Launched 9 June 2026, Fable 5 is the first Mythos-class model anyone can use. It ships wrapped in safety classifiers: separate AI systems that watch for queries about cybersecurity, biology and chemistry, or attempts to distil the model into competing systems. Flagged queries are answered by Opus 4.8 instead, and the user is told; Anthropic reports this touches fewer than 5% of sessions. Early evidence of the capability jump is striking: Stripe compressed a two-month, 50-million-line code migration into a day, and Simon Willison called it “something of a beast”. Our full breakdown, including the until-June-22 inclusion window on paid plans, is in the Fable 5 employer post.

Mythos 5 is the same model with some safeguards lifted, which gives it what Anthropic calls the strongest cybersecurity capabilities of any model in the world. It is restricted to vetted organisations, most visibly the Project Glasswing cyberdefence consortium (AWS, Apple, Cisco, CrowdStrike, Google, JPMorganChase, Microsoft, NVIDIA and others), which has used Mythos-class models to surface more than 10,000 high or critical severity vulnerabilities in production software. Anthropic plans a broader trusted-access programme for cybersecurity organisations, and a separate one for biology researchers. The governance story, and what it means for UK security teams, is covered in our Mythos and Glasswing post.

Fable 5.1 and Mythos 5.1 (September 2026). Same model, two safeguard levels, as before. Fable 5.1 is generally available on every platform, including AWS, Google Cloud and Microsoft Azure, as claude-fable-5-1. Token prices are unchanged from Fable 5 ($10 per million input, $50 per million output), but cache reads, where the model re-reads context it has already processed, were cut by 75% to $0.25 per million. Anthropic puts the effect at about 25% cheaper for typical workloads and up to about 45% cheaper for long agentic runs. Two changes matter for UK employers beyond price: Fable 5.1 can now be used to find software vulnerabilities (though not to build exploits), with roughly 60% fewer safeguard interventions in Claude Code; and a new Enterprise Frontier Safeguards scheme gives enterprise customers the privacy of zero data retention by keeping data on the customer's own cloud, rolling out in phases from this autumn, with eligible customers able to use Fable 5.1 under zero data retention in the meantime.

Mythos 5.1 is identical to Fable 5.1 with more permissive cyber and life-sciences safeguards. It is available only through Anthropic's Cyber Verification Program and a new Life Sciences Verification Program run with the US government, and, in Anthropic's words, "currently it is only available to a set of US organisations". So for a UK business the practical position is unchanged: your team will meet Fable 5.1 or the 5.5 models, not Mythos.

The 5.5 family. Three weeks after Fable 5.1, Anthropic refreshed every tier below it. Opus 5.5 (22 September) is described on Anthropic's transparency hub as "our new leading model" and "our most capable general release model to date, matching or exceeding Claude Fable 5.1 and Claude Mythos 5.1 on many of our evaluations", which blurs the old line between the Fable and Opus tiers. Sonnet 5.5 (28 September) is pitched as the everyday work partner, about 30% faster than Sonnet 5 at $2 per million input and $10 per million output, and it is the first Sonnet to carry the same cyber safeguards as Fable and Opus. Haiku 5.5 (7 October) is the fast, cheap tier at $0.10 per million input and $0.50 per million output for prompts up to 100K tokens, available to Free, Pro, Max, Team and Enterprise users. Anthropic's own guidance on which to use: "for complex coding and knowledge work, Opus 5.5 is the daily driver, and Sonnet 5.5 is a good fit for well-scoped tasks."

Watermarking. In July 2026 Anthropic signed the EU AI Act's Code of Practice on transparency of AI-generated content. As a result, text from models released after 2 August 2026, so Fable 5.1 and the 5.5 family, carries an invisible watermark readable only through a detection API that is in private preview. We cover what that does and does not prove in the watermark post.

Is there a Claude Fable 6? No. As of 15 October 2026 Anthropic has announced nothing called Fable 6. The reports circulating since August trace back to a single unsourced article and a speculative video; Anthropic's current models are the ones listed above. If that changes, this page will.

On the API, Fable 5 and Mythos 5 cost around £7.50 per million input tokens and £37 per million output tokens (converted from dollar pricing at the June 2026 rate), roughly twice Opus 4.8 and less than half what Mythos Preview cost. On subscriptions, Fable 5 is included on Pro, Max, Team and seat-based Enterprise plans at no extra cost until June 22; from June 23 it needs usage credits until capacity lets Anthropic restore it as standard. CNBC reported the commercial logic from Anthropic’s product lead: customers buy accuracy and return per task, not cheap tokens, and the more intelligent model wins on ROI.

October 2026 update: Fable 5.1 keeps the same token prices as Fable 5, so roughly £7.50 in and £37.50 out per million at £0.75 to the dollar, with cache reads now about £0.19 per million. Opus 5.5 is reported at $4 in and $20 out (about £3 and £15), Sonnet 5.5 at $2 and $10 (about £1.50 and £7.50), and Haiku 5.5 at $0.10 and $0.50 (about 8p and 38p) for prompts up to 100K tokens. Opus 5.5 pricing is from secondary coverage; the others are Anthropic's published figures.

The names will keep changing; the release cadence is roughly every two months and Anthropic says more capable models arrive in the coming months. What does not change is the work around the model: spotting the process, commissioning the task, wiring the tools, checking the output, governing the result. That is the skill the Claude Apprenticeship builds in your doers (a live automation by month 3, a working agent by month 4), and that the free AI Leadership taster days introduce to your leaders. Train for the discipline, and every name on this page becomes an upgrade rather than a disruption.

Sources & further reading

Anthropic: Introducing Claude Fable 5.1 and Claude Mythos 5.1 (September 2026), Transparency Hub (Opus 5.5 and Sonnet 5.5 summaries) and Claude Haiku 5.5 (7 October 2026). TechCrunch: Anthropic releases Sonnet 5.5. CNBC: export controls lifted, 30 June 2026. Windows Forum: Claude Fable 6 rumour: no release or roadmap confirmed. Opus 5.5 token pricing as reported by secondary coverage, not yet checked against an Anthropic pricing page. Anthropic: Claude Fable 5 and Claude Mythos 5. CNBC: Anthropic releases Mythos-like AI model to the public. Simon Willison: Initial impressions of Claude Fable 5. TESS: the Opus 4.8 launch post.