| Context window | 1M tokens |
|---|
| Max output | 128K tokens |
|---|
| Input / output price | $10 / $50 per 1M tokens |
|---|
| Thinking | Adaptive (always on), default effort high |
|---|
| Knowledge cutoff | Jan 2026 |
|---|
| Status now | Legacy; succeeded by Claude[2] Fable 5.1 |
|---|
From Anthropic's platform docs. The launch benchmark table is an image on the announcement page and could not be read as text, so no benchmark graph is saved. · source
Estimate from the version's published specs, benchmarks and reviews, 0-5; scores are only comparable within one criteria year. Scored as launched (Claude Fable 5 / Mythos 5, 9 Jun 2026; same model, different safeguards; scored on Fable 5): Artificial Analysis Intelligence Index 64.9 (max effort, Opus 4.8 fallback, #1) per its 9 Jun 2026 launch article, the index version then current; the live page now shows 50 on v4.3.2. Coding null: AA's Terminal-Bench 4.0 run post-dates launch; Anthropic reports Terminal-Bench 2.1 (88.0%, starred, falls back to Opus). 1M-token context; $10/$50 per 1M (about $20 blended). 69.7 output tokens/s on Artificial Analysis (live page, 24 Oct 2026). Text and image input. 4 reviews, string-matched; the TechCrunch quote is a customer statement it reports.[1]
Claude Fable 5 scores 64.9 on the Artificial Analysis Intelligence Index, claiming the #1 rank overall.
I'm really impressed with the quality of API design, tests, code and documentation that Fable put together for this.
Fable automatically falls back to the weaker Opus 4.8 or refuses to respond when it suspects sensitive topics.
Vibe-coding platform Base44 noted in a statement that Fable is better at "one-shotting full apps" and has excellent tool-calling.
| Claude Fable 5 Launches at #1 on the Artificial Analysis Intelligence Index · Jun 9, 2026 | 0.80 |
|---|---|
| Initial impressions of Claude Fable 5 · Jun 9, 2026 | 0.90 |
| Anthropic's Claude Fable 5 is a version of Mythos the public can access today · Jun 9, 2026 | 0.70 |
| Claude Fable 5: The first Mythos model is powerful, expensive, and heavily filtered · Jun 10, 2026 | -0.40 |
Photos from readers, kept apart from the pin's sourced media. Rate what you ate or saw, and on a restaurant pin tag the dishes right on your photo.
The first entry is always the pin's source. Overall confidence is a weighted average of how firmly each reference supports the start and end times used above; a reference counts half as much for every 180 days older than the newest.
Anthropic's[4] announcement is dated Jun 9, 2026 and begins "Today we're launching Claude[2] Fable 5"; the Claude platform docs list "Released June 9, 2026" for claude-fable-5, and CNBC's[6] report is dated 9 June 2026. Access was suspended on 12 June and restored on 1 July 2026; those are later steps, not the release.
Model reference: claude-fable-5 released June 9, 2026, 1M-token context, 128K max output, $10 and $50 per 1 million tokens, adaptive thinking always on with high default effort, a January 2026 knowledge cutoff, available on the Claude API, Amazon Bedrock, Google Cloud and Microsoft Foundry; now marked legacy after Fable 5.1 with retirement not sooner than June 9, 2027.
Summarises the Fable 5 and Mythos 5 release on 9 June 2026, the 12 June US letter barring non-US nationals, access restored to some US organizations on 26 June and generally on 1 July, Fable 5 free to all subscription plans from 1 to 19 July and a standard feature of higher tiers from 20 July, and Fable 5.1 on 1 September.[1]
Anthropic's[1] timeline: launch on 9 June, a US export directive on 12 June that led it to suspend all access, Mythos 5 approved for some US organizations on 26 June, controls lifted on 30 June and Fable 5 restored globally on 1 July, with a new classifier that blocks the Amazon-reported bypass in over 99% of cases.
Andrew Hoblitzell's report: first publicly available Mythos-class model, available the same day on the Claude[2] API, AWS and Microsoft Foundry, the mandatory 30-day data retention that led Microsoft to pull it from its internal Copilot model picker, an external bug bounty of over 1,000 hours with no universal jailbreak, and the export directive that took it offline within three days.
Something missing or wrong? Say it in your own words: a link that backs this pin up, a different start or end date and why, or a fact it lacks or gets wrong. The AI checks it against this pin's sources, searches for better ones, and adds any page that backs you up. The pin's own sources still count most. A picture that shows something else, or shows it badly, is looked at too, and moved down or replaced.