At a Glance
| Vendor | Anthropic |
| Family | Claude 5 series |
| Launched | Anthropic released Claude Fable 5.1 on September 1, 2026, alongside Claude Mythos 5.1. Anthropic describes the two as the same model with different levels of safeguards. Fable 5.1 is the generally available version and sits at the top of the current Claude lineup, above Opus 5, Sonnet 5, and Haiku 4.5. |
| Context window | 1,000,000 tokens, with a maximum output of 128,000 tokens on the synchronous Messages API. Reliable knowledge cutoff is June 2026, the most recent of any current Claude model. Thinking is adaptive and always on, with high as the default effort. |
| Pricing | List pricing is unchanged from Fable 5 at $10 per million input tokens and $50 per million output tokens, with batch at half price. The change is in caching: cache reads fall 75 percent to $0.25 per million tokens. Anthropic estimates this makes Fable 5.1 about 25 percent cheaper than Fable 5 on typical workloads and around 45 percent cheaper on highly agentic ones that re-read long contexts. |
| Access channels | Claude API (claude-fable-5-1), Claude.ai, Claude Code, Claude Enterprise, and Claude Cowork. Cloud availability covers Amazon Bedrock, Claude Platform on AWS, Google Cloud, and Microsoft Foundry. Anthropic commits to not retiring it before September 1, 2027. |
Notable Benchmarks
Anthropic reports Terminal-Bench 4.0 at 55.8 percent (Fable 5: 42.0, Opus 5: 52.3, GPT-5.6 Sol: 37.3), Terminal-Bench-Science 0.1 at 52.6 percent against 24.7 for Fable 5, Humanity's Last Exam at 60.9 percent without tools and 65.0 percent with tools, OSWorld 2.0 at 77.9 percent partial and 41.7 percent strict, AutomationBench at 31.4 percent (Fable 5: 17.1), and CursorBench 3.2.0 at 73.4 percent. Anthropic also says Fable 5.1 matches or beats Fable 5 at low and medium effort, so the gains are not only a function of spending more tokens.
Strengths
Long-running agentic work, terminal and computer use, and knowledge work where the model has to stay on task across many steps. Cybersecurity safeguards now produce about 60 percent fewer false positives, which Anthropic says means roughly 60 percent fewer interruptions per Claude Code session, and it can now be used to find software vulnerabilities, though not to build exploits for them.
Limitations
It is the slowest and most expensive model in the current Claude lineup, and Anthropic still points most workloads at Opus 5 first. New API accounts can no longer manually edit Claude's prior context while keeping earlier thinking in the transcript, an anti-distillation change that Anthropic says will extend to all users with future releases. Pipelines built on rewriting conversation history need checking before migration.
Brand-Visibility Implications
The number that matters most for brands is not a benchmark but the June 2026 knowledge cutoff. Fable 5.1 recalls a newer web than any other Claude model, so brands that launched, rebranded, or changed pricing in the first half of 2026 may be described correctly for the first time, or described with a fresh set of errors. Re-baseline Claude answers against Fable 5.1 rather than assuming Fable 5 results carry over. Cheaper cache reads also make long-context research agents built on Fable 5.1 more economical, which favours brands with deep, consistent documentation that an agent can read end to end. See Claude family lineage, how training cutoffs affect brand recall, and Claude brand visibility.
How Presenc AI Tracks This Model
Presenc AI monitors brand visibility on Anthropic's Claude 5 series as part of continuous multi-platform AI visibility tracking. We sample Claude Fable 5.1 across representative prompt sets daily, compare against competitor performance on the same prompts, and flag material mention-rate changes so brand teams can respond quickly when AI representation shifts.