Prompting Fable 5

# 09 — Prompting Claude Fable 5

Behavioral differences and prompting patterns for **Claude Fable 5** and **Claude Mythos 5**,
covering effort, instruction following, long runs, memory, and scaffolding changes. This page
is the model-specific layer that sits under the rest of this kit: the loop-selection,
harness, context, and failure-mode guidance in `docs/00`–`docs/08` still applies — this page
covers what changes when the model underneath is Fable 5.

This guide covers the prompting and scaffolding patterns specific to Claude Fable 5 and Claude
Mythos 5. For the model's capabilities, API changes, pricing, and availability, see
[*Introducing Claude Fable 5 and Claude Mythos 5*](https://platform.claude.com/docs/en/about-claude/models/introducing-claude-fable-5-and-claude-mythos-5).
For techniques that apply across all current Claude models, see
[*Prompting best practices*](https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/claude-prompting-best-practices).

Claude Fable 5 takes on problems that were previously too complex, long-running, or ambiguous
for prior models, and is particularly effective at end-to-end work that takes a person hours,
days, or weeks to complete. The teams seeing the best outcomes apply Claude Fable 5 to their
hardest unsolved problems; testing it only on simpler workloads tends to undersell its
capability range. It also performs reliably on more straightforward tasks.

Claude Fable 5 has several behavioral differences from Claude Opus 4.8 that may require prompt
or scaffolding updates. Capability improvements at this level are also a good prompt to
re-evaluate which instructions, tools, and guardrails are still needed. The patterns below
cover the behaviors that most often require tuning.

> **Note**
>
> For API parameter changes specific to Claude Fable 5 and Claude Mythos 5 (adaptive thinking
> only, summarized-only thinking output, no extended thinking budgets, the `refusal` stop
> reason and fallback handling), see
> [*Introducing Claude Fable 5 and Claude Mythos 5*](https://platform.claude.com/docs/en/about-claude/models/introducing-claude-fable-5-and-claude-mythos-5).
>
> Claude Fable 5 runs safety classifiers that target offensive cybersecurity techniques (such
> as building exploits, malware, or attack tooling), biology and life sciences content (such
> as lab methods or molecular mechanisms), and extraction of the model's summarized thinking.
> Benign cybersecurity work and beneficial life sciences tasks may also trigger these
> safeguards. To re-route declined requests automatically, configure
> [server-side or client-side fallback](https://platform.claude.com/docs/en/build-with-claude/refusals-and-fallback)
> to Claude Opus 4.8.

## Capability improvements

Compared with Claude Opus 4.8, Claude Fable 5 shows improvement in:

- **Long-horizon autonomy.** Claude Fable 5 sustains productive output over extended periods,
  completing multi-day, goal-directed runs with strong instruction retention across long,
  complex tasks.
- **First-shot correctness on complex, well-specified problems.** Early testers reported
  single-pass implementations of systems that previously took days of iteration.
- **Vision.** Claude Fable 5 interprets dense technical images, web applications, and detailed
  screenshots with substantially higher accuracy, often while using fewer output tokens, and
  is trained to use bash and crop tools to handle flipped, blurry, or noisy images.
- **Enterprise workflows.** Claude Fable 5 follows instructions, stays in scope, and produces
  professional-grade output on financial analysis, spreadsheets, slides, and documents.
- **Code review and debugging.** Bug-finding recall (outside the cybersecurity domains the
  safety classifiers cover) is noticeably higher than Claude Opus 4.8, including search across
  codebases and repository history.
- **Navigating ambiguity.** Claude Fable 5 performs well when given complex, multi-threaded
  requests and asked to determine next steps.
- **Delegation and collaboration.** Claude Fable 5 is significantly more dependable at
  dispatching and sustaining parallel subagents, and reliably manages ongoing communication
  with long-running subagents and peer agents.

Beyond these specific improvements, Claude Fable 5 is generally more capable than prior models
on almost all tasks. Claude Fable 5 is not intended for offensive cybersecurity or biology and
life sciences work; requests in those domains can return
[`stop_reason: "refusal"`](https://platform.claude.com/docs/en/build-with-claude/refusals-and-fallback).

## Longer turns by default

Individual requests on hard tasks can run for many minutes at higher
[effort](https://platform.claude.com/docs/en/build-with-claude/effort) settings, especially
when the task requires gathering context, building, and self-verifying, and autonomous runs
can extend for hours. This is one of the largest shifts teams encounter when adjusting to
Claude Fable 5. Adjust client timeouts, streaming, and user-facing progress indicators before
migrating, and consider restructuring harnesses to check on runs asynchronously, for example
through scheduled jobs, rather than blocking — this is the **outer, self-prompting loop** (see
[`01-the-agent-loop.md`](01-the-agent-loop.md#the-outer-loop-scheduled-self-prompting-runs)). To
keep Claude Fable 5 from overplanning when a
task is ambiguous: