UPDATE 14 June, 2026
Anthropic – Claude Mythos / Fable 5 pulled from market – The U.S. government ordered Anthropic to restrict access to their Mythos and Fable 5 models, leading Anthropic to remove these models entirely; this matters for users who had started to rely on their long‑context reasoning.
Anthropic released Claude Fable 5 on June 9, 2026 as its first “Mythos‑class” model that regular users can access broadly through the Claude chat interface, Claude Code, and the Claude API.
It is described as Anthropic’s most capable generally available model, matching the private Mythos 5 model in performance while adding stricter safeguards in high‑risk domains.
Fable 5 supports a 1‑million‑token context window with up to 128,000 output tokens, making it well suited to long‑horizon reasoning and complex, multi‑step tasks such as large research syntheses.
Pricing is currently about double Claude Opus (around 10 USD per million input tokens and 50 USD per million output tokens), but Anthropic is temporarily including Fable 5 at no extra cost on Pro, Max, Team, and certain enterprise plans through June 22 before switching it to usage‑credit billing.
Why it matters for genealogists
With its huge context window and stronger long‑form reasoning, Fable 5 looks especially promising for:
Ingesting a large body of notes and transcriptions (for example, a complete county research file) and helping you draft narrative reports or proof arguments without losing track of nuance.
Orchestrating more complex, agentic workflows, where the model has to plan steps, check its own work, and update a plan over time as you feed it more evidence.
Here are concrete, Claude‑Fable‑specific things you can do right now as a working genealogist.
How Fable 5 changes the game for genealogy
Claude Fable 5 is Anthropic’s first Mythos‑class model released for general use, with a 1‑million‑token context window and up to 128k output tokens. It is explicitly designed for long, complex “knowledge work” and multi‑step tasks that earlier models struggled to hold in working memory, and it’s temporarily available on Pro/Max/Team plans at no extra cost until June 22.[simonwillison]
For genealogy, that combination maps almost perfectly onto three pain points:
You can finally drop an entire county‑or‑case file (thousands of words of notes, timelines, and transcripts) into a single conversation and ask coherent, sustained questions about it.
You can ask Fable to plan and revise multi‑step work over a longer period—iteratively refining a research plan as you add more sources, rather than starting over each session.[anthropic]
You can use it as a “devil’s‑advocate assistant” on your densest arguments because it can see more of the underlying evidence at once.[constellationr]
Below are 10 specific micro‑workflows that make good use of Fable’s strengths and price point.
10 Claude Fable micro‑workflows for genealogists
1. Whole‑project case ingest and mapping
Goal: Let Fable “see” a full, multi‑generation research project at once and map what you’ve already done.
Assemble your compiled notes: research logs, timelines, key transcriptions, and short source summaries for one complex problem (for example, a 3‑generation Oklahoma Territory family you’ve been working on for years).
Paste as much as you can into one Claude Fable 5 chat, taking advantage of the million‑token context.[platform.claude]
Prompt: “Please read everything above as my complete working file on the [Surname] family. Create: (1) a chronological timeline of events; (2) a table of sources with what each one actually proves; (3) a list of unresolved questions or conflicts; and (4) a short paragraph that states my current main hypothesis in your own words.”
This is exactly the kind of long‑running knowledge work Fable is marketed for: staying focused across millions of tokens and improving its outputs using its own notes.[vellum]
2. Fable as a “proof argument second reader”
Goal: Use Fable as a meticulous editor on dense proof arguments.
Paste your full argument (including embedded citations, if possible) into Fable in one go. The large context window can hold both your analysis and a substantial portion of cited excerpts.[simonwillison]
Prompt: “Treat this as a genealogical proof argument in draft. Identify: (1) every place where I am relying on indirect evidence; (2) every conflict I have mentioned; (3) any implied assumptions; and (4) any point where a skeptical reviewer could reasonably demand more evidence. Respond as if you’re reviewing for a peer‑reviewed genealogical journal.”
Because Fable is tuned for analytical, long‑form reasoning, it is well‑suited to spotting hidden assumptions and under‑explained leaps.[anthropic]
3. Multi‑day research‑plan refinement in one long thread
Goal: Keep an evolving research plan “alive” as you work over days.
Start a dedicated Fable chat for a single research question and draft an initial plan together (collections to search, repositories to contact, negative searches to record).
Each research session, paste your new log entries and findings into the same thread and ask Fable to update the plan, mark completed items, and suggest next steps given the new evidence.
Because Fable is designed for multi‑day agentic work and can stay with a complex task for long stretches, you can avoid constantly re‑summarizing what you’ve already done.[youtube][truefoundry]
This approximates having a project‑memory assistant that never forgets what you tried two weeks ago.
4. “All‑sources on one ancestor” synthesis
Goal: Feed every major source for one ancestor into a single context and get a structured synthesis.
Compile transcriptions or detailed abstracts for every record relating to one ancestor (census, land, probate, military, court, church, etc.).
Drop them into Fable and prompt: “Group these sources by type and by repository; for each group, summarize what they collectively show about identity, relationships, residence, and chronology. Then highlight conflicts and propose at least two plausible resolutions for each conflict.”
Here you’re using Fable’s long‑context and reasoning capabilities to see patterns across a bigger slice of evidence than Sonnet/Opus comfortably hold in a single chat.[vellum]
5. Teaching/demo: Fable vs. your current daily driver
Goal: Decide where Fable belongs in your personal toolkit and show others.
Take one complex prompt you already use in teaching or practice (for example, a brick‑wall locality research plan or a proof‑argument critique) and run it in both Fable and your current “best” model (Claude Sonnet, GPT‑5.5, or Gemini).
Compare responses on: depth of analysis, handling of contradictions, and explicitness about uncertainty.
Because Fable is priced and positioned as a premium reasoning model—roughly twice Opus and GPT‑5.5 on input—it’s better reserved for the hardest jobs if you’re cost‑sensitive.[finout]
This quick bake‑off helps you decide when Fable’s extra cost is justified in real genealogy work.
6. Cross‑jurisdiction land and probate strategy
Goal: Use Fable’s long context to coordinate complex jurisdiction shifts (for example, colonial/territorial/state changes).
Create a document that lists all the jurisdictions and boundary changes affecting a research area (such as Indian Territory into Oklahoma), plus bullet summaries of known land and probate record sets per period.
Feed the entire document into Fable and prompt: “Using everything above, design a step‑by‑step research strategy for land and probate records for the [Surname] family from 1870–1930, noting which repositories or online platforms are most likely to hold each record type, and pointing out where records were lost or never created.”
Fable’s long‑horizon planning strength is well‑matched to this kind of multi‑layer jurisdictional work, where you want one coherent strategy instead of piecemeal advice.[constellationr]
7. “Devil’s advocate” on a tough identification problem
Goal: Explicitly ask Fable to argue against your preferred identity or relationship conclusion.
Paste a focused section of your notes where you’ve resolved a same‑name or FAN‑club cluster problem.
Prompt: “Take the role of a skeptical genealogist on a peer‑review board. Based only on the evidence and reasoning shown here, list every plausible alternative identification or relationship, and for each, explain what additional evidence you would seek to support or refute it.”
Because Fable is tuned for deeper reasoning and is state‑of‑the‑art on complex tasks, you’re leveraging it not to agree with you but to strengthen your argument by surfacing alternate hypotheses.[anthropic]
8. Long‑form locality guide with embedded record‑creation context
Goal: Generate a detailed, method‑oriented locality guide informed by a big excerpt of your own notes and outside references.
Combine snippets from published guides, catalog notes, and your own experience for a specific locality (e.g., “Probate and land records in X County, 1850–1930”).
Paste all of that into Fable and prompt: “Rewrite this as a single, structured locality guide aimed at advanced genealogists, focusing on record‑creation processes, gaps, and pitfalls. Keep the structure: Overview, Jurisdiction changes, Record types and where to find them, Known gaps or losses, and Research strategies.”
The large context window means Fable can harmonize multiple sources and your own commentary without losing details or repeating itself excessively.[simonwillison]
9. Multi‑source oral‑history and document alignment
Goal: Align a long oral‑history transcript with your document trail.
Transcribe an extended oral‑history interview (or use another model/tool to do so), then copy the full transcript plus key document summaries into a Fable chat.
Prompt: “From the oral history above, extract a timeline of events, people, and places. Then compare that timeline with the documentary evidence summarized here, highlighting all agreements, conflicts, and places where the oral history suggests leads not yet documented. Propose prioritized follow‑up research questions.”
Fable’s “knowledge‑work” positioning is ideal for this kind of cross‑source reconciliation over a large volume of text.[constellationr]
10. “AI‑augmented editor” for a book‑length family narrative
Goal: Use Fable to help edit or restructure a long family history manuscript.
Paste one chapter at a time of a book‑length ancestral narrative into Fable, keeping the whole table of contents and a synopsis of previous chapters in the same thread so it can see the big picture.
Prompt: “You are a structural editor for a historical family narrative. Given the synopsis above and this chapter draft, identify continuity problems, missing context for non‑specialist readers, and opportunities to integrate mini‑explanations of record types and research challenges. Suggest specific edits and where to move or expand sections.”
Because Fable can maintain more global context than earlier models, it can give feedback that respects the arc of the whole work, not just a single chapter in isolation.[vellum]
No comments:
Post a Comment