共有:

Model Layer Update

Felo's deep-research tool is now running on Claude Fable 5.1

The feature name — "Fable Research" — hasn't changed. What changed is the model underneath it. The upgrade Felo announced on its official blog isn't a new product feature; it's an inherited swap from Anthropic's own model update. Here's what actually changed, and what didn't.

AI Navigate Editorial2026.09.066 min read

Felo (search / LiveDoc / Fable Research UI) Look and workflow: unchanged Fable Research agent Produces structured, source-backed reports Claude Fable 5 Previous foundation Claude Fable 5.1 Current foundation, since 2026.09
FIG. Felo's search experience stays the same; only the model underneath swapped from Fable 5 to Fable 5.1
01
What Actually Changed

This isn't a new feature.
It's a model swap underneath one.

Get the subject of this story wrong, and you'll misread the whole thing.

Felo is an AI search product that bundles search, summarization, and slide generation, and it has offered a deep-research feature called "Fable Research" — which pulls from multiple sources to build structured, source-backed reports — powered by Anthropic's "Fable" model family. On its official blog, Felo announced that it has upgraded Fable Research's underlying model from Claude Fable 5 to Claude Fable 5.1. Nothing about the screen or the workflow users see has changed at all.

Which is exactly why the real question isn't "what did Felo build," but "what did Anthropic actually change in Claude Fable 5.1 itself." Covering the release of Fable 5.1 around September 1, 2026, TechCrunch's reporting frames two changes as the core of the update relative to the original Fable 5: it's cheaper, and it's less restrictive. That's not a flashy new-capability announcement — it's a cost and usage-policy shift that shows up directly in day-to-day use.

Claude Fable 5 (previous foundation)Claude Fable 5.1 (current)
Early testers praised its long reasoning chains for tracking multi-part questionsPer TechCrunch, delivers comparable or better reasoning at lower cost
Relatively stricter usage constraintsReported to be less restrictive, widening what it can be used for
The foundation Fable Research ran on until nowFable Research fully migrated onto this in September 2026
02
The Numbers

A concrete perk — for exactly seven days

The one place this story gives you a hard, actionable number is the discount.

2026.09.01 Claude Fable 5.1 released 2026.09.06 Felo's Fable Research migrates +7 days 50% off for Pro users
FIG. From the model's release to Felo's migration to the time-limited discount
50%
Discount on Fable Research usage credit
7 days
Duration of the Felo Pro discount
09.01
Claude Fable 5.1 release date (2026)

To mark the migration, Felo's official blog says Pro users get a 7-day, 50% discount on Fable Research usage credit. This is not a permanent price cut — it's explicitly a launch-window promotion, a distinction worth keeping in mind, covered further below.

03
Who Benefits

Multi-part research wins.
Quick lookups won't notice.

Two Felo users can have completely different experiences depending on how they use it.

Business / competitive research

A multi-part teardown like "pricing, features, and recent announcements across ten competitors" benefits directly from reasoning coherence. The more sub-questions involved, the more it matters whether the model stays on-topic all the way through the report.

PM / product planning

When scoping a new feature, PMs often need to pull together market trends, comparable products, and technical constraints in one pass — exactly the kind of high-sub-question research task where the upgrade should show up most.

Marketer (casual search)

A quick lookup — "what does this term mean," "where's this stat from" — doesn't give the model much room to show off multi-step reasoning, so the difference is basically imperceptible. As the impact note puts it, this group can safely ignore the change.

04
What To Do Next

What to do in this 7-day window

01

Run that deep-research task you've been putting off

If you've been holding back a ten-question-class multi-part research job because of cost, now is the window to run it through Fable Research. This is a limited-time promotion, not an ongoing rate — the longer you wait, the less of the discount is left.

02

Split tasks between Fable Research and regular search

Keep single-question lookups on regular search, and reserve Fable Research for jobs that genuinely need a coherent report across multiple angles. That split keeps cost and payoff in balance.

03

Check the model-layer change at the primary source

"Cheaper" and "less restrictive" are worth verifying yourself via TechCrunch's reporting. Before you route an important decision through it, it's worth reading the model-side coverage, not just Felo's own announcement.


What Felo changed wasn't the screen.
It was the model underneath it.


05
Counterpoint

Three reasons not to get too optimistic

First, this isn't a story about Felo building something new — it's an existing feature getting a model swap underneath it. Since the front-end experience is identical, the substance of the change is only visible if you read Anthropic's own announcement, not just Felo's. Don't over-credit the product-layer post alone.

Second, the 50% discount lasts exactly seven days. Read it as "Fable Research just got permanently cheaper" and you'll be surprised by the bill once the window closes. Treat the discount purely as a one-time nudge to try the migration, not a new baseline price.

Third, because Felo is a wrapper product built on top of Anthropic's models, its quality and its constraints are ultimately set by the layer underneath. This time the change happened to be favorable — cheaper, less restrictive — but the same structure means that if Anthropic ever updates the model in a different direction, Felo's user experience can shift regardless of anything Felo itself does.