This Claude Fable 5.1 review covers the first seven days after Anthropic shipped the model on 1 September 2026. We ran no benchmark on it. Every number below is traced to the source that produced it, and two of the week's most repeated figures mean close to the opposite of how they were quoted.
The short version for anyone deciding today: Fable 5.1 is a real step up on long, hard, single-shot work, and a step sideways or backwards on short tasks. Its price fell and its cost rose, and both of those are true at once. Where a claim below rests on one person, it says so.
The short answer
Claude Fable 5.1 costs 10 dollars per million input tokens and 50 dollars per million output, unchanged from Fable 5. It carries a one million token context window, 128,000 tokens of output, and a June 2026 knowledge cutoff. The whole price story sits in one line of the pricing table: cache reads dropped to 25 cents per million, which is 2.5% of base input, against 10% for every other current Claude model. In the Claude apps it needs the 100 dollar Max plan, and the Pro plan does not include it.
The useful finding of week one is not a benchmark. It is that the two most quoted numbers about this model, a cost cut and a hallucination rate, both reverse when you read the denominator underneath them. One says the model got cheaper when many workloads got dearer. The other says the model got worse when it measurably got more accurate.
What shipped
Fable 5.1 arrived with a restricted sibling, Claude Mythos 5.1, which Anthropic describes as the same model with different safeguard levels and gates behind a verification programme. The specifications below come from Anthropic's own documentation and the model card, read on 6 September 2026.
| Specification | Fable 5.1 | Why it matters |
|---|---|---|
| API model ID | claude-fable-5-1 | A pinned snapshot despite the dateless name. No dated alias exists. |
| Input and output | $10 and $50 per million | Identical to Fable 5. Nothing moved here. |
| Cache reads | $0.25 per million | 2.5% of base input, against 10% on other current Claude models. |
| Context and output | 1,000,000 and 128,000 | The output ceiling drives the cost story further down this page. |
| Adaptive thinking | Always on, cannot be disabled | Effort runs low, medium, high, xhigh, max. The default is high. |
| Knowledge cutoff | June 2026 | Anything after that needs to be in the prompt or a tool call. |
One line in the model card changes how you write a client. Anthropic states that blocked requests return a standard HTTP 200 response carrying a refusal stop reason and a category. It adds that refusal rates on this model run materially higher than on previous Claude models, and that customers should treat refusal as a primary response path, not as an error. Billing follows the same split. A refusal caught before inference costs nothing, and one caught partway through the answer bills for the tokens already generated.
What people built in week one
Filtering out the demos that only prove a model can produce something shaped like software, six builds stand up. Each names its source so you can go and look.
A production bug that had survived five years
The hedge fund Millennium handed Fable 5.1 a crash dump for a defect that had escaped its engineers for five years. The model disassembled a closed-source vendor library into raw assembly and traced the cause. That reached us through a launch review with over two million views, not through Anthropic, which makes it a second-hand account of a first-hand result, so treat the detail as indicative and the shape as real.
Source: Fireship on YouTube.
A racing game from one prompt, by five sub-agents
The repository turbo-kart-rush was assembled by five Fable 5.1 sub-agents working from a single prompt, with entirely procedural assets and no art pipeline. It sat at 35 stars in week one. No human sequenced the agents.
A hundred thousand line codebase review for about two dollars
One developer ran Fable 5.1 across roughly 100,000 lines that earlier Claude models had largely written, and reported the review cost around two dollars and surfaced real problems. This is the cheapest useful thing anyone reported doing with the model all week.
A 74,000 line semantic layer, analysed in sixteen minutes
In a three-way head-to-head, Fable 5.1 read a real client codebase plus a meeting transcript, consulted thirteen sources, and produced a four-option comparison matrix, a 250-line design document and a migration plan. It took sixteen minutes and eight dollars, and an independent judge called it the most complete document of the three at the highest cost and slowest speed.
Source: Duke Pan on YouTube.
A walkable city block, and a house from its floor plan
One project rendered an explorable three-dimensional Union Square and drew 327 points on Hacker News. Separately, a developer handed the model the floor plan of a house they were considering buying and got back a three-dimensional model they could walk through. Spatial reasoning from a two-dimensional drawing is the common thread.
Sources: github.com/PhiloLabs/fable51-worlds and the floor plan, on X.
A physics platformer, one-shotted in an hour
A solo developer reported one-shotting a rope-swinging two-dimensional platformer, graphics and music and mechanics together, in about an hour, then spending a few days polishing it. Single source, and consistent with the pattern in every other item here.
Source: r/aigamedev.
Read those six together and a shape appears. Every one is a long, hard, single-pass task. Not one is a quick edit. That is the same conclusion the tips section reaches from the other direction, and it is the most useful thing anyone learned about this model in seven days.
Prompting practices
Each practice below carries its source. One was measured by a third party. Four rest on a single person, which is worth knowing before you rebuild a workflow around them.
| Practice | Source | Strength |
|---|---|---|
| Start at low effort and raise it only when the output forces you to | Code Rabbit measured low effort catching 61% of real problems against 57.1% at high, and running faster. | Third party measured |
| Tell it to edit only the span that changes, never the whole file | It defaults to rewriting entire files for small edits, which is where the extra output tokens go. | Mechanism confirmed |
| Say explicitly what it must not touch | It over-scopes, editing test files nobody asked about, on ordinary feature work. | Single source |
| Delete instructions written for older Claude models | They survive the upgrade and spend tokens solving problems 5.1 no longer has. | Single source |
| Give it one large task, not several small ones | On one team's client work it lost to Fable 5 on simple tasks and won the hardest by 40% on time at under half the cost. | Single source |
| Orchestrate with 5.1, implement with something cheaper | One published build had Fable at 3% of total spend, with sub-agents carrying the rest. | Single source |
The second row is the one to act on first, and it is also the explanation for the cost section below. A model that rewrites a whole file to change four lines emits the entire file as output tokens, and output is the expensive half of the bill at fifty dollars per million. The tip and the cost story are the same fact seen from two directions.
Notice what just happened to the source column. Four of six rows rest on one person. They are still worth trying, because the cost of testing a prompting practice is a single run, but a table that presented all six as settled would be lying about four of them. That is the same discipline we apply to model output itself. Paste a draft into TrueStandard and four frontier models from different vendors check it in parallel, and every place they disagree is shown, never averaged away.
The cost claim needs its denominator
Anthropic's launch page makes a specific promise. Fable 5.1 will cost an estimated 25% less than Fable 5 for typical workloads wherever usage is billed by token, and for highly agentic work the saving often reaches approximately 45%. The page says plainly where that comes from, which is the lower price on cache reads. Independent measurement then found cost per task went up. Both are true, and they measure different things.
| Who measured | What they found | What that number is about |
|---|---|---|
| Anthropic | 25% less, up to about 45% on agentic work | Per-token billing, where a cache read costs a quarter of what it did |
| Artificial Analysis | Cost per task rose about 20%, $3.14 to $3.76 | A fixed evaluation suite, where 5.1 emits roughly 1.7x the output of Fable 5 |
| Artificial Analysis | Total evaluation cost rose 56%, $5,455 to $8,523 | The same suite, whole-run total instead of per task |
Two things pull against each other. Cache reads got 75% cheaper, which helps any workload that re-reads a large stable context, so long-running agents and repeated passes over one codebase pay less. Output grew about 70%, which hurts any workload that generates a lot of new text. Artificial Analysis put numbers on the tug of war and found the cache saving worth roughly 1.40 dollars per task while the extra output outweighed it.
So the answer to whether Fable 5.1 is cheaper is that it depends which half of your bill is larger, and you already know which that is. The 56% figure circulating on Hacker News is real but it is the whole-suite total and not the per-task number, and quoting it beside Anthropic's 25% compares two different measurements as though one refutes the other. Neither side is lying. The claim just never travelled with its denominator attached.
The number everyone quoted wrong
Two independent reviewers reported the same pair of figures during launch week: a hallucination rate of 72.6% for Fable 5.1 against 63.6% for Fable 5, framed as the newer model guessing instead of admitting it does not know. The figures are real. The framing drops the part that reverses the conclusion.
| Model | Correct | Hallucinated | Declined |
|---|---|---|---|
| Claude Fable 5.1 | 67% | 24% | 9% |
| Claude Fable 5 | 65% | 22% | 13% |
Artificial Analysis figures, 1 September 2026, rearranged to sum to 100. Partial answers, about 2%, sit in Declined. The 72.6% and 63.6% that circulated count hallucinations against only what the model got wrong, not against everything it was asked.
The metric divides incorrect answers by everything the model did not get right, which is the incorrect answers plus the partial ones plus the times it declined. It asks a narrow question. Of the questions this model failed to answer correctly, how often did it produce a confident wrong answer instead of stepping back. Fable 5.1 attempts 93.4% of questions where Fable 5 attempted 87.8%, and it is the more accurate model of the two at 67.2% against 65.4%. It answers more, gets more right, and because it declines less often the bottom of that fraction shrinks.
That is worth converting before anyone repeats it. Take Fable 5.1's own published figures: 32.8% of questions came back non-correct, and 72.6% of those were confident wrong answers. So roughly 24% of all questions got a wrong answer stated as fact, against roughly 22% for Fable 5. The gap between the two models is about two points. The gap between 72.6% and how that number reads at a glance is about fifty.
Artificial Analysis put the attempt rate in the same sentence as the rate, and said the higher attempt rate was the reason. Both reviews carried the rate and dropped the reason.
That is somebody else's benchmark. We ran our own arm on the same model and asked a narrower question, whether it invents citations: Fable 5.1 invented 2 of the 144 DOIs it offered against Fable 5's 3 of 137, which is not a difference this test can separate.
We have written this exact mechanism before, running the other way. In our own sycophancy work, two models posted the best scores in the field by barely answering at all, because an unanswered item leaves the denominator and cannot be caught being wrong. Here the sign flips. A model that answers more is punished by the same arithmetic that rewarded the quiet ones. Any rate built as a fraction can be improved by refusing to play, and made to look worse by playing harder.
The honest sentence is longer than the one that travelled, which is why the short one won. Of the questions Fable 5.1 does not get right, it guesses instead of declining 72.6% of the time, up from 63.6%, while also attempting more questions and answering more of them correctly. Its composite score on that benchmark lands level with Fable 5, not below. If you need a model that says it does not know, that is a real regression worth planning around. If you read the bare rate as the model getting worse, you read it backwards.
What bites
Five things surfaced repeatedly enough to plan around. The first is a breaking change, the second is a billing surprise, and the third has five independent reports across two platforms behind it.
Context is now append-only
Preserved Thinking makes the context append-only, so code that edited or pruned conversation history no longer works the way it did. One developer said publicly it was enough to move their API usage elsewhere. If you trim history to control cost, this is a rewrite, not an upgrade.
A refusal arrives as a success
Blocked requests return HTTP 200 with a refusal stop reason, not an error status. A client that only inspects the status code will treat a refusal as a completed answer and pass an empty or truncated result downstream. Anthropic says refusal rates are materially higher here than on earlier models, so this path will be exercised.
Sessions get silently downgraded
Several users reported being moved to a weaker model mid-session with no notice, triggered by benign context. One review saw it on ordinary leukaemia and Alzheimer's questions. Another team had a game project downgraded for an entire session because a file contained the word biological. Five independent reports across two platforms, with enough people reporting no issues that it reads as inconsistent, not universal.
Sub-agent fan-out ignores soft limits
One developer had a configured ceiling of roughly fifteen agents ignored, turning 87 findings into 328 agents, of which 275 died against the session limit. Another reported 92% of a weekly quota burned by sub-agents failing and restarting with large contexts. Cap fan-out somewhere the model cannot reach.
The usage maths is contested and about to change
A proposed class action filed in June 2026 in the Northern District of California alleges the Max 5x and 20x plans deliver closer to 3.5x and 6x to 8x. It is filed and unadjudicated, so it is an allegation and not a finding. Separately and independently, the temporary weekly boost expired on 13 September 2026 and the permanent 25% increase that replaced it lands about 17% below the level users had, which Anthropic has itself acknowledged.
One more thing changed quietly and it is not a Fable 5.1 feature. Every Claude model released after 2 August 2026 now carries an invisible statistical watermark for EU AI Act compliance, which Anthropic says is undetectable without its detection API and carries no information about the user or their conversations. That sweeps in Fable 5.1 and Mythos 5.1 by release date, and it changes the answer our earlier post gave on whether Claude watermarks your writing.
None of this makes Fable 5.1 a bad model. Two of the five items are the price of tighter safety machinery, and the capability gain on long single-pass work is real and visible in every build people published. It does mean the launch post and the first week describe different products, and only one of them is the one you will operate.
Frequently asked questions
What is Claude Fable 5.1?
A frontier model Anthropic released on 1 September 2026, available as claude-fable-5-1. It carries a one million token context window, 128,000 tokens of maximum output and a June 2026 knowledge cutoff, and it costs 10 dollars per million input tokens and 50 dollars per million output. Adaptive thinking is always on and cannot be turned off, with effort levels from low through max and a default of high. A restricted sibling called Mythos 5.1 is the same model with different safeguards, gated behind a verification programme.
Is Claude Fable 5.1 cheaper than Fable 5?
Per token, yes, and only through one line. Input and output prices are unchanged at 10 and 50 dollars per million. Cache reads fell to 25 cents per million, which is 2.5% of base input where other current Claude models sit at 10%. Anthropic estimates 25% lower cost for typical workloads and up to about 45% for agentic work. Independent measurement on a fixed evaluation suite found cost per task rose about 20%, because the model emits roughly 1.7 times the output of Fable 5. Cache-heavy workloads get cheaper and output-heavy ones get dearer.
Does Claude Fable 5.1 hallucinate more than Fable 5?
Its hallucination rate on the AA-Omniscience benchmark rose from 63.6% to 72.6%, but that metric counts only the questions the model did not answer correctly and asks how often it guessed instead of declining. Over the same test Fable 5.1 attempted more questions, 93.4% against 87.8%, and was more accurate, 67.2% against 65.4%. Its composite score lands level with Fable 5. The regression is real if you need a model that admits uncertainty, and the bare rate overstates it badly.
Is Claude Fable 5.1 good for coding?
It is strongest on long, hard, single-pass work and weaker on small edits. One team found it lost to Fable 5 on simple tasks and won the hardest task by 40% on time at under half the cost. It also defaults to rewriting whole files for small changes, so tell it to edit only the span that changes or you pay for the rest of the file in output tokens.
What effort level should I use with Claude Fable 5.1?
Start low and raise it only when the output forces you to, which is the opposite of the instinct. Code Rabbit measured low effort catching 61% of real problems against 57.1% at high, while running faster. The default is high, so this is a setting worth changing deliberately instead of leaving alone.
Why does Claude Fable 5.1 return an empty response?
Most likely a refusal. Blocked requests come back as HTTP 200 carrying a refusal stop reason and a category, with no error status, and Anthropic states refusal rates are materially higher on this model than on previous ones. Client code that only checks the status code will read that as success. Handle the refusal stop reason as a primary response path. Refusals caught before inference are not billed, and ones caught mid-answer bill for the tokens already produced.
What is Preserved Thinking in Claude Fable 5.1?
A change that makes the conversation context append-only, so code that edited or pruned history no longer behaves as it did. It is the breaking change in this release for anyone managing context manually, and at least one developer has said publicly it was enough to move their API usage off the platform.
Does Claude Fable 5.1 watermark its output?
Yes, and so does every Claude model released after 2 August 2026, as EU AI Act compliance, and not as a Fable 5.1 feature. Anthropic describes it as invisible to anyone without its detection API, with no effect on output quality and no information about the user, their organisation or their conversations. A detection API is in private preview for regulators, law enforcement, media and researchers.
Which Claude plan includes Fable 5.1?
In the Claude apps it requires the 100 dollar per month Max plan, and the Pro plan does not include it. It is also available through the API, Amazon Bedrock, Google Cloud Vertex and Microsoft Foundry. Note that a permanent 25% weekly limit increase took effect on 14 September 2026, replacing a temporary boost that expired the day before, which nets out about 17% below the limits users had immediately prior.
Is Claude Fable 5.1 worth upgrading to in 2026?
If your work is long single-pass reasoning over large contexts, yes, and the week-one builds support that. If it is short iterative edits, the gain is small and the extra output tokens cost real money. If you manually prune context, budget for a rewrite because of Preserved Thinking. If you depend on the model declining when unsure, test that yourself first, because it attempts noticeably more than Fable 5 did.
Keep reading
Is There a Most Accurate AI Model?
The honest answer is no. The ranking changes with the task, the benchmark, and the month, and even the leader still hallucinates.
How Accurate Is ChatGPT?
Accurate enough to trust for everyday questions, and wrong often enough to get you sued if you publish it unchecked. Here is what the measurements actually say, and what to do about it.
Why AI Is Confidently Wrong
Models sound certain every time, even when wrong. The confident tone you trust in people is worthless here. Here is the fix.
AI Detectors Ask the Wrong Question
They flag honest writers, clear famous human documents as AI, and OpenAI quietly killed its own detector. But the deeper problem is not that detection is unreliable. It is that 'was this written by AI?' was never the question that protects you.
Should You Stop Using ChatGPT?
Researchers found AI made experts measurably worse on hard tasks. Here is when to trust ChatGPT, and when it is just telling you what you want to hear.
Two of this week's most repeated numbers meant the opposite of how they read.
Both survived because nobody checked them against their own source. TrueStandard runs your draft past four frontier models from different vendors at once and shows you every place they disagree, before it ships rather than after.
Start Checking →