 |
VOOZH | about |
|
VOOZH | about |
On June 9, 2026, Anthropic released Claude Fable 5, the first model in a brand new top tier the company calls the Mythos class. This is not another point upgrade in the Opus line.
Anthropic describes Fable 5 as a model that sits above Opus in raw capability. It is state of the art on nearly every benchmark it was tested on, and it is the most powerful model the company has ever made available to the general public.
Introducing Claude Fable 5: a Mythos-class model that we’ve made safe for general use.
— Claude (@claudeai) June 9, 2026
Its capabilities exceed those of any model we’ve ever made generally available. pic.twitter.com/2AvmEjHIX8
It launched alongside a sibling model, Claude Mythos 5. That model is the same underlying system with the safety guardrails lifted, available only to a small set of vetted users. Fable 5 even became a punchline in the viral Le Chaton Fat meme, where users joked you would not need Fable if Mistral shipped a fat kitten model.
The new tier is the real headline. For two years Anthropic ran a three step ladder of Haiku, Sonnet, and Opus. Mythos is a fourth step above all of them, and Fable 5 is the version of that step the public can actually use.
The timing is its own story. Fable 5 arrived only days after Anthropic warned that frontier AI is approaching recursive self improvement and urged the industry to agree on a coordinated brake on development. Releasing your most powerful model yet, days after that warning, struck a lot of observers as either a contradiction or a carefully chosen strategy.
In this article we cover what Fable 5 is, how the Mythos class works, the full benchmark picture against Opus 4.8, GPT-5.5, and Gemini 3.1 Pro, the safeguard system, pricing, availability, and how the restricted Mythos 5 model fits in.
Claude Fable 5 is a frontier reasoning model and the first generally available member of Anthropic’s Mythos class. Here are the basics at a glance:
claude-fable-5, with anthropic.claude-fable-5 and global.anthropic.claude-fable-5 on AWSThe most important fact is that Fable 5 and Mythos 5 are the same model. The difference is not the weights or the training. It is the safety layer wrapped around Fable 5 that intercepts high risk requests. So every benchmark number Fable 5 posts is the Mythos class capability operating with a safety net underneath it.
Anthropic positions the model for autonomous work that runs for hours or even days inside an agent harness. In that setting the model plans a multi step job, calls tools, reads the results, validates its own output, and corrects course without a human in the loop.
The trait partners keep citing is self verification. Rakuten told Anthropic the model reflects on and validates its own work, which is what makes the autonomous operation practical rather than risky.
On context size, the picture firmed up after launch. Independent benchmarking site Artificial Analysis lists a 1 million token window, matching the figure that circulated in early summaries. The companion 128,000 token maximum output number is still not confirmed in Anthropic’s own materials, so treat the output ceiling as unverified for now.
Anthropic’s lineup has always been a ladder:
Mythos represents a capability level Anthropic had been holding back from general release, because of the risks a model that strong could pose in the wrong hands.
Fable 5 is what you get when Anthropic takes a Mythos class model and adds the guardrails needed to make it safe for a broad audience. The pitch is that the safeguards are not a tax on the model. They are the thing that makes shipping it possible at all.
That framing matters for the price. You are not paying twice the cost of Opus 4.8 for a slightly better Opus. You are paying for a step up to a new tier, delivered with a safety system the older tiers never needed.
Mythos did not appear out of nowhere on launch day. The first Mythos model, Claude Mythos Preview, shipped quietly in April 2026 through a limited program named Project Glasswing. An early Mythos build later made headlines when it bypassed Apple’s M5 memory protections in five days.
Project Glasswing was not a public product. It was a controlled access program aimed at a narrow set of users, mainly cyber defenders and critical infrastructure providers. These are organizations that could use a frontier model’s offensive security knowledge to strengthen their own defenses.
Anthropic used the program to learn how a Mythos class model behaves in the field before deciding whether to release it more widely. The June 9 launch is the graduation of that experiment into two official products:
Anthropic has also signaled that the trusted access program for vetted cybersecurity organizations will broaden, and that enrollment for a biology research program is beginning.
Anthropic published a full benchmark table with the launch, putting the Mythos class against Opus 4.8, GPT-5.5, and Gemini 3.1 Pro. Launch coverage from VentureBeat and Tom’s Hardware echoed the topline that it is state of the art on nearly all of them.
Two things matter for reading the table. The Anthropic column combines Mythos 5 and Fable 5, showing the higher of the two, which are normally within 1 to 3 points of each other. The starred rows are the exception. On those, Fable 5’s safeguards route requests to Opus 4.8, so the public Fable 5 score lands closer to Opus 4.8 while the starred figure reflects the unrestricted Mythos 5.
| Benchmark | Mythos 5 / Fable 5 | Mythos Preview | Opus 4.8 | GPT-5.5 | Gemini 3.1 Pro |
|---|---|---|---|---|---|
| Agentic coding (SWE-Bench Pro) | 80.3% | 77.8% | 69.2% | 58.6% | 54.2% |
| Agentic coding (FrontierCode Diamond) | 29.3% | — | 13.4% | 5.7% | — |
| Knowledge work (GDPval-AA) | 1932 | — | 1890 | 1769 | 1314 |
| Knowledge work vision (GDPpdf, no tools) | 29.8% | — | 22.5% | 24.9% | 16.7% |
| Spatial reasoning (Blueprint Bench 2) | 38.6% | — | 14.5% | 36.2% | 26.5% |
| Tool use (AutomationBench) | 17.4% | — | 15.5% | 12.9% | 9.6% |
| Computer use (OSWorld Verified) | 85.0% | 85.4% | 83.4% | 78.7% | 76.2% |
| Legal (Legal Agent Benchmark) | 13.3% | — | 10.4% | 2.1% | 0.0% |
| Multidisciplinary reasoning (Humanity’s Last Exam, no tools) | 59.0% | 56.6% | 49.8% | 41.4% | 44.4% |
| Humanity’s Last Exam (with tools) | 64.5%* | 64.7% | 57.9% | 52.2% | 51.4% |
| Biology hard (BioMysteryBench) | 46.1%* | 29.6% | 40.0% | not listed | not listed |
| Biology human solved (BioMysteryBench) | 83.9%* | 82.6% | 80.4% | not listed | not listed |
| Agentic coding (Terminal Bench 2.1) | 88.0%* | — | 82.7% | 83.4% | 70.7% |
| Cybersecurity (ExploitBench, capture %) | 78.0%* | 69.0% | 40.0% | 34.0% | not listed |
| Health (HealthBench Professional) | 66.0%* | 64.7% | 56.9% | 51.8% | not listed |
Starred rows are where Fable 5’s safeguards trigger a fallback. The number shown is Mythos 5, and public Fable 5 performs closer to Opus 4.8 on those tasks. On Terminal Bench 2.1, GPT-5.5’s 83.4% is via Codex CLI and Gemini’s 70.7% via Gemini CLI.
The standouts are the unstarred rows, where Fable 5 and Mythos 5 are effectively the same. Fable 5 leads every rival on agentic coding, knowledge work, vision, tool use, computer use, and legal reasoning. The SWE-Bench Pro gap is the clearest signal, with Fable 5 at 80.3% sitting more than 11 points above Opus 4.8 and nearly 22 above GPT-5.5, a margin that compounds across every step of a long autonomous job. Even the strongest open-weight coders, like Nex-N2-Pro at 58.8% on SWE-Bench Pro, sit far behind Fable 5’s 80.3% here. On the hardest FrontierCode Diamond split it more than doubles Opus 4.8 and roughly quintuples GPT-5.5.
A few results outside the official table round out the picture:
Stripping the benchmarks down to positioning, the four models line up like this.
| Attribute | Claude Fable 5 | Claude Opus 4.8 | GPT-5.5 | Gemini 3.1 Pro |
|---|---|---|---|---|
| Tier | Mythos class | Opus class | flagship | flagship |
| Benchmark standing | leads most categories | closest follower | mid pack | trails |
| Input price per million tokens | $10 | $5 | varies | varies |
| Output price per million tokens | $50 | $25 | varies | varies |
| Built in hard safeguards | yes | no | no | no |
The pattern from the benchmark table is consistent. Fable 5 leads on the unsafeguarded categories, with Opus 4.8 as the closest follower. GPT-5.5 sits a step behind on coding but stays competitive on vision and spatial reasoning, and Gemini 3.1 Pro trails across the board.
The tradeoff is price and the safety layer. Fable 5 costs twice what Opus 4.8 does and routes certain requests away from itself, which none of the other models here do. If you are weighing the two biggest labs more broadly, our Anthropic vs OpenAI breakdown covers pricing, safety, and ecosystem beyond a single model.
The first independent confirmation came from Artificial Analysis, which folded Fable 5 into its public leaderboard within a day of launch. Fable 5 debuted at the very top.
The headline numbers:
The independent data also surfaced weaknesses the partner testimonials gloss over. The biggest one is speed:
That latency is a direct consequence of the heavy chain of thought reasoning the model runs before answering. This is a model built for deep, long horizon work, not snappy back and forth chat.
Cost is the other caveat. Artificial Analysis lists Fable 5 at the expensive end of its chart, quoting an input rate of $12.50 per million tokens against the $10 in Anthropic’s own materials, with a blended rate around $8.20 once cache hits are factored in. Either way, it is among the priciest models on the board.
One scoreboard is still blank. Fable 5 has not yet appeared on LMArena’s Chatbot Arena leaderboard, which ranks models by head to head human preference votes. That is expected for a model this new, since Arena needs a large volume of comparisons before assigning a reliable rating. A human preference ranking will be one of the more interesting data points to watch in the coming weeks.
Benchmarks are one signal. The launch partner reports point in the same direction.
Stripe produced the most cited result. Anthropic says Fable 5 compressed months of engineering into days. The specific example is a migration across a 50 million line Ruby codebase that the model completed in a single day, work that would otherwise have taken a full engineering team over two months.
Rakuten emphasized reliability over raw speed. The company reported that Fable 5 reflects on and validates its own work, letting it run autonomous operations where the model is trusted to check itself rather than handing every step back to a human.
Hex contributed the 90% analytics milestone, the first time any model cleared that bar on its suite of complex tasks. AWS framed the model as purpose built for long running, asynchronous execution, the kind of job that can run for days inside a harness before producing a result.
The safeguards are what make Fable 5 a public product, so they deserve a close look.
Fable 5 ships with classifiers covering four high risk domains. When a request trips one, the model blocks its own response and falls back to Claude Opus 4.8, the older and more conservative model, to handle the query safely. The domains are:
The key number is how often this happens. Anthropic says more than 95% of Fable 5 sessions involve no fallback at all. Only around 5% hit a classifier and defer to Opus 4.8, so for the overwhelming majority of normal use you are talking to the full Mythos class model.
The red team results are the other half of the safety case:
There is a cost dimension too. When a request is routed to Opus 4.8, you pay Opus prices for that portion. Anthropic also requires a mandatory 30 day retention of inputs and outputs for all users, with human review capability, which it frames as a defense against novel attacks. On AWS this is enforced through a provider_data_sharing setting that must be enabled before the model can be invoked.
The starred rows in the benchmark table above are exactly where the two models diverge. On those, Mythos 5 answers at full capability while Fable 5’s safeguards route the request to Opus 4.8, so the public model performs closer to Opus 4.8 than to the starred score. The gap is widest in the high risk domains:
A general user hitting these topics through Fable 5 is redirected to Opus 4.8, which is why public Fable 5 lands near the Opus column on the starred rows. Mythos 5, available only to vetted cyber defenders and approved biology researchers, is the version that answers in full.
This is the core of Anthropic’s argument for the whole release. A model this capable in cybersecurity and biology is genuinely dangerous in the wrong hands, so that capability is locked behind trusted access while Fable 5 lets everyone else use the same intelligence for everything that is not high risk.
Claude Fable 5 and Claude Mythos 5 are both priced at $10 per million input tokens and $50 per million output tokens.
For context, Claude Opus 4.8 costs $5 and $25, which makes Fable 5 exactly twice the price of the model just below it. Anthropic prefers to compare up rather than down. Against the earlier Claude Mythos Preview, Fable 5 costs less than half as much, so within the Mythos class the price has dropped sharply even as capability went up.
Whether the premium is worth it depends on the work. For routine chat, Opus 4.8 or a cheaper tier is the obvious pick. For long autonomous jobs where a higher pass rate compounds across hundreds of steps, like the Stripe migration, the premium can pay for itself by finishing tasks a cheaper model would fail.
Anthropic shipped Fable 5 across a wide set of platforms on launch day.
There is a rollout window worth noting. Through June 22, 2026, Fable 5 is included at no extra cost on Pro, Max, Team, and seat based Enterprise plans. From June 23, access shifts to a usage credit model, so the free inclusion is effectively a two week introductory period.
Claude Mythos 5 is the exception to all this breadth. It is restricted to Project Glasswing partners and a small set of approved biology researchers, and it is not part of any general subscription or public API tier.
The release cannot be separated from its context. Just days before launching Fable 5, Anthropic warned that AI systems are advancing toward recursive self improvement, the point at which models can improve themselves without human intervention. The company paired that warning with a call for major AI labs to agree on a coordinated brake on frontier development.
Then it shipped its most powerful model yet. The apparent contradiction was not lost on anyone covering the launch. Releasing a Mythos class model to the public, days after arguing that AI is becoming too dangerous, reads as either a reversal or a deliberate strategy.
Anthropic’s answer is that the safeguards are the entire point. The argument runs that a Mythos class capability is coming whether Anthropic ships it or not, so it is better for the first public model at this tier to arrive with hard guardrails, fallback routing, and mandatory monitoring than to wait for a competitor to release something equally capable with no protections.
Whether that holds up is fair to debate. The safeguards are real and the red team numbers are strong, but they depend on classifiers that have to catch every dangerous request, data retention that raises privacy questions, and a fallback model users do not control. The launch is a live test of whether a model this capable can be released responsibly.
A few things are not fully nailed down yet:
None of this undercuts the headline. On the benchmarks Anthropic did publish, and on Artificial Analysis’s independent leaderboard, Fable 5 is the strongest model the company has released to the public. The full picture will sharpen as the model card and more independent evaluations land.
Three things make this launch significant beyond the benchmark numbers.
A new tier, public on day one. For two years the frontier conversation has been about incremental gains within the Opus, GPT, and Gemini flagship lines, the models we track in our best AI models guide. Mythos is a deliberate step above that ladder, and Fable 5 makes it generally available rather than locking it to a research preview.
A safeguard template the industry will study. Anthropic is betting that the way to ship a dangerous capability is to wrap it in classifiers, fallback routing, red teaming, and monitoring, then release the safe version broadly while the unrestricted version stays behind a trusted access wall. If it works, it becomes the playbook. If it fails, it becomes the cautionary tale.
A coding gap big enough to change what agents can do. An 80.3% SWE-Bench Pro pass rate and a Stripe migration finished in a day point to a model that can carry long, complex jobs end to end rather than assisting step by step. That is the difference between a coding assistant and an autonomous engineer.
For anyone who wants to see Claude together with other AI models, Fello AI puts Claude, GPT, Gemini, DeepSeek, Perplexity, and more into a single native app for Mac, iPhone, and iPad, so you can test the same task across models and pick the right one for the job without managing a stack of subscriptions.
Stay ahead with expert AI insights trusted by top tech professionals!
Join thousands of AI fans & professionals benefiting from exclusive tips and insights from industry leaders.
