Anthropic released two new models today: Claude Fable 5 and Claude Mythos 5. If you follow AI at all, you’ve probably seen the announcement. If you don’t, here’s the short version, and more importantly, what it actually means for small businesses.
What Fable 5 is
Claude Fable 5 is Anthropic’s new flagship model, available to anyone through Claude.ai and the API. It replaces Claude Opus 4.8 as their most capable general-use model.
The headline claim: state-of-the-art across nearly every tested benchmark. Strongest performance in software engineering, knowledge work, vision tasks, and long-horizon reasoning. The longer and more complex the task, the bigger the gap between Fable 5 and everything that came before it.
Pricing landed at $10 per million input tokens and $50 per million output tokens. That’s less than half the price of the previous Mythos Preview model, which was restricted to a small group of cybersecurity researchers. For comparison, GPT-5.5 currently runs at similar or higher prices depending on the tier.
Claude Mythos 5 is the same underlying model but with fewer safety restrictions, available only to vetted cybersecurity researchers and government partners through Anthropic’s Project Glasswing program. Regular users don’t have access to it.
How it compares to the competition
Anthropic published benchmark numbers against Claude Mythos Preview, Claude Opus 4.8, GPT-5.5, and Gemini 3.1 Pro. Here’s what the table actually shows:
| Benchmark | Claude Fable/Mythos 5 | Claude Opus 4.8 | GPT-5.5 | Gemini 3.1 Pro |
|---|---|---|---|---|
| Agentic coding (SWE-Bench Pro) | 80.3% | 69.2% | 58.6% | 54.2% |
| Agentic coding (FrontierCode Diamond) | 29.3% | 13.4% | 5.7% | N/A |
| Knowledge work (GDPval-AA) | 1932 | 1890 | 1769 | 1314 |
| Knowledge work vision (GDP.pdf) | 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% | 83.4% | 78.7% | 76.2% |
| Legal (Legal Agent Benchmark) | 13.3% | 10.4% | 2.1% | 0.0% |
| Multidisciplinary reasoning (HLE, no tools) | 59.0% | 49.8% | 41.4% | 44.4% |
| Health (HealthBench Professional) | 64.7% | 56.9% | 51.8% | N/A |
| Cybersecurity (ExploitBench) | 78.0% | 40.0% | 34.0% | N/A |
A few things stand out. The coding gap is substantial: 80.3% vs GPT-5.5’s 58.6% on SWE-Bench Pro. The legal benchmark is almost embarrassing for the competition: 13.3% vs 2.1% for GPT-5.5 and 0.0% for Gemini. Knowledge work shows steady leads across every test.
One honest caveat: the cybersecurity and biology scores for Fable 5 have an asterisk. Fable 5 has safety classifiers that kick in on those topics and route to Opus 4.8 instead, so the Fable 5 scores on those benchmarks reflect the Mythos 5 numbers. For everyday use, the difference won’t matter.
The safety piece
This is worth understanding because it’s different from how most model releases work.
Anthropic built classifiers into Fable 5 that detect certain types of risky queries, primarily offensive cybersecurity and biology research, and automatically route those requests to Claude Opus 4.8 instead. You’re told when that happens. About 95% of sessions never trigger a fallback at all.
The idea is that the same capabilities that make Fable 5 exceptional at coding also make it dangerous in the wrong hands for cyberattacks. So instead of either refusing to release the model or releasing it with no guardrails, they built a switch.
Whether you find that reassuring or frustrating probably depends on what you do. For most small business use cases like writing, analysis, customer communication, and content drafts, you’ll never notice it.
What this actually means for small businesses
Honest answer: if you’re already using Claude, ChatGPT, or any AI tool for your business, today’s release is a modest upgrade, not a revolution.
The improvements that matter most for small business work are:
Writing and analysis are noticeably better. Longer documents, more nuanced drafts, fewer edits needed. If you use AI to draft emails, proposals, or blog content, Fable 5 will do it better.
Coding tasks are substantially better. If you have a developer or use AI to help with your website, automations, or app, the gap between Fable 5 and GPT-5.5 is real. For small businesses who use Claude Code or similar tools, this matters.
Long tasks stay on track. Fable 5 handles longer, multi-step work without losing context. If you’ve tried using AI for complex research or multi-page documents and found it fell apart halfway through, this model holds together better.
The price is competitive. $10/$50 per million tokens is roughly in line with current market rates for top-tier models, and cheaper than where Mythos Preview was.
What it doesn’t do: fix a broken business process, replace a skilled person, or make AI worth it if AI wasn’t already worth it for your situation. The tool got better. The strategy question is still yours to answer.
How to try it
If you have a Claude.ai subscription, Fable 5 is available now. Select it from the model picker.
If you use the API, the model ID is claude-fable-5-20260609.
If you’ve been on the fence about whether to try Claude at all, the free plan gives you limited access. It’s enough to get a feel for what the model does and whether it fits your workflow.
If you want to think through whether AI tools like this actually make sense for your business, the 90-second quiz is still the fastest way to find out. A better model doesn’t change the underlying question.
Questions about how AI fits into your business? Book the free 30-minute chat. I’ll give you a straight answer.
