Anthropic has launched Claude Sonnet 5, a faster, cheaper model that brings near-flagship agentic performance to everyone, while its most powerful model, the Mythos-class Claude Fable 5, has returned to general availability. In short: Sonnet 5 is the new everyday workhorse, and Fable 5 is the frontier heavyweight for the hardest, longest tasks. Together they mark one of the clearest snapshots yet of how fast capable AI is becoming both cheaper and more powerful at the same time.
It is a genuinely interesting moment, because these two releases pull in complementary directions — one makes strong AI affordable and ubiquitous, the other pushes the ceiling of what a model can do. Here is what is new in each, and how to think about which one fits a given job.
What’s new in Claude Sonnet 5
Launched on June 30, 2026, Claude Sonnet 5 is built to be the most agentic Sonnet model yet. That means it can make multi-step plans, use tools like browsers and terminals, and run autonomously in ways that, just a few months ago, required larger and more expensive models. It also verifies its own output without being explicitly told to, follows instructions more reliably, and searches across large codebases more effectively.
The headline is that its performance lands close to Anthropic’s top-tier Opus 4.8, but at a much lower price. Compared with Sonnet 4.6 from February, it shows clear gains in reasoning, tool use, software coding, and knowledge work. It is now the default model on the Free and Pro plans and is available to Max, Team, and Enterprise users as well.
Sonnet 5 pricing and safety
Sonnet 5 launched with introductory pricing of $2 per million input tokens and $10 per million output tokens through August 31, 2026, after which it moves to $3 per million input and $15 per million output. That is aggressive pricing for a model this capable, aimed squarely at making agents cheaper to run at scale.
On safety, Anthropic says Sonnet 5 shows a lower rate of undesirable behaviors — things like cooperating with misuse or being deceptive — than its predecessor. It is also better at refusing malicious requests and resisting prompt-injection attacks, which matters more than ever as models are handed real tools and left to act on their own.
What’s new in Claude Fable 5
Claude Fable 5 sits at the opposite end of the range. It is a Mythos-class model — Anthropic’s most capable tier — made safe for general use, and its capabilities exceed those of any model the company had previously released broadly, with state-of-the-art results across software engineering, knowledge work, vision, and scientific research.
The standout feature is stamina. Fable 5 can tackle days-long, complex, asynchronous tasks that earlier models simply could not sustain. Run inside an agent harness like Claude Code, it can work for days at a stretch — planning across stages, delegating to sub-agents, and checking its own work as it goes. That is a different kind of tool from a chatbot you prompt one question at a time.
Fable 5’s strengths and safeguards
Fable 5 is especially strong in three areas. In coding, it handles ambitious projects like large migrations and multi-day autonomous sessions, writing its own tests and implementing designs with high fidelity. In vision, it reads diagrams, charts, and tables nested inside files and PDFs, which opens up document-heavy work in finance, legal, analytics, and architecture. In knowledge work, it can carry complex, multi-stage projects from research to review-ready deliverables with minimal oversight.
It also ships with notable safeguards: queries on sensitive topics such as cybersecurity, biology, chemistry, and health are quietly routed to Claude Opus 4.8 instead, a fallback that Anthropic says triggers in under 5% of sessions on average. Fable 5 recently returned to general availability after the US lifted export controls that had been linked to jailbreak concerns, and it is now offered on the Claude API, AWS, Amazon Bedrock, Google Cloud, and Microsoft Foundry. Its pricing reflects its position at the top: $10 per million input tokens and $50 per million output, with a 90% input discount for prompt caching.
Sonnet 5 vs Fable 5: how they compare
| Factor | Sonnet 5 | Fable 5 |
|---|---|---|
| Role | Efficient everyday default | Frontier flagship |
| Best at | Fast, affordable agentic work | Long, complex, multi-day tasks |
| Performance | Near Opus 4.8 | State-of-the-art, Mythos-class |
| Autonomy | Strong multi-step tasks | Days-long autonomous sessions |
| Input price (per M tokens) | $2 (intro), then $3 | $10 |
| Output price (per M tokens) | $10 (intro), then $15 | $50 |
| Availability | Default on Free and Pro | API and major clouds |
Which one should you use?
For the vast majority of tasks — drafting, research, coding help, everyday agents — Sonnet 5 is the sensible default. It is fast, remarkably capable for the price, and now built into the free experience, so most people will use it without thinking about it. If you want to get productive with any of these models, our guide on how to use an AI chatbot applies just as well to Claude.
Reach for Fable 5 when the job is genuinely hard or long: a sprawling code migration, a multi-day research project, or document-heavy analysis that would overwhelm a smaller model. You pay considerably more per token, but for work that would otherwise take a person days, that can be a bargain. The rise of models that plan and act on their own is exactly the shift we explored in our explainer on what an AI agent is.
The bigger picture
What strikes me about this pair of releases is the direction of travel. A year ago, the capabilities now sitting in a cheap default model were reserved for the most expensive tier. Sonnet 5 pushing near-Opus performance down to a couple of dollars per million tokens tells you how quickly the floor is rising, while Fable 5 shows the ceiling climbing just as fast toward autonomous, days-long work. This is the same broad wave driving the boom in free everyday assistants we covered in our roundup of free AI tools. For the primary sources, see Anthropic’s announcements for Claude Sonnet 5 and Claude Fable 5 and Mythos 5.
Frequently asked questions
What is the difference between Claude Sonnet 5 and Fable 5?
Sonnet 5 is the efficient, affordable everyday model with near-Opus performance, now the default on Free and Pro. Fable 5 is Anthropic’s frontier Mythos-class model built for the hardest, longest, multi-day tasks, at a much higher price.
How much does Claude Sonnet 5 cost?
It launched at an introductory $2 per million input tokens and $10 per million output tokens through August 31, 2026, then rises to $3 input and $15 output. It is also included as the default in the Free and Pro plans.
What makes Fable 5 special?
Fable 5 can run autonomously for days — planning across stages, delegating to sub-agents, and checking its own work — with state-of-the-art results in coding, vision, and knowledge work. It is Anthropic’s most capable generally available model.
Is Claude Sonnet 5 free to use?
Yes. Sonnet 5 is the default model on Anthropic’s Free plan, so most users can access it at no cost, with higher limits and additional models available on paid plans.
Why was Fable 5 unavailable before?
Fable 5 returned to general availability after the US lifted export controls that had been linked to jailbreak concerns. It now ships with safeguards that route sensitive queries to Claude Opus 4.8 in a small share of sessions.
Sonnet 5 and Fable 5 capture this AI moment neatly: powerful capability is getting dramatically cheaper, and the frontier is racing ahead toward genuinely autonomous work. For most people, Sonnet 5 is the new default worth reaching for — and when a task is big enough to need it, Fable 5 is waiting at the top.
Sofia follows emerging technology, from AI and VR to IoT and blockchain, and translates the hype into plain language. She cares about what these tools mean for everyday users, not just the headlines.