Chat with Claude Fable 5 alongside Opus 4.8 and other top models — one workspace for research, coding, and knowledge work.
Start chatting with Fable 5
Context window, output limits, and API pricing for Fable 5 next to Claude Opus 4.8 and Sonnet 4.6 — public data, clearly dated and sourced.
| Spec | Fable 5 | Opus 4.8 · Sonnet 4.6 |
|---|---|---|
| Context window | 1M tokens | 1M · 1M |
| Max output | 128K tokens | 128K · 64K |
| Input price / 1M tokens | $10 | $5 · $3 |
| Output price / 1M tokens | $50 | $25 · $15 |
| Thinking | Adaptive only | Adaptive |
| Tier | Flagship — above Opus | Opus · Sonnet |
Source: Anthropic public pricing and model documentation. Data as of June 10, 2026. Anthropic Docs
A plain-language guide to Anthropic’s Fable 5 model: what it is, what it costs, how to access it, and how to prompt it well. Facts dated June 10, 2026.
Claude Fable 5 is Anthropic's flagship AI model, released on June 9, 2026. Anthropic describes it as a Mythos-class model made safe for general use: its capabilities exceed any model the company has previously made generally available, with state-of-the-art results in software engineering, knowledge work, vision, and scientific research. New safeguards block responses in specific high-risk areas, and a small share of sensitive queries are answered by Claude Opus 4.8 instead.
Fable 5 offers a 1M-token context window and up to 128K output tokens. API pricing is $10 per million input tokens and $50 per million output tokens — double Opus 4.8 — with a 90% input discount on cached prompt content. The model uses adaptive thinking only: classic sampling knobs like temperature and top_p are removed, so output style is steered through prompting, and an effort parameter (low to max) trades depth against speed and cost.
Official channels: the Claude apps on paid plans (included on Pro, Max, Team, and seat-based Enterprise until June 22, 2026, then via usage credits), the Claude API under the model ID claude-fable-5, AWS, Google Cloud, and Moely AI. Routers such as OpenRouter list it as anthropic/claude-fable-5.
Fable 5 follows instructions more literally than older models, which rewards precise prompts: state the role, one line of context, the task, the output format, and hard constraints. For long agentic tasks, put the full specification in the first message rather than drip-feeding it. Because sampling parameters are gone, ask for variation explicitly when you want it.
Start free and upgrade when you need more. You can cancel anytime.
Credit cards, Alipay, and WeChat Pay
For everyday tasks—chat without limits
Do more with advanced intelligence
For more capacity to stay productive
By subscribing, you agree to our Terms of Service
Join the community of researchers and developers independently exploring Fable 5 AI capabilities.
