Published June 10, 2026
Claude Fable 5 is Anthropic's new frontier model, generally available since June 9, 2026, priced at $10 per million input tokens and $50 per million output, with a 1M-token context window and 128k max output. It sits above the Opus 4.8 / Sonnet 4.6 / Haiku 4.5 tier as Anthropic's most capable widely released model. Here is what it actually costs, what it does, how it compares, and the caveats that matter before you move production traffic onto it.
For the wider landscape see Claude, GPT, Gemini compared and Free LLM APIs in 2026.
What is Claude Fable 5?
Claude Fable 5 (API model ID claude-fable-5) is the new top of Anthropic's lineup, launched June 9, 2026 as a generally available model. Anthropic describes it as a "Mythos-class model made safe for general use" and its "most capable widely released model," positioned above the Opus, Sonnet, and Haiku tiers rather than replacing them.
It takes text and image input and returns text. Adaptive thinking is always on, and unlike earlier Claude models it does not expose an extended-thinking toggle, the raw chain of thought is never returned. It supports vision, the memory tool, context editing, and the effort parameter at launch.
How much does Claude Fable 5 cost?
Pricing per million tokens, verified 2026-06-10 against Anthropic's pricing page:
| Usage | Price (per MTok) |
|---|---|
| Input (base) | $10 |
| Output (base) | $50 |
| Cache read (hit) | $1 |
| Cache write (5-min) | $12.50 |
| Cache write (1-hour) | $20 |
| Batch input (50% off) | $5 |
| Batch output (50% off) | $25 |
The full 1M-token context is billed at the standard rate, there is no long-context surcharge. One caveat on cost: Fable 5 uses the Opus 4.7 tokenizer, so the same text produces roughly 30 to 35% more tokens than older Claude models. Your effective bill per page of text is higher than the sticker price suggests, budget for it before you switch.
Claude Fable 5 vs Opus 4.8, Sonnet 4.6, Haiku 4.5
Fable 5 is the frontier option, but it is not the right default for most work. Opus 4.8 remains Anthropic's recommended model for complex tasks at half the price.
| Model | API ID | Input | Output | Context | Max output |
|---|---|---|---|---|---|
| Claude Fable 5 | claude-fable-5 | $10 | $50 | 1M | 128k |
| Claude Opus 4.8 | claude-opus-4-8 | $5 | $25 | 1M | 128k |
| Claude Sonnet 4.6 | claude-sonnet-4-6 | $3 | $15 | 1M | 64k |
| Claude Haiku 4.5 | claude-haiku-4-5 | $1 | $5 | 200k | 64k |
Reach for Fable 5 on the hardest long-horizon work, large software migrations, deep research, vision-heavy analysis, where top capability earns its 2x-Opus price. For everyday coding, agents, and chat, Opus 4.8 or Sonnet 4.6 is the better cost-to-capability call.
How do you access Claude Fable 5?
Fable 5 is generally available from launch day on:
- The Claude API
- Amazon Bedrock (
anthropic.claude-fable-5) - Google Vertex AI (
claude-fable-5) - Microsoft Foundry
It is also included at no extra cost on Pro, Max, Team, and seat-based Enterprise plans during a promotional window, June 9 to 22, 2026. Note that Fable 5 is a Covered Model with 30-day data retention, it is not available under zero-data-retention terms.
What about Claude Mythos 5?
Anthropic launched a sibling, Claude Mythos 5 (claude-mythos-5), alongside Fable 5. It is the same underlying model with the safety classifiers lifted in some areas. Mythos 5 is not generally available, it is invitation-only through Project Glasswing, aimed at cyberdefenders and infrastructure providers, with limited availability on Bedrock and Vertex via account teams. If you are a normal developer, Fable 5 is the version you can actually use.
Is Claude Fable 5 worth switching to?
For most teams, not yet as a default. Three reasons to wait before you move core traffic:
- Anthropic published almost no hard benchmark numbers at launch. The only specific figure stated is that Fable 5 is the first model to break 90% on Anthropic's internal analytics benchmark. Everything else is qualitative ("state-of-the-art," "highest"), there are no official SWE-bench or GPQA scores yet to compare against GPT or Gemini.
- Always-on adaptive thinking with no toggle means less control over latency and token spend than you have on Opus.
- It ships with safety classifiers. A refused request returns
stop_reason: "refusal"(still HTTP 200), and Anthropic says safeguards trigger in under 5% of sessions on average. Afallbacksparameter can re-serve those on another model such as Opus 4.8, which is worth wiring up before you depend on Fable 5.
In practice: keep Opus 4.8 as your default, route only your hardest tasks to Fable 5, and measure whether the extra capability justifies the price and the tokenizer overhead on your own workload.
Frequently asked questions
What is Claude Fable 5's API model ID? claude-fable-5. It uses a dateless ID that is also a pinned snapshot, so there is no separate dated version string.
How much does Claude Fable 5 cost? $10 per million input tokens and $50 per million output. Cache hits are $1/MTok, and the Batch API halves the rate to $5 input / $25 output. The 1M context is billed at the standard rate.
Is Claude Fable 5 better than Claude Opus 4.8? Anthropic positions Fable 5 as its most capable widely released model, above the Opus tier. But Opus 4.8 is half the price and is still the recommended default for complex tasks. Fable 5 is for the hardest frontier workloads.
How is Claude Fable 5 different from Claude Mythos 5? Mythos 5 is the same model with safety classifiers lifted in some areas. Fable 5 is the version made safe for general use and is generally available, Mythos 5 is invitation-only via Project Glasswing.
Does Claude Fable 5 support extended thinking? No toggle. Adaptive thinking is always on and cannot be disabled, and the raw chain of thought is never returned.
Is Claude Fable 5 free? The API is paid. Fable 5 is included at no extra cost on Pro, Max, Team, and seat-based Enterprise plans during the June 9 to 22, 2026 promotional window.
