Anthropic/Model/Claude/Fable/5
Key Features
- Anthropic's most powerful and most intelligent model; new tier above Opus
- Context window: 1M input tokens at standard API pricing
- Max output: 128K tokens on the synchronous Messages API
- Adaptive thinking only: uses
thinking: {"type": "adaptive"}; explicit thinking: {"type": "disabled"} returns a 400 — omit the thinking param entirely to run without thinking - Extended thinking with explicit
budget_tokens not supported (returns 400) - Sampling parameters (
temperature, top_p, top_k) removed — passing them returns 400 - Effort parameter:
low | medium | high | xhigh | max; default high; xhigh recommended for coding and agentic workloads - Task Budgets (beta,
task-budgets-2026-03-13): output_config: {task_budget: {type: "tokens", total: N}}; minimum 20,000 tokens - Thinking content omitted by default; set
thinking: {type: "adaptive", display: "summarized"} to restore visible reasoning - Model version status: Active (current flagship tier)
Performance Highlights
- Highest-capability tier in the Claude model family, exceeding Anthropic/Model/Claude/4.8/Opus across all evaluated dimensions
- Anthropic's most intelligent model
Model Tiers
- New flagship tier above the Opus line in the Claude model hierarchy
- Previous top tier: Anthropic/Model/Claude/4.8/Opus (
claude-opus-4-8) - Earlier Opus pages in this garden:
Access Details
- API availability: Claude Messages API model string
claude-fable-5 - Pricing (standard tier, per million tokens): $10 input · $50 output
- Batch API: 50% discount; prompt caching: up to 90% cost savings
- SDK support:
- Python:
pip install -U "anthropic" - TypeScript:
@anthropic-ai/sdk
Technical Specifications
- Model architecture: Proprietary Anthropic large language model (undisclosed weights)
- Input/output formats: Text and images in; text (and tool use) out
- Special capabilities:
- Adaptive thinking (only mode; omit
thinking key entirely to suppress — cannot pass {type: "disabled"}) - Task Budgets for agentic loop token awareness (beta)
- Effort parameter including
xhigh and max levels
- Model version code:
claude-fable-5
References