Transparent pricing with no hidden fees. Pay as you go.
claude-fable-5-1claude-fable-5-1claude-fable-5-1| Model | Spec | PoYo price | Official price | You save |
|---|---|---|---|---|
| claude-fable-5-1 | Input | $8.00/1M tokens 1600 credits /1M tokens | $10.00/1M tokens Official | 20% |
Cache read | $0.200/1M tokens 40 credits /1M tokens | $0.250/1M tokens Official | 20% | |
Output | $40.00/1M tokens 8000 credits /1M tokens | $50.00/1M tokens Official | 20% |
* Actual fees are based on the final output.
Complete guide to using Claude Fable 5.1 API for Long-Horizon Coding and Knowledge Work
Claude Fable 5.1 is Anthropic's most capable generally available model, released on September 1, 2026. It is built for demanding reasoning, multi-day agentic coding, multistep research, and document, spreadsheet, and slide work.
claude-fable-5-1. Anthropic positions Fable 5.1 for demanding reasoning and long-horizon agentic work when Claude Opus 5 still falls short.Official Anthropic launch materials show a large jump over Fable 5 on agentic science, coding, computer use, and knowledge work, plus 75% cheaper cache reads.
01
Anthropic reports Claude Fable 5.1 at 52.6% on Terminal-Bench-Science 0.1, versus 24.7% for Fable 5, 29.0% for Opus 5, and 22.4% for GPT-5.6 Sol. It also leads the published Fable 5.1 table on Terminal-Bench 4.0, AutomationBench, CursorBench 3.2.0, and Humanity's Last Exam.
02
Anthropic used Fable 5.1 and Mythos 5.1 on molecular design, computational biology, and planetary mapping. In protein design tests, lab-confirmed binders reached nearly a 50% hit rate across 12 targets, compared with typical 10–15% hit rates in protein design today.
03
Claude Fable 5.1 trained a neural network on NASA Magellan radar and an existing map covering one-fifth of Venus. Anthropic says the new elevation map covers a third of the planet, resolves details down to two to three kilometers instead of 10 to 20, and improves height accuracy by up to 25%.
04
Base input and output prices stay at the Fable 5 rates, but cache reads drop 75% to $0.25 per million tokens officially. Anthropic estimates typical workloads cost about 25% less than Fable 5, and highly agentic workloads up to about 45% less. PoYo keeps the same 20% discount on those official rates.
Use Fable 5.1 for repository-scale features, code review, performance work, and multi-day autonomous coding sessions that need root-cause debugging.
Run jobs that take hours, span tools and apps, recover from failed steps, and keep sending progress updates without constant supervision.
Hand off deep research, analysis, documents, spreadsheets, and slides, then review completed deliverables instead of supervising every step.
Analyze technical papers, experimental notes, diagrams, and long research trails where earlier models lose the thread.
Read charts, nested tables, PDFs, and design screenshots, then check coding output against the original visual goal.
Keep Fable-class quality on cached, tool-heavy loops. Official cache reads are 75% cheaper than Fable 5, which is where long agent runs spend most of the bill.
Anthropic's docs compare Claude Fable 5.1 with Claude Opus 5 first, then with Claude Fable 5. Start with Opus 5 for most workloads and move to Fable 5.1 for demanding reasoning and long-horizon agentic work, or when Opus 5 at higher effort still falls short. Uncached, Fable 5.1 is twice Opus 5 ($10 / $50 vs $5 / $25). Cache reads invert: $0.25 vs $0.50. Versus Fable 5, input and output stay the same while cache reads drop 75%.
| Feature | Claude Fable 5.1 | Claude Opus 5 | Claude Fable 5 |
|---|---|---|---|
| Model ID | claude-fable-5-1 | claude-opus-5 | claude-fable-5 |
| Released | September 1, 2026 | July 24, 2026 | June 9, 2026 |
| Positioning | Demanding reasoning and long-horizon agentic work | Recommended default for most Claude workloads | Previous Fable-class general-use model |
| Context window | 1M tokens | 1M tokens | 1M tokens |
| Max output | 128K tokens | 128K tokens | 128K tokens |
| Official input / output | $10 / $50 per 1M tokens | $5 / $25 per 1M tokens | $10 / $50 per 1M tokens |
| Cache read | $0.25 per 1M tokens (0.025x input) | $0.50 per 1M tokens (0.1x input) | $1.00 per 1M tokens (0.1x input) |
| Cache write (5m / 1h) | $12.50 / $20 per 1M tokens | $6.25 / $10 per 1M tokens | $12.50 / $20 per 1M tokens |
| PoYo price | 1600 / 8000 / 40 credits per 1M tokens ($8 / $40 / $0.20) | 400 / 2000 credits per 1M tokens ($2 / $10) | Not listed on PoYo |
| Latency | Slower | Moderate | Slower |
| Thinking | Adaptive, always on; cannot be disabled | Adaptive; can be disabled at high or below | Adaptive, always on |
| Default effort | high | high | high |
| Knowledge cutoff | June 2026 | May 2026 | January 2026 |
| Fast mode | No | Yes (research preview, $10 / $50) | No |
| Terminal-Bench-Science 0.1 | 52.6% | 29.0% | 24.7% |
| Terminal-Bench 4.0 | 55.8% | 52.3% | 42.0% |
| AutomationBench | 31.4% | 26.9% | 17.1% |
| Best fit | Long-horizon coding, research, and prefix-heavy agents | Most production coding, chat, and enterprise work | Earlier Fable-class workloads before the 5.1 upgrade |
Sources: Anthropic Claude Fable 5.1 overview and pricing (September 1, 2026). Benchmark scores are Anthropic's published figures with production safeguards enabled. PoYo prices are 20% below official Fable 5.1 rates; Opus 5 PoYo rates follow the live catalog.
Create a PoYo API key
Get API Key
Set the model ID
Use claude-fable-5-1 as the model ID in your request.
Choose an endpoint
Use the OpenAI-compatible endpoint /v1/chat/completions, or the Anthropic-native endpoint /v1/messages.
Open API docs
View API docs
claude-fable-5-1.claude-fable-5-1. There is no date suffix. The same ID works on PoYo through OpenAI-compatible /v1/chat/completions and Anthropic-native /v1/messages.tool_choice returns 400, older models cannot read Fable 5.1 thinking blocks, and editing earlier turns can invalidate later thinking blocks.low, medium, high (default), xhigh, and max. Anthropic says gains over Fable 5 are largest at higher effort, while medium often matches Fable 5 at lower cost./v1/chat/completions for OpenAI-compatible chat requests or /v1/messages for Anthropic-native Messages API requests. Set the model to claude-fable-5-1 in either case.Call Fable 5.1 with the same PoYo key you already use for other chat, image, and video models.
Use OpenAI-compatible Chat Completions or Anthropic-native Messages with one model ID.
Input, output, and cache-read rates stay aligned to Anthropic's published Fable 5.1 prices, then discounted 20%.
Test prompts on the model page, watch service status, and review pricing before moving the same model ID into production.