---
format: "aidr-story-markdown/v1"
id: "4d5861773f6bf96e057ec87d8586637f2db48d2674f7029decd7c550bcd83fef"
canonical_url: "https://aidr.today/4d586177?lang=en"
title: "Claude Fable 5.1 Hits 1765 Points to Top Code Arena WebDev With 77 Point Lead"
lang: "en"
requested_lang: "en"
available_langs: ["en","vi"]
translation_fallback: null
fallback_fields: []
published_at: "2026-09-02T20:54:48.000Z"
category: "Models"
topics: ["anthropic","claude","coding","benchmark","agent"]
source_urls: ["https://x.com/arena/status/2095194515984585171","https://x.com/ShayneRedford/status/2095194296639185197","https://x.com/jerryjliu0/status/2095185262003265790","https://x.com/skalskip92/status/2095235295323074646","https://x.com/sarahzorah/status/2095128952402563542","https://x.com/btibor91/status/2095084209329385631","https://x.com/WesRoth/status/2095127097761304720"]
summary: "GitHub Copilot now supports Anthropic's Claude Fable 5.1 model for long running coding tasks and agentic workflows. The model ranks first in the Code Arena: WebDev benchmark with 1,765 points, a 77 point margin over Qwen3.8 Max and a jump from Fable 5's previous rank of 8. Blended pricing for the model is $40 per million tokens, which reduces the cost per task by 25% for typical workloads. Vision capabilities in Fable 5.1 improve object detection and table parsing on ParseBench compared to the previous generation. The model reached 52.6% on the Terminal Bench Science 0.1 benchmark, more than double the result of Fable 5, while document OCR costs are 15 cents per page. New Enterprise Frontier Safeguards allow eligible customers to keep activity data on their own cloud infrastructure for privacy instead of using Anthropic's servers."
---

# Claude Fable 5\.1 Hits 1765 Points to Top Code Arena WebDev With 77 Point Lead

> [Open the canonical story](<https://aidr.today/4d586177?lang=en>)

**Published:** 2026-09-02T20:54:48.000Z
**Category:** Models
**Topics:** anthropic, claude, coding, benchmark, agent

## Summary

GitHub Copilot now supports Anthropic's Claude Fable 5\.1 model for long running coding tasks and agentic workflows\. The model ranks first in the Code Arena: WebDev benchmark with 1,765 points, a 77 point margin over Qwen3\.8 Max and a jump from Fable 5's previous rank of 8\. Blended pricing for the model is $40 per million tokens, which reduces the cost per task by 25% for typical workloads\. Vision capabilities in Fable 5\.1 improve object detection and table parsing on ParseBench compared to the previous generation\. The model reached 52\.6% on the Terminal Bench Science 0\.1 benchmark, more than double the result of Fable 5, while document OCR costs are 15 cents per page\. New Enterprise Frontier Safeguards allow eligible customers to keep activity data on their own cloud infrastructure for privacy instead of using Anthropic's servers\.

## Sources

- [Supporting source](<https://x.com/arena/status/2095194515984585171>)
- [Supporting source](<https://x.com/ShayneRedford/status/2095194296639185197>)
- [Supporting source](<https://x.com/jerryjliu0/status/2095185262003265790>)
- [Supporting source](<https://x.com/skalskip92/status/2095235295323074646>)
- [Story source](<https://x.com/sarahzorah/status/2095128952402563542>)
- [Supporting source](<https://x.com/btibor91/status/2095084209329385631>)
- [Supporting source](<https://x.com/WesRoth/status/2095127097761304720>)