Anthropic Deploys Claude Opus 4.8: Setting New Records on NL2Repo (69.7%) and Multi-Hour SWE-Marathon

By Marcus Sterling • Published on 2026-09-09 • 5 min read •Source: Anthropic Blog

Anthropic introduces Claude Opus 4.8, specialized in sustained repository-level code synthesis and complex architectural refactoring across multi-hour autonomous SWE-Marathon sessions.

Anthropic has begun the general rollout of Claude Opus 4.8, tailored for enterprise engineering teams executing high-complexity software refactors.

SWE-Marathon & NL2Repo Leadership

While smaller models excel at one-line fixes, Opus 4.8 dominates long-horizon repository creation:

NL2Repo: 69.7% (#1 in turning natural language design specs into complete, compilable repositories).
SWE-Marathon (v1.1): 48.8%, sustaining focus across 8-hour agentic sessions without degradation.

Opus 4.8 pairs natively with the Claude Code CLI and supports extensive steerability through extended system prompt schemas.

Source & Fact Check

This technical dispatch was verified against primary documentation released by Anthropic Blog.

Read Original Announcement on Anthropic Blog →

Go deeper: related guide

This story makes more sense when you build it. Here is the practical step-by-step guide:

Claude Code Complete Guide: build your first autonomous coding agent →

Get 1 AI agent blueprint every week

Free newsletter: one practical AI agent or automation you can build and sell, every week. No spam.