Agents & dev tools · 9 November 2026

Alibaba's Co-Design Bench shows AI coding agents fully solve only 26.4% of tasks

Alibaba's Co-Design Bench benchmark tests AI agents on real user change requests. GPT-6 Astra, the top model, meets about 70% of task requirements on average but fully solves only 26.4% of tasks, whereas older benchmarks report >99% scores.

Benchmarks are the company's own; not independently verified.

Why it matters

Alibaba’s Co‑Design Bench shows GPT‑6 Astra fully solves only 26.4% of coding tasks; temper automation promises accordingly.

For product managers

For your product

What does this mean for your product?

Try it on your product

Read at The Decoder

More from this issueAll AI news from 9 Nov →

Apple pushes Siri upgrades for iOS 27.3 and Spring third‑party integration

Google launches Elephant‑Goldfish plugin for spec‑first coding with five zero‑context reviewsAvailable now

Brava Security raises $15M, launches multi-agent AI platform

Open‑weight DeepSeek V4.1 Flash achieves 72.8% cyber‑attack success at one‑tenth costResearch

Get this every morning, written for your product.

Deltaday reads the whole AI news cycle and tells you what changed, whether it's real yet, and what to do about it.

Start free See how it works