DataArticle

Qwen3.8-Max demonstrates autonomous long-horizon and multimodal capabilities including 10+ days of unattended coding, a 125-hour autonomous research loop that beat an original paper's benchmark by +2.71 points, silicon design from RTL to physical layout, and product-grade outputs across professional workflows.

Alibaba frames Qwen3.8-Max around headline autonomous capabilities: building a self-evolving coding harness over 10+ days, inventing a data-selection method beating a paper by +2.71 points across 125 hours, completing a full silicon design flow, and beating 87% of human teams in a data science challenge. ✦ AI generated

AINews · Latent Space · 2026-08-04 · original ↗

Share card ↓Export .md Export .txt

10+ Days Unattended Coding: Built a self-evolving coding harness from scratch over a multi-week autonomous run. Autonomous AI Research: Rebuilt a complete paper's pipeline (Unified Data Selection for LLM Reasoning) from scratch, then autonomously ran an iterative research loop over 125 hours to invent a new data selection method beating the original paper's benchmark by +2.71 points. Competitive Data Science: Competed against 526 human teams in the WWW2025 Multimodal Dialogue Intent Recognition Challenge, placing in the top 13% (outperforming 87% of human teams) within 24 hours. Executed a complete silicon design flow (GCD/RSA cryptographic accelerator) from RTL editing to simulation, synthesis, and physical layout. Reduced gate count from 8,298 to 678 gates while achieving an 81% die area reduction and meeting physical timing closure at 500 MHz. Demonstrated production-grade outputs across hundreds of professional workflows (e.g., corporate legal reviews, UI/UX design, structural engineering models, and automated ETF quant research). Outperformed competing models in the E-Commerce Bench (a 365-day store operation simulation), generating a 4.16x return (¥416,252 balance) through continuous game-theoretic negotiation and inventory planning. Integrates native visual feedback across planning, coding, and GUI interaction, enabling direct application recreation across platforms (desktop, mobile, web).

Read full article ↗excerpt · fair-use quotation

✳ Qwen

Around this claim

Evidence · 1

Qwen 3.8 27B has sufficient capability in long context, code generation, and tool-calling to successfully drive a coding agent loop on consumer hardware.Simon Willison · Simon Willison's Weblog · conf 65%

Context · 2

Qwen 3.8 27B is highly accurate at generating bounding box coordinates for objects in photographs, producing near-perfect results on its first attempt.Simon Willison · Simon Willison's Weblog · conf 60%Kimi K3 autonomously built a Triton-like GPU compiler (MiniTriton) that matches or beats existing tools on roofline benchmarks, and separately designed a chip for a nano model in a single 48-hour autonomous run.Kimi (Moonshot AI blog) · Import AI · conf 60%

Counterpoint · 1

Current frontier AI agents remain far from reliable at long-horizon, multi-program computer-use tasks, with the best-performing setup reaching only 20.6% binary accuracy on OSWORLD 2.0, though rapid improvement is likely given how quickly scores rose on OSWORLD 1.0.OSWorld 2.0 researchers · Import AI · conf 60%

This moment responds to

explains mechanism → Qwen3.8-Max is a 2.4T-parameter flagship model that would have been the top open model in the world but for the Kimi K3 release, and Alibaba has promised to open-weight both it and the 27B variant.AINews · Latent Space