Qwen4-Max hands-on: 2.4T MoE with 1M context passes game, report and finance tests
Div_pradeep · x · 2026-11-18
A hands-on thread tests the newly launched Qwen4-Max (2.4T-parameter MoE, native vision, 1M-token context) on three tasks: building a playable Three.js space-runner game from one prompt, finding 12 planted contradictions in a fake 199-page annual report with page references, and building a full DCF with sensitivity table from 3 years of financials while flagging red flags like ballooning receivables and negative operating cash flow.
More from Models
- AI agents sometimes produce unreadable but correct math — burny_tech · 2026-11-18
- Code model eval: Rust hits 95-100% accuracy as 13 languages see big gains — shuding · 2026-11-18
- Asked Opus to make a video explaining consciousness and existence — "no wiser but cool" — flowersslop · 2026-11-18
- theo retests GPT-6.2 Astra: better, but still 'directionally very bad' at debugging — airesearch12 · 2026-11-18
- Claude correctly folds ubiquitin from a plain prompt, no specialized tools needed — iScienceLuvr · 2026-11-18
- GPT-6.2 Astra Tops Bug-Hunt Coding Benchmark at 62.7/105, One-Fifth Sonnet's Cost — PawelHuryn · 2026-11-18