Burkov Gives Up on DeepSeek 4.1 Flash for Coding: Best Kept as a Frontier-Model Subagent
burkov · x · 2026-09-18
After wasting half a day, ML author Andriy Burkov gave up on DeepSeek 4.1 Flash for coding work: it needs hand-holding and verification of every change and its side effects. His verdict — it works as a subagent controlled by a frontier model, or as an executor of extremely detailed specs, not as an autonomous coding lead.
Related event: Burkov Tests DeepSeek 4.1 Flash: Falls Short for Coding(2 posts)→
More from Models
- LightOnOCR-3 Reads 16th-Century Manuscripts and Ancient Greek Without Training — IgorCarron · 2026-11-02
- 'You Can Feel the Gradient Slopes': Dev Contrasts Benchmaxxed RL Models vs General Ones — theteknosaur · 2026-11-02
- LightOnOCR-3 Launches: Tops FRBench, Adds Layout Structure and Chart Extraction — IgorCarron · 2026-11-02
- NVIDIA's Kumo Is a Family of Four Tabular Foundation Models, Not Just One — pandeyparul · 2026-11-02
- SemiAnalysis: safety classifiers are underrated — Meta's Muse tops cyber benchmark at 35.2% — ThomasScialom · 2026-11-02
- DeepSeek V4.1 Flash 1017 wins over a longtime closed-source-only user: fast, cheap, reliable — _bachrc · 2026-11-02