DeepSeek V4.1 Flash 1017 fixes 34/105 hidden bugs in real-repo test, weights on Hugging Face
PawelHuryn · x · 2026-10-17
Pawel Huryn tested the newly released DeepSeek V4.1 Flash 1017 on real work: 2 repos with 105 planted hidden bugs.
- Rankings: Muse Wonder Full 1.0 (42) > Opus 5.5 (41.7) > Gemini 4 Argon (40.7) > DeepSeek V4.1 Flash 1017 (34) > Grok 4.7 (29) > Qwen3.8-Max (25.7) > DeepSeek V4.1 Flash (21.7) > GLM-5.3 (19).
- The attached leaderboard shows Sonnet 5.5 MAX leading overall at 51.3 (287 min, $154), with GPT-6 Astra at 45; Flash 1017 MAX cost just $0.90 in 50.6 min, standing out on cost-efficiency.
- The model has 763B params, weights already on Hugging Face. Huryn calls it a great improvement, finds it odd it wasn't labeled V4.2, and expects ±1.7 points of variance across repeated runs, with more data coming in the thread.
Related event: Open-Source 763B Model Tested on 105 Hidden Bugs(3 posts)→
More from Models
- Qwen-Image-Edit shows East Asian facial prior causing identity drift in edits — Gen-Talks · 2026-10-17
- AI turns a nuclear engineer's PWR rod-bow challenge from X into a structured analysis — burhop · 2026-10-17
- Voice models still give themselves away: smooth but repetitive prosody, masked latency filler phrases — dioscuri · 2026-10-17
- Image model showdown: Gemini 4 Argon takes 13 min at $1.2, DeepSeek matches quality in 50 min at $0.4 — teortaxesTex · 2026-10-17
- Users hit Grok Bot's long-context limits: 'retire' each bot after one project — alexcovo_eth · 2026-10-17
- DeepSeek 4.1 Flash generates a realistic animated grassy hill with sound in one turn — zizhpan · 2026-10-17