DeepSeek V4.1 Flash uses 4.3x tokens and 2.4x time vs GPT-6 Sol, gap within noise
teortaxesTex · x · 2026-10-20
teortaxesTex analyzes DeepSeek V4.1 Flash vs GPT-6 Sol on a multi-objective benchmark:
- Cost: DS takes 4.3x more tokens and 2.4x more time.
- Best single run: Sol's best reached Objective Group 6 (19 objectives); no V4.1F run did — DS's best stopped at 18.
- Statistics: mean 16.3 vs 15.7, median 16 vs 14 — the winner depends on the summary statistic, gap within noise.
- DS's lead rests on one late high-variance jump (12→18 in a single step, likely one credential/domain compromise completing several objectives).
He also notes Sol 6.1 is 2x slower than Sol 6 and >4x slower than Whale, asking whether looping V4.1 two or three times could close the gap.
More from Models
- Claude users report subscription limits running out much faster than before — prasenx · 2026-10-20
- User reports ChatGPT todos silently vanish between messages, threatens switch to Grok — koltregaskes · 2026-10-20
- doubleAI, co-founded by Mobileye CEO, claims its model cuts compute costs by 67% — pstAsiatech · 2026-10-20
- Longtime Anthropic critic returns: Claude 5.5 lineup is the best in months — intellectronica · 2026-10-20
- Vals AI: DeepSeek V4.1 Flash 1017 beats GPT-6 Sol on offensive cybersecurity benchmark — Nunki08 · 2026-10-20
- ModelScope releases MemT-1.0: models with built-in persistent memory, 1.86x faster at 128K context — kalyan_kpl · 2026-10-20