GPT-6 Sol (max) vs DeepSeek V4.1 Flash (max)
Head-to-head on the Towards AI writing benchmark: same 10 real YouTube scripts, five runs each, scored blind by a three-family judge panel. DeepSeek V4.1 Flash (max) leads overall, 88.1 to 88.0. Their per-model Elo intervals overlap on this board.
The verdict
This one is closer than the ranks suggest. DeepSeek V4.1 Flash (max) sits at #28 with 2183.1 Elo and 88.11 overall; GPT-6 Sol (max) is at #31 with 2165.6 Elo and 87.95 overall. Their 95% Elo intervals overlap (2147.5 to 2214.6 against 2127.0 to 2200.7), so do not read much into the exact order. Where DeepSeek V4.1 Flash (max) pulls ahead: Tone & Voice Match (87.97 vs 85.84) and Hook Strength (88.31 vs 86.91). GPT-6 Sol (max) still wins on Visual Cue Quality (89.15 vs 84.51) and Slop Score (EQ-Bench + ours) (94.08 vs 91.27), so it is not a clean sweep. The trade is price: DeepSeek V4.1 Flash (max) costs about $0.013 per article against $0.112 for GPT-6 Sol (max), so you pay less for the higher rank. DeepSeek V4.1 Flash (max) publishes open weights; GPT-6 Sol (max) does not. Default to DeepSeek V4.1 Flash (max). Reach for GPT-6 Sol (max) when visual cues and slop score matter more to you than the headline rank.
Pick DeepSeek V4.1 Flash (max) for the stronger board result, tone, and the hook.
Pick GPT-6 Sol (max) for visual cues and slop score.
Reading the Elo intervals. Each range estimates uncertainty around one model's Elo. We repeatedly sample from every model's recorded run scores on each task and recalculate the full ranking. Comparing these separate ranges does not test the Elo gap between two models. The tasks stay fixed, so the ranges do not measure how the ranking would change on new tasks.
Metric by metric
Blue bars: GPT-6 Sol (max). Orange bars: DeepSeek V4.1 Flash (max). Same 0–100 scale; the bold bar wins that metric.
Everything else that differs
| GPT-6 Sol (max) | DeepSeek V4.1 Flash (max) | |
|---|---|---|
| Overall / 100 | 88.0 | 88.1 |
| Writing Elo | 2166 | 2183 |
| Run-to-run spread (± overall std) | 1.330 | 1.120 |
| Cost per script (USD) | 0.112 | 0.013 |
| Avg latency (s) | 171.0 | 76.2 |
| Open weights | No | Yes |
Full scorecards: GPT-6 Sol (max) · DeepSeek V4.1 Flash (max). How scoring works: methodology.
← All comparisons