GPT-5.6 Sol (ultra) vs Qwen3.8 Flash
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. GPT-5.6 Sol (ultra) leads overall, 88.0 to 83.0.
The verdict
GPT-5.6 Sol (ultra) wins this one, and the board is not close: #25 at 2186.8 Elo and 88.05 overall, against #65, 1698.8 Elo and 82.97 overall for Qwen3.8 Flash. The 95% Elo intervals do not overlap (2146.7 to 2223.2 against 1650.3 to 1754.4). Where GPT-5.6 Sol (ultra) pulls ahead: Visual Cue Quality (90.11 vs 79.55) and Continuity & Emotion (86.03 vs 78.74). Qwen3.8 Flash still wins on Length Adherence (92.46 vs 91.90), so it is not a clean sweep. The trade is price: Qwen3.8 Flash costs about $0.013 per article against $0.363 for GPT-5.6 Sol (ultra), so you pay for the higher rank. Both are closed. Default to GPT-5.6 Sol (ultra). Reach for Qwen3.8 Flash when length adherence and the lower price matter more to you than the headline rank.
Pick GPT-5.6 Sol (ultra) for the stronger board result, visual cues, and continuity.
Pick Qwen3.8 Flash for length adherence and the lower price.
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-5.6 Sol (ultra). Orange bars: Qwen3.8 Flash. Same 0–100 scale; the bold bar wins that metric.
Everything else that differs
| GPT-5.6 Sol (ultra) | Qwen3.8 Flash | |
|---|---|---|
| Overall / 100 | 88.0 | 83.0 |
| Writing Elo | 2187 | 1699 |
| Run-to-run spread (± overall std) | 1.230 | 2.420 |
| Cost per script (USD) | 0.363 | 0.013 |
| Avg latency (s) | 195.5 | 341.7 |
| Open weights | No | No |
Full scorecards: GPT-5.6 Sol (ultra) · Qwen3.8 Flash. How scoring works: methodology.
← All comparisons