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. MiniMax M3 leads overall, 83.4 to 82.6. Their confidence intervals overlap, so treat the order as close rather than settled.
MiniMax M3 has the stronger current result: #46 at 1766.5 Elo and 83.44 overall, compared with Muse Spark 1.3 (thinking) at #49, 1699.2 Elo, and 82.63 overall. Their 95% Elo confidence intervals overlap, so the exact ordering should be treated as uncertain (1731.20–1804.20 and 1653.10–1736.10). MiniMax M3 has its clearest metric edges in Length Adherence (89.72 versus 63.88) and Visual Cue Quality (81.21 versus 75.93). Muse Spark 1.3 (thinking) counters on YouTube Best Practices (83.72 versus 80.43) and Slop Score (EQ-Bench + ours) (89.08 versus 86.04). At the measured run mix, MiniMax M3 costs $0.023 per article versus $0.041 for Muse Spark 1.3 (thinking); MiniMax M3 is the cheaper route. MiniMax M3 is the open-weights option; Muse Spark 1.3 (thinking) is closed. On the current automated evidence, MiniMax M3 is the stronger default; Muse Spark 1.3 (thinking) remains a defensible choice when its specific strengths, price, or deployment profile matter more than the headline rank.
Pick MiniMax M3 when you prioritize the stronger current board result, length adherence, visual cue quality, open weights, and lower measured cost.
Pick Muse Spark 1.3 (thinking) when you prioritize youtube best practices, and slop score (eq-bench + ours).
Blue bars: MiniMax M3. Orange bars: Muse Spark 1.3 (thinking). Same 0–100 scale; the bold bar wins that metric.
| MiniMax M3 | Muse Spark 1.3 (thinking) | |
|---|---|---|
| Overall / 100 | 83.4 | 82.6 |
| Writing Elo | 1766 | 1699 |
| Run-to-run spread (± overall std) | 2.030 | 2.770 |
| Cost per script (USD) | 0.023 | 0.041 |
| Avg latency (s) | 109.9 | 78.0 |
| Open weights | Yes | No |
Full scorecards: MiniMax M3 · Muse Spark 1.3 (thinking). How scoring works: methodology.
← All comparisons