PodBench - Thinking Mode - Script Quality: leaderboard
PodBench's 100-point audio-first script-quality rubric (content depth 45, structure 30, conversational language 25; Claude-4.5-Opus judge; higher is better) for open models run with reasoning enabled, with the reasoning trace removed before judging. Tencent, ACL 2026.
Metric: Podcast Script Quality (0-100 rubric: depth 45, structure 30, language 25). Source: arxiv.org. Saturation forecast: Estimated already saturated. 12 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | DeepSeek R1 0528 | 61.15 | #425 |
| 2 | Qwen 3 235B A22B (Thinking) | 60.42 | #622 (Qwen 3 235B A22B) |
| 3 | Qwen 3 32B (Thinking) | 59.92 | #960 (Qwen 3 32B) |
| 4 | Qwen 3 14B (Reasoning) | 56.86 | #1235 (Qwen 3 14B) |
| 5 | Qwen 3 30B A3B (Thinking) | 54.78 | #1124 (Qwen 3 30B A3B) |
| 6 | Qwen 3 8B (Thinking) | 53.89 | #1838 (Qwen 3 8B) |
| 7 | DeepSeek R1 0528 Qwen3 8B | 51.3 | #2101 |
| 8 | Qwen 3 4B (Reasoning) | 51.15 | #2297 (Qwen 3 4B) |
| 9 | DeepSeek R1 Distill Qwen 32B | 48.05 | #1780 |
| 10 | DeepSeek R1 Distill Llama 8B | 45.22 | #6773 |
| 11 | Qwen 3 1.7B (Thinking) | 42.29 | #5551 (Qwen 3 1.7B) |
| 12 | DeepSeek-R1-Distill-Qwen-7B | 36.21 | #8045 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=podbench-thinking-mode-script-quality · How It Works · Data refreshed daily, snapshot 2026-11-12.