TASC-m - Clarification: leaderboard
Time-sensitive safety awareness in authored conversations. TASC v1 uses one sample per authored conversation through OpenRouter, with provider-default temperature, top-p and reasoning effort. These are published observations; this intake makes no model calls. Condition A uses a duration parser. Conditions B and C use GPT-5.6 Luna as the judge of record, Claude Sonnet 5 as a second judge and Claude Opus 5 for adjudication. TASC-m Table 3 uses 3,397 shared inputs, with column denominators 234, 877, 1,347, 275 and 79. TASC-p has 33 complete scenario tuples per model, except DeepSeek-V4-Pro with 32. Non-violation catches are false alarms, so lower is better. Recovery is conditional on Condition B misses; the undefined Opus recovery remains missing. Exact API snapshots are unreported. Question access: Primary result reviewed; exact scored release availability not confirmed.
Metric: Published rate (%). Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 13 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 5 (Benchmark Sweep Run: source=TASC; provider defaults; one sample; full authored history) | 16.7 | |
| 2 | Kimi K3 (Benchmark Sweep Run: source=TASC; provider defaults; one sample; full authored history) | 6.5 | |
| 3 | GPT-5 mini (Benchmark Sweep Run: source=TASC; provider defaults; one sample; full authored history) | 5.8 | |
| 4 | Qwen3.8-Max (Benchmark Sweep Run: source=TASC; provider defaults; one sample; full authored history) | 4 | |
| 5 | GPT-5.6 Terra (Benchmark Sweep Run: source=TASC; provider defaults; one sample; full authored history) | 3.6 | |
| 6 | GPT-5.5 (Benchmark Sweep Run: source=TASC; provider defaults; one sample; full authored history) | 3.3 | |
| 7 | Claude Sonnet 5 (Benchmark Sweep Run: source=TASC; provider defaults; one sample; full authored history) | 2.5 | |
| 8 | Gemini 3.7 Flash (Benchmark Sweep Run: source=TASC; provider defaults; one sample; full authored history) | 1.5 | |
| 9 | GLM-5.2 (Benchmark Sweep Run: source=TASC; provider defaults; one sample; full authored history) | 1.5 | |
| 10 | DeepSeek-V4-Pro (Benchmark Sweep Run: source=TASC; provider defaults; one sample; full authored history) | 1.1 |
Interactive version: theaggregate.ai/benchmark?slug=tasc-m-clarification · How It Works · Data refreshed daily, snapshot 2026-11-06.