Benchmark RadarAI BENCHMARK PROFILE
SagaQA
SagaQA is a long-form video benchmark for multi-hop reasoning over full-length TV series, requiring reasoning across episodes and multimodal narrative understanding.
- Released
- 2026-06-02
- Readiness
- Paper only
- Primary field
- General AI
Why it matters
No public artifacts or reuse path are provided, and the benchmark lacks a clear scoring contract for independent runs.
Motivation
We introduce SagaQA, a long-form video benchmark for multi-hop reasoning over full-length TV series.
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.