Second Workshop on Video Large Language Models

Challenge Tracks

Challenge tracks for CVPR 2026 are being hosted on Eval.ai!

  1. Reasoned-Aware Composed Video Retrieval (CoVR-R) - CoVR-R challenge page (eval.ai)

    2. Reasoning-based video retrieval requiring models to interpret causal, temporal, and semantic modifications across video pairs.

  2. See Beyond Frames: The Implicit Video Relational Reasoning (VRR) Challenge - VRR Challenge page (eval.ai)

    3. Testing models' ability to infer spatial, narrative, and causal relationships that extend beyond what is directly visible in video frames.

  3. TimeLogic QA: Evaluating Temporal Reasoning in Videos - TimeLogic challenge page (eval.ai)

    4. Assessing temporal logical understanding including operators like Before, After, Until, Since, and event ordering across video sequences.

Each track tests a different aspect of VidLLMs. Challenge links are available on Eval.ai.

Challenge Prizes

Prizes will be awarded across the challenge tracks. Details TBD. Winners will be decided by leaderboard ranking and review.

Key Dates