Today for AI
HOT RADAR
toolsHEAT 11.8°

AI CURATED HOT STORY · 11/10/2026

Airbnb Cuts Recommendation Embedding Latency from Two Days to 30 Seconds

1 reports archived1 independent sourcesupdated 11/10/2026, 09:19:27
Synthesis & Latest Updates
1 Sources Cross-Validated

Airbnb added Push Mode and near-real-time model transformation to its open-source feature platform Chronon, cutting sequence-recommendation embedding lag from nearly two days to 10-30 seconds. The change lifted offline NDCG by 1.67% and increased non-cancelled bookings by about 0.33% in online A/B tests, with the capabilities contributed back to the public Chronon repository.

LATEST/Airbnb added Push Mode and near-real-time model transformation to its open-source feature platform Chronon, cutting sequence-recommendation embedding lag from nearly two days to 10-30 seconds. The change lifted offline NDCG by 1.67% and increased non-cancelled bookings by about 0.33% in online A/B tests, with the capabilities contributed back to the public Chronon repository.

Hourly heat curve

2 fully observed hours
Fewer than 3 fully observed hours; trend not plotted yet.

Coverage timeline

Total 1 reports · Latest first
  1. InfoQ 中文T2·78 pts

    Airbnb Cuts Recommendation Embedding Latency from Two Days to 30 Seconds

    • Push Mode proactively notifies downstream consumers when streaming features commit, eliminating polling or scheduled waits
    • Near-real-time model transformation reuses the same Transformer encoder, avoiding a separate streaming inference implementation
    • The online ranking path stays largely unchanged, gaining only from fresher features in shared low-latency storage