We’ve made ChromaDB operations more reliable by automatically splitting large upsert and query requests into smaller batches at runtime. This prevents failures that used to happen when requests exceeded ChromaDB’s per-request limits.
You can continue calling upsert and query operations the same way as before. The system now handles batching behind the scenes, so large payloads process smoothly without extra work.
View the ChromaDB Vector Database docs for more.