classifier.dev introduces infinite context for zero-shot classification of documents of any length.
classifier.dev has released an update adding "infinite context" support, allowing users to classify documents of any length over plain HTTP. For documents exceeding standard limits, the service automatically chunks the text, screens for relevant evidence using parallel calls, and synthesizes a final structured classification, handling documents up to 10 million tokens.
classifier.dev just killed the need to build custom RAG pipelines for basic document routing and classification. By abstracting away chunking and context window management into a single HTTP call, it makes processing massive texts as easy as sorting short strings.
- –Supports up to 10 million original context tokens per job
- –Automatically handles chunking and relevant evidence screening
- –Server-side processing allows asynchronous polling for massive documents
- –Costs are tied directly to token usage, providing transparent pricing
DISCOVERED
1h ago
2026-09-25
PUBLISHED
2h ago
2026-09-25
RELEVANCE
AUTHOR
michael_chomsky