ScaleDown's Summarize model goes live on OpenRouter alongside Extract and Classify-1

OpenRouter has made ScaleDown's Summarize model available, alongside Extract and Classify-1, which the company describes as constrained-decoding models each built for one task.

Summarize turns long text, images and PDFs into short summaries, rather than the fluent rewrite of the original text described in the earlier listing. Extract returns JSON from plain-English field descriptions with confidence scores, and Classify-1 gives calibrated probabilities for user-defined labels.

OpenRouter's post gives launch pricing of $0.04 per million input tokens for the models, with output free, and says this is 20% off for launch. The earlier report gave no pricing, context length or benchmark results, and the later post does not include benchmarks.