Microsoft: Microsoft-Decision-1 vs ScaleDown: Classify-1

Microsoft: Microsoft-Decision-1 is a decisions model from Microsoft, released on 2026-10-09. ScaleDown: Classify-1 is a decisions model from ScaleDown, released on 2026-10-13. Both belong to the same category, so they are aimed at the same broad job: taking an input and returning a decision rather than a paragraph of prose. Neither provider is a newcomer to serving models through APIs, so the choice between them comes down to fit for your specific inputs and context size rather than brand.

The clearest difference is the context window. Microsoft-Decision-1 accepts 32768 tokens, while ScaleDown: Classify-1 accepts 1000000 tokens, so ScaleDown gives you the bigger context by a wide margin. The input modalities also differ. Microsoft-Decision-1 takes text only. ScaleDown: Classify-1 takes text, images and files, so if your decision depends on a scanned form, a screenshot or an uploaded document, ScaleDown can consume it directly while Microsoft would need you to convert everything to text first. Both return decisions as their output modality, so there is no difference on the output side, and neither model lists supported parameters, which means you should not expect extra knobs like temperature or top_p to be honored in the request body.

For straightforward, text-only decision tasks where the prompt fits comfortably in 32768 tokens, Microsoft-Decision-1 is a reasonable default. It is a Microsoft model, which may matter if your stack already leans that way, and the smaller context is not a limitation for short classification style inputs. Reach for ScaleDown: Classify-1 when the decision has to account for a long document, a large batch of text, or non-text inputs such as an image or an attached file in one call, since it is the only one of the two that can see those modalities and it accepts far more tokens. If your inputs are text and small, the two are about the same price on the input side, so cost is not the deciding factor there.

You call either model through api14's OpenAI-compatible API. Point your existing client at the api14 base URL with your api14 key, then set the model field to the id you want. For Microsoft the id is microsoft/microsoft-decision-1, and for ScaleDown it is scaledown/classify-1. Everything else in the request stays the same, so the switch is a one line change in the model id, which makes it easy to A/B the two against your own inputs. Remember that neither model lists supported parameters, so keep request bodies minimal and avoid passing sampling settings that may be rejected or ignored.

There is no fact here that makes one model strictly better in every case. ScaleDown: Classify-1 wins on context length and input modalities and is not more expensive on input, but Microsoft-Decision-1 remains a fine pick for short, text-only decisions, particularly if you prefer a Microsoft-served model. Start with the model whose limits match your inputs, and if you are unsure, run the same request against both ids and compare the decisions on your own data.