Jev 1.14 Day 0 Support in LiteLLM

Both decision routes: POST /v1/systemone and POST /v1/decisions, plus the /typesafe pass-through. Request jev-1.14.0 through the proxy and the same virtual keys, budgets, spend logs and fallbacks you already use for decision models apply.

Four edges, top, bottom, left and right, as fractions of the image height and width measured from the top left corner, plus a confidence. Because the edges are fractions, the same answer is correct at any image size, so it can drive a crop, a highlight or a placement check.

With the headers TypeSafe-Danger-Zone-Images: true and TypeSafe-Danger-Zone-Long-Context: true on the request. LiteLLM forwards any client header that starts with TypeSafe-Danger-Zone- to TypeSafe deployments and to no other provider, so nothing leaks to other decision providers behind the same model name.

A box question to a model that cannot answer it, jev-1.13.0 and jev-latest included, or an image sent without the images header, returns HTTP 400 before any provider is called, so the failure is predictable and free.

With drop_params on, LiteLLM drops the box question or the image instead of failing and lets the rest of the request through. On /v1/decisions a dropped question comes back as a refusal type, and on /v1/systemone it is simply left out of the answers.

On /v1/systemone you name images in a top-level assets map and reference them from state. On /v1/decisions you send them OpenAI's way as input_image parts with an image URL or a data URL; LiteLLM turns each PNG or JPEG into an asset, swaps the asset name into the text and downloads image URLs first.

No. Spend for jev-1.14.0 lands in the spend logs at TypeSafe's $0.042 per 1M input tokens with no output charge, and cost tracking works on the LiteLLM version you already run after you reload the model cost map.