Add reference examples to the Gemini v6 and OpenAI v7 detection prompts

workflows/CHANGELOG.md

@@ -18,6 +18,7 @@ for contributor and maintainer responsibilities.
 
 ### Added
 
+- Google Gemini v6 (`roboflow_core/google_gemini@v6`) and OpenAI v7 (`roboflow_core/open_ai@v7`) gain an optional `reference_examples` input for `object-detection`: up to 10 labeled example images, each `{"image": ..., "boxes": [{"x", "y", "width", "height", "class"}]}` in pixels of that image. The examples go before the image to label in every request, so the model sees what each class looks like and how tightly it is boxed. Gemini gets the example images with their boxes and labels drawn on them; OpenAI gets up to 3 crops per label, grouped by label (the output labels, so `output_classes` when set). Example images load through the host's image codec, with the same URL policy as any workflow image. Other tasks reject the input. Without it the requests are unchanged.
 - Vision Events (`roboflow_core/roboflow_vision_events@v1`) and Vision Event Bundle (`roboflow_core/vision_event_bundle@v1`) blocks gain an optional `value_base_units` input: what the event is worth, in base units of the use case's currency (e.g. cents for USD). The currency always comes from the use case. The value is sent as `valueBaseUnits` to the Vision Events API and in the bundle `payload.json`, and as `value_base_units` to the local event store. A typed value or workflow input must be a whole number (`true`/`false` are rejected). A value from another step is rounded to whole base units; one that is not a finite number or falls outside ±10,000,000,000 is dropped with a warning and the event is still recorded. Left empty, the use case's default value applies. `operator_feedback` events never carry a value, so the blocks omit it for them.
 
 ## `0.2.14`