⚡️Did you know that you can use Gemini 1.5 Flash 8B + function calling as a model router, to pick the best open-source or proprietary model based on your data + use case? This example uses function calling to select the most appropriate @HuggingFace model - then downloads the model, uses the open-source model with the image, and returns the result, all with a single Gemini inference call. Since Gemini 1.5's context window is ~1M tokens, you could register *thousands* of different OSS or proprietary models as potential tools. "Is this question challenging, or does it need powerful performance on a given eval?" -> automatically route it to the best model option "Is this question easier, or could it be solved with a specialized OSS model?" -> automatically have that function selected, and that model called Mind = blown. ‼️

🤯⚡️ Gemini 1.5 Flash-8B + code execution is insane. I just uploaded the text for a short story - Harlan Ellison's "Repent, Harlequin!" - and it was able to generate Python not only to tell me how many times Harlequin occurred in the text (✅ correct) but also to generate a dictionary for how many times each word occurred.