AI Developer Experience at Google DeepMind 🔵 prev: Tech Lead at Hugging Face, AWS ML Hero 🤗 Sharing my own views and AI News
Control your reasoning efforts! The Gemini OpenAI compatibility layer now supports `reasoning_efforts`: "low", "medium", "high", and "none"! We map those to 1K, 8K, and 24K thinking token budgets. None == no thinking. Swap to Google DeepMind Gemini 2.5 Flash with 3 lines of code changed. Docs: https://lnkd.in/esa-pQU9
Great news Philipp Schmid! I will test it today. Aside of this I have a question. Is there any very good quality text to speech model via Gemini API which could be used in reading polish texts? I know that you pulled it in Gemini UI and NotebookLM, but can I use it in applications?
Thanks. Do you have context caching in Gemini OpenAI SDK?
custom K thinking tokens would also be nice
oh nice! Michele Sama
When is the reasoning available in the API? :)
Perfect, even More control.
Hi Philipp Schmid Can you please accept my connection request ? i have some queries on finetuning gemma3
Compatibility ftw👏👏
Data Science Leader | Empowering Business Innovation with Advanced AI Capabilities
1dHi Philipp Schmid Will this work with vertexAI Gemini models for OpenAI compatibility layer.