Two newsletters were sent consecutively, lol

OpenAI latest update:

1. Generate structured outputs that conform to a JSON Schema

   option).

2. Save costs by using the latest GPT-4o snapshot

   The input token cost of is 50% cheaper than the original GPT-4o, and the output token cost is 33% cheaper. This version also supports structured output functionality.

3. Access GPT-4o mini via the API—and get free fine-tuning until September 23rd.

   It is currently OpenAI's most advanced small model and also the cheapest (60% cheaper than GPT-3.5 Turbo, but smarter and just as fast). Fine-tuning for GPT-4o mini has just been made available to all API users at access levels 1-5 who are paid customers. Fine-tuning is free until September 23rd (up to 2 million training tokens per day).

   

Google's latest update:

1. Gemma 2 2B

   for enhanced AI support on edge devices.

   

2. ShieldGemma

   .

3. Gemma Scope

   Gemma Scope is an interpretability tool for research purposes. It provides transparency into the decision-making processes of the Gemma 2 models. By utilizing an emerging technology—Sparse Autoencoders—Gemma Scope acts like a powerful microscope, allowing users to zoom in on specific points within the model while it runs, making its internal operations more comprehensible.