Ollama Adds D3 as OpenAI Expands Its Decisions API
Developers have another downloadable model for judging documents and images locally. Alongside OpenAI's now generally available Decisions API, it makes two choices clearer: where evidence is processed, and which work can be reused.
Edition File
All editions ›Recent Editions
Full history ›Phonon’s Offline Speech Is Now a Kit You Can Download
Gradium has opened a practical route from an app’s text to locally generated speech: download the runtime, model and a reusable voice. Preparing a new voice is a separate job.
Pydantic Updates Agent Instructions Without Rewriting the Opening
An agent’s task state can change without changing the start of its next model request. Pydantic AI’s new append mode puts the update at the end of the conversation, preserving the earlier prefix for possible cache reuse.
OpenAI Gives 20 Languages the Same Database Exam
MultiLangBench asks a model to build a database engine in a chosen language. The revealing part is the examiner: reference SQLite reads the file that implementation leaves behind.
Unsloth Can Reject a New Model to Keep Another Loaded
New Studio code gives chat, image and video models a choice: share GPU memory when possible, make room, or refuse a disruptive load. When media does displace chat, it can also save the primary GGUF model’s cached work.
How AI Agents Work
The long-form map behind the daily paper: context, tools, loops, memory, delegation, safety, interface, and trust.