Tootfinder

OpenAI makes Realtime API generally available with new features, including MCP support, and launches gpt-realtime, its most advanced speech-to-speech model (Sabrina Ortiz/ZDNET)
https://www.zdnet.com/article/openai-gives-its-voice…

OpenAI gives its voice agent superpowers to developers - look for more apps soon
The company's AI voice offerings just got several new capabilities, including MCP support.

Saw this "Followed hashtags" collapsible list on mastodon.social (v4.4.0 nightly), wondering how long the list of hashtags will be.
Turns out it's hard-coded to 4 🤷‍♂️
- Initial PR: https://github.com/mastodon/mastodon/p

TIL: #lhp
Der Vorteil im Vergleich zu

Client--Library Compatibility Testing with API Interaction Snapshots
Gustave Monce (LaBRI), Thomas Degueule (LaBRI), Jean-R\'emy Falleri (LaBRI), Romain Robbes (LaBRI)
https://arxiv.org/abs/2507.20814

Client--Library Compatibility Testing with API Interaction Snapshots
Modern software development heavily relies on third-party libraries to speed up development and enhance quality. As libraries evolve, they may break the tacit contract established with their clients by introducing behavioral breaking changes (BBCs) that alter run-time behavior and silently break client applications without being detected at compile time. Traditional regression tests on the client side often fail to detect such BBCs, either due to limited library coverage or weak assertions that…

Provable Benefits of In-Tool Learning for Large Language Models
Sam Houliston, Ambroise Odonnat, Charles Arnal, Vivien Cabannes
https://arxiv.org/abs/2508.20755 https://

Provable Benefits of In-Tool Learning for Large Language Models
Tool-augmented language models, equipped with retrieval, memory, or external APIs, are reshaping AI, yet their theoretical advantages remain underexplored. In this paper, we address this question by demonstrating the benefits of in-tool learning (external retrieval) over in-weight learning (memorization) for factual recall. We show that the number of facts a model can memorize solely in its weights is fundamentally limited by its parameter count. In contrast, we prove that tool-use enables unbo…

The new EU age verification app is currently aiming to use Google Play Integrity devices because NO-ONE LEARNS ANYTHING. EVER. 😖
Here’s the GitHub issue https://github.com/eu-digital-identity-wallet/av-app-android-wallet-ui/issues/10

Wo ist die Protest-Aktion?
"The European Union will ban apps not licensed by [Google's Play Integrity API] under its new age verification system"
https://www.talkandroid.com/512861-eu-age-verification-app-google/

EU Age Verification App Sparks Ban on Non Google Licensed Android Apps - Talk Android
The European Union will ban apps not licensed by Google under its new age verification system, which has been met with controversy. Set for pilot testing

heise | KI-Code mit Metal parallelisieren: So geht es mit Apples API
Large-Language-Modelle haben einen immensen Bedarf an Rechenleistung. Wir zeigen, wie man mit Apples Metal-API sequenziellen Code von der CPU auf die GPU lädt.

Den som får svenska myndigheter att bygga samma API'n får en kram och en glass av mig shitloads med respekt

Enhancing Project-Specific Code Completion by Inferring Internal API Information
Le Deng, Xiaoxue Ren, Chao Ni, Ming Liang, David Lo, Zhongxin Liu
https://arxiv.org/abs/2507.20888

Enhancing Project-Specific Code Completion by Inferring Internal API Information
Project-specific code completion is a critical task that leverages context from a project to generate accurate code. State-of-the-art methods use retrieval-augmented generation (RAG) with large language models (LLMs) and project information for code completion. However, they often struggle to incorporate internal API information, which is crucial for accuracy, especially when APIs are not explicitly imported in the file. To address this, we propose a method to infer internal API information w…