OpenRouter Does Not Charge for Failed Generations
OpenRouter states that failed generations incur zero credits, including empty outputs, error finish reasons, and broken image outputs, and this is enabled by de…
Global AI Digest
-
-
Google Releases Gemini 4 Carbon, a Frontier Model Focused on Cybersecurity
Google announced Gemini 4 Carbon, its next frontier model, two weeks after Gemini 4 Argon. It ranks #1 on Artificial Analysis Intelligence Index v5.0 with a sco…
-
NVIDIA Releases Nemotron 3.5 Transcribe, a 30B Open-Weight Speech Model
NVIDIA announced Nemotron 3.5 Transcribe, an open-weight 30B Mixture-of-Experts model with about 4B active parameters that handles transcription in 40 languages…
-
OpenCode Now Runs on Cloudflare Enterprise, Absorbing Worker Costs for Side Projects
dax, an OpenCode developer, said the project now has Cloudflare enterprise access, which means it covers the cost of users' cheap Worker deployments on side pro…
-
LOUPZ Routes AI Chats and Agents Through a Single European Endpoint
LOUPZ uses Requesty to point its AI chats, apps and agents at one European endpoint, spanning 5 to 15 models, with US data transfer only when strictly required…
-
Perplexity Updates Voice Mode With Multi-Session Computer Control
Perplexity has updated voice mode to let users work across multiple Computer sessions, start new tasks, and create and manage Projects within a single voice ses…
-
Claude for Legal Adds Ready-Made Artifacts Gallery
Anthropic introduced a gallery of ready-made artifacts in Claude for Legal covering contract review, regulatory horizon scans, and cite checks, which users can…
-
OpenCode Developer Builds Custom Linux Notification System on New Native UI Framework
An OpenCode developer created a custom Linux notification solution built on the team's new native UI framework, highlighting high customizability, including con…
-
OpenCode Developer Pushes Back on Hidden Enterprise Demand for Open Source Models
An OpenCode developer argues that hidden enterprise demand for open source models is overstated, citing the small number of top inference providers and the scar…
-
Mistral Launches Responses API for Agents and Adds Two Fast Models
Mistral introduced the Responses API in public preview for building agents, returning answers, reasoning and tool calls as separate items, and added DeepSeek V4…
