Command Code Releases Agr Open Decision Model Family
Command Code announced Agr, an open decision model family with Agr (31B) and Agr-flash (360M), scoring 58.15 on Decision Index 0.2.1, focused on tool calls and…
Global AI Digest
-
-
GPT-6.1 Sol Now Available on GMI Cloud
GMI Cloud announced that GPT-6.1 Sol is now available to all its users, positioning it as near-Astra intelligence at a fifth of the price.
-
OpenAI Begins Text Watermarking in the EU Under AI Act, Offers API Option Worldwide
OpenAI announced it will watermark eligible ChatGPT and Codex text in the EU to comply with the EU AI Act, with an opt-in watermarking option for API customers…
-
Muse Agent Ported to Run Natively on a Sony PSP
A user ported the Muse gadget client to C so their AI agent runs natively on a 15-year-old PlayStation Portable, using the built-in mic for voice input.
-
Mistral Vibe Adds Smart Approve, GLM 5.3 Model, and Subagent Management
Mistral's terminal coding agent Vibe shipped three new features: smart approve mode, the GLM 5.3 model, and a new subagent management view for starting and re-i…
-
GPT-6 Astra and GPT-6.1 Sol Get ~50% Faster Default Speed via Sign in With ChatGPT
OpenAI's Codex team announces a Day 1 improvement: default response speed is ~50% faster across GPT-6 Astra and GPT-6.1 Sol for subscription users on all produc…
-
llama.cpp Gains Decision Models with /v1/systemone Endpoint
The latest llama.cpp builds now include a /v1/systemone endpoint for Jev-style decision inference, supporting multiple open models locally and privately, demons…
-
Cursor SDK Adds Agent Steering, Background Subagent Results, MCP Hints, and Custom System Prompts
Cursor announced a batch of SDK updates including runtime steering of running agents, background subagent result reporting, MCP hints on custom tools, and the a…
-
Claude Opus 5.5 Cache Pricing: 20x Cost Gap Between Hits and Misses
On Claude Opus 5.5, a cache hit costs $0.20 per million input tokens versus $4 for a miss, a 20x difference that dominates bills for agents resending full conve…
-
Hugging Face Explores Multi-Harness RL for More Portable Agents
Hugging Face researchers tested multi-harness reinforcement learning, training an open-weight model across multiple agent interfaces like Claude Code, Codex, an…
