OpenRouter Launches Gateway Security Center for API Key Auditing
OpenRouter introduces a Security Center that lets users audit all API keys in one place, identify risky keys, and manage an IP allowlist, after finding over 1,0…
Global AI Digest
-
-
Perplexity Computer Adds Interactive Charts and Visualizations With TradingView Integration
Perplexity Computer now generates interactive charts and visualizations directly in threads, using TradingView Lightweight Charts for financial data such as can…
-
OpenAI's 6.1 Sol Was Fastest-Growing Model, Slowdown Now Fixed
Sam Altman reveals that 6.1 Sol was OpenAI's fastest-growing model ever and had been slow under heavy load, but says performance should be much better now.
-
Altman Sees Untapped Potential in Sign In With ChatGPT and Plugin Extensions
After DevDay, Sam Altman reflects on builders creating startups in a day and argues there is far more untapped potential in Sign In With ChatGPT and Plugin Exte…
-
Open Guide Released for Training Agents with RL Across Multiple Harnesses
An open guide and method for training any model with RL inside real-world agent harnesses like Claude Code, Codex, and OpenCode, without modifying harness or tr…
-
OpenRouter Launches Security Center to Manage and Audit API Keys
OpenRouter announces a new Security Center that surfaces every API key across workspaces, flags risky keys, and lets teams disable, archive, or cap up to 500 ke…
-
GLM 5.3 Flash Credits Raised to $60 in Command Code GOAT Plan
Command Code raises GLM 5.3 Flash usage credits to $60 on its GOAT plan, offering 6x usage with a 1M context window and multimodal support.
-
OpenAI Launches Codex Cloud Environments for Reusable Agent Workspaces
OpenAI announced Codex cloud environments, reusable workspaces that keep agents running even after the laptop is closed, with repo, dependencies, scripts, and s…
-
Pareto 26.10 Preview Launches on OpenRouter at 68% Lower Price Than Current Version
Theunbiasedco's Pareto 26.10 Preview is now available on OpenRouter at $0.80/M input and $3.20/M output, significantly cheaper than the current Pareto, offering…
-
Custom Jev-Style Agent Trained via Single HuggingChat Prompt Runs on A100 in 16 Minutes for $6.60
A researcher reports training their own Jev-style agent API with a single prompt in Hugging Chat, completed by an ML intern on an NVIDIA A100 in 16 minutes, cos…
