OpenAI ships a prompt-caching dashboard: watch your cache hit rate
New token-spend visibility for API builders, plus Copilot's coding agent in Microsoft Teams and Vercel's always-on production tracing.
Copy markdown
See your cache hit rate — and stop overpaying
OpenAI's new prompt-caching dashboard charts cache hit rate over time, cache reads per write, and the split of cache-read, cache-write, and uncached tokens. Cached input bills far cheaper than fresh input, so the view shows exactly where restructuring a prompt could turn expensive uncached tokens into cheap cached ones.
GitHub Copilot's coding agent lands in Microsoft Teams
After arriving in Slack, GitHub's coding agent now does shared agentic work inside Microsoft Teams: assign it a task from a channel and it triages, writes the fix, and opens a PR your team can watch. One less context-switch out of chat.
Vercel turns on always-on production tracing
Vercel added sampled tracing that captures real production and preview traffic, so you can debug a live incident from actual request traces instead of trying to reproduce it locally. It pairs with new custom metrics you can emit and analyze straight from your functions.