Referrals, Cache Controls & Product Updates
Share a referral link and reward new signups, toggle provider cache writes per project, see cache and audio costs in your logs, plus a faster Chat and more.

A roundup of everything else that shipped recently — a new referral program, finer cost controls, and a stack of product polish.
Refer Friends, Reward Signups
Every organization now has a shareable referral link. Send it to a friend and they get a bonus credit on their first top-up.
1https://llmgateway.io/ref/<your-org-id>1https://llmgateway.io/ref/<your-org-id>- Grab your link from the dashboard and share it anywhere
- New users who sign up through it get a bonus percentage on their first top-up
- Track who you've referred right from your org settings
Provider Cache-Write Controls
You can now enable or disable provider-level cache writes as an organization or project setting. Keep cache writes on to save on repeated prompts, or turn them off when you'd rather not persist prompt content upstream — your call, per project.
Clearer Cost Breakdowns in Logs
Request logs now break out cache-write and audio tokens and cost separately, so you can see exactly where spend is going — not just a single blended number.
Chat & Playground
- Faster long conversations — Chat now virtualizes the message list, so threads with hundreds of messages scroll smoothly
- Comparison mode remembers your setup — your side-by-side model comparison persists between sessions
- Image & video history — generated images and videos now have a dedicated history you can browse and revisit
Enterprise
- Per-project routing overrides — apply different routing rules to different projects within the same org
- Per-project source usage — break down usage by source (
x-source) for each project - IP CIDR IAM rules — restrict access to specific IP ranges, available on the Enterprise plan