Hi everyone 
We’re excited to share another major update from the Genum team — this one is big.
| Update | Description |
|---|---|
| Placeholders | Genum now supports dynamic prompt blocks  Add a {{key}} to your prompt, define the variants it can switch between, and pick one per run in the playground, a test case or the API. Placeholders are versioned with every commit and replace the old Memory feature. Learn more |
| Agent Testing & OpenTelemetry | Send traces from your own agents over OpenTelemetry and see every run as a session in Logs. Turn a session into a test case that checks which tools were called, with which arguments, and what the model answered — replayed with the recorded tool results, across multiple turns. Learn more |
| Prompt API | Manage prompts end to end through the API: create a prompt with its model and settings and run it right away, then fetch the exact committed instruction with the new render endpoint. Learn more |
| DeepSeek & New Models | DeepSeek is now a built-in provider with DeepSeek V4 Flash and DeepSeek V4 Pro. We’ve also added the GPT-6 and GPT-5.6 families, Claude Opus 5.5, Sonnet 5.5, Haiku 5.5 and Fable 5.1, and Gemini 3.5–3.8 Flash — 30 new models in total, many with selectable reasoning effort. Shut-down models are now blocked with a suggested replacement, and deprecated ones show their shutdown date. |
| Cost & Usage Insights | Run costs now account for cached input and reasoning tokens, and the playground and logs show the split for every run. Track usage per API key with the new API Key Activity card on your dashboard. |
| Continuous Optimization | We’re continuously improving Genum under the hood — the first page load is about four times smaller, security is tighter, and many fixes make the platform more stable for production use. |
What’s Next
We’re not slowing down.
More improvements for agent workflows, integrations, and enterprise-grade capabilities are on the way.
Thanks for building with us 
Let’s keep pushing practical, production-ready AI forward.
The Genum Team