{"type":"video","version":"1.0","html":"<iframe src=\"https://www.loom.com/embed/bcd48dcbd1ba46968a9ad05dcf82ab74\" frameborder=\"0\" width=\"2580\" height=\"1935\" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>","height":1935,"width":2580,"provider_name":"Loom","provider_url":"https://www.loom.com","thumbnail_height":1935,"thumbnail_width":2580,"thumbnail_url":"https://cdn.loom.com/sessions/thumbnails/bcd48dcbd1ba46968a9ad05dcf82ab74-b3b1c5b1356a9dc7.gif","duration":199.809,"title":"Set Model Budget Rules for CodingAgents","description":"This Loom demonstrates how a Workspace Admin creates and tests a CodingAgentConfig named LeaseBudget that controls which model users can run based on budget usage. The configuration points to a specific budget ID and sets a $1,000 threshold with default model options of Opus 4a or GPT 5.6 sol; after crossing 50% it allows Sonnet 5 or GPT 5.5, and over 0.9 it restricts users to Kemi only. The creator then simulates being well under the budget to show the default options, updates the budget to test behavior after surpassing 50%, and finally tests that beyond 90% only Kemi is selectable. The Loom also notes that after surpassing the threshold, the session opens the corresponding model interface (Codex, then Claude, then Kemi)."}