A cluster of OpenAI API changes landed 9/8-9/10: GPT-Live 1 voice sessions (v1/live/sessions) went GA — full-duplex conversation that continues while a background model or agent reasons and calls tools, with Responses delegation or delegation to your own backend, priced at $0.05/minute billed by the second (backend model/tool usage billed separately). Prompt Cache Diagnostics went GA on Responses (GPT-5.6+): compares cache reuse between responses, identifies why misses happen, and produces troubleshooting guidance. Project API keys gained expiry dates with org- or project-level maximum-lifetime enforcement for admins. Two image models shipped (9/8): GPT Image 2.5 Sunburst for editing precision and Flare for fast everyday generation, both with xhigh/max quality settings on GPT Image 2 token rates.
GPT-Live 1's speak-while-thinking primitive is the threshold capability for real-time voice agents — the conversation doesn't pause for tool calls, and backend delegation means the voice loop can wrap your existing agent. Prompt-cache diagnostics addresses the single most common cost/debugging pain in long agent loops.
| Gpt Live 1 | GA 2026-09-10; v1/live/sessions; full-duplex; continues while backend model/agent reasons and calls tools; Responses delegation or own backend; $0.05/min billed per second; backend usage billed separately |
|---|---|
| Prompt Cache Diagnostics | GA 2026-09-08; Responses API, GPT-5.6+; compares current vs previous response cache reuse, identifies miss causes, troubleshooting guidance |
| API Keys | 2026-09-10: project API key expiry dates; admin-enforced maximum key lifetime at org or project level |
| Gpt Image 2 5 | 2026-09-08: Sunburst (editing precision) + Flare (fast everyday); xhigh/max quality settings; GPT Image 2 token rates |