OpenAI released its Decisions API to all users. It competes with TypeSafe’s Jev model. The API uses a reconfigured ChatGPT Luna for speed and cost. Response times drop from seconds to 100–180 milliseconds. OpenAI added voice and visual support at launch. Early users like Box and Speak cut latency from 1.4 to 0.13 seconds. Public benchmarks rank OpenAI’s model eighth and fourteenth. TypeSafe’s Jev leads on OpenRouter queries. OpenAI sees multi-modality as a key advantage. It targets 60-millisecond speeds for future video use.

• 100 to 180 milliseconds response time
• 1.4 to 0.13 seconds latency cut at Speak
• $7.6 billion TypeSafe valuation