Verified75% confidenceReleaseExact time
OpenAI previewed an Ultra Fast speed service mode in the API, enabling GPT-5.6 to achieve up to a 14x speed improvement with an output throughput of 750 tokens per second.
5
Sources
75%
Confidence
Medium-term (~90 days)
Relevance
8/15/2026
First Seen
Valid until: 11/13/2026
Sources
AI大厂降价混战：Gemini、Grok、DeepSeek同日发力，开发者红利期来临
bilibiliAI天天酱8/14/2026
Related Claims
UnverifiedThe capabilities of GPT-5.6 Ultra Fast officially entered preview on August 13 and were first made available through the OpenAI API.78% similarUnverifiedOpenAI GPT-4o's API response speed is typically around 80–100 T/s.77% similarUnverifiedOpenAI has allegedly reset GPT-6 Astra's rate limits for paid users.75% similarUnverifiedGPT-5.6's Ultra-Fast mode can output a peak of up to 750 tokens per second.74% similarUnverifiedOpenAI launched the Decisions API, which uses GPT-6 Luna for single-step judgment and returns answers in 150 milliseconds—approximately 10 times faster than standard calls—and is currently in limited preview.74% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/754434API
curl https://kongchang.com/api/v1/knowledge/claims/754434MCP
get_claim(id=754434)