OpenAI details Control Room for stopping internal research inference
OpenAI’s November 9 announcement says Control Room blocks further model output for a stopped research workload within 35 seconds of receiving a stop instruction. Workload processes may remain running.
OpenAI says the system is deployed in research infrastructure. Inference engines deny further model requests for stopped workloads. The 35 seconds starts when Control Room receives a stop instruction, not when a problem first occurs, and does not promise that all processes terminate. This is not a new ChatGPT web, mobile, desktop or public API feature.
This is internal infrastructure, with no public product or pricing announced. Hong Kong, Macau and Taiwan access and interface languages are inapplicable. November 9 is the announcement date shown in OpenAI’s news index; the initial deployment date is unknown. The timing claim has not been independently tested here.
https://openai.com/news/
This is internal infrastructure, with no public product or pricing announced. Hong Kong, Macau and Taiwan access and interface languages are inapplicable. November 9 is the announcement date shown in OpenAI’s news index; the initial deployment date is unknown. The timing claim has not been independently tested here.
https://openai.com/news/
Usage conditions
- Hong Kong
- Public access eligibility, interface language and subscription pricing do not apply to Hong Kong: this is internal research infrastructure, with no consumer product announced. Not tested.
- Macau
- Public access eligibility, interface language and subscription pricing do not apply to Macau: this is internal research infrastructure, with no consumer product announced. Not tested.
- Taiwan
- Public access eligibility, interface language and subscription pricing do not apply to Taiwan: this is internal research infrastructure, with no consumer product announced. Not tested.
Official source: Introducing Control Room: Human control over AI workloads