OpenAI officially launches the GPT-6.1 Sol Ultrafast version in API, Codex, and ChatGPT Work. Building on the standard version, this release boosts response speed to 8 times faster, priced at US$12 per million input tokens (approx. HK$94) and US$60 per million output tokens (approx. HK$468). GPT-6.1 Sol Ultrafast was announced at DevDay on September 29, positioned as delivering capabilities near the flagship Astra model at one-fifth the price.
The standard API is priced at US$2 per million input tokens (approx. HK$16) and US$10 per million output tokens (approx. HK$78). The Ultrafast mode is priced at exactly six times the standard version.
OpenAI developer relations engineer Dominik Kundel noted that Ultrafast is positioned at near-Astra-level intelligence, with speed 8 times that of Sol, while costing only 1.2 times Astra. Compared to Astra's standard API pricing of US$10 per million input tokens (approx. HK$78) and US$50 per million output tokens (approx. HK$390), Ultrafast is only slightly more expensive. This marks OpenAI's second iteration on the Ultrafast concept.
Previously, OpenAI partnered with Cerebras to release a GPT-5.6 Sol Ultrafast preview, claiming output speeds of 750 tokens per second, 14 times the standard speed at the time.
GPT-6.1 Sol Ultrafast delivers more efficient real-time interaction capabilities
This iteration of GPT-6.1 Sol Ultrafast achieves an 8-fold speedup, but the baseline model's capabilities have also improved substantially. Typical use cases cited by OpenAI include online troubleshooting, real-time agent navigation applications, and various real-time interaction scenarios. The API imposes no subscription-tier restrictions; developers can call the gpt-6.1-sol model directly and enable Ultrafast mode on a pay-as-you-go basis.
In Codex and ChatGPT Work, Ultrafast is available only to Pro 500 subscribers, eligible pay-as-you-go Enterprise users, and Edu users. Pro 500 is a US$500-per-month (approx. HK$3,900) subscription plan OpenAI introduced this year, meaning ChatGPT Plus and Pro users cannot access the feature directly. Some developers report that Ultrafast shows a clear speed improvement in XHigh effort mode, but token consumption also accelerates; the monthly allowance tied to the US$500 (approx. HK$3,900) plan can be depleted by 1% within minutes under that mode.