{"rewrite":{"id":"r_6d04149b730fd2b1da14dd4b","clusterId":"c_85bb1b913a1bef81dfee7c9b","slug":"openai-launches-ultrafast-mode-for-gpt-5-6-sol","model":"deepseek-v4-flash:free","headline":"OpenAI Launches Ultrafast Mode for GPT-5.6 Sol","summary":"OpenAI announced Ultrafast, a service that runs GPT-5.6 Sol up to 14 times faster, using Cerebras inference technology. The preview is open to some users, with a waitlist for broader access. OpenAI says Ultrafast reaches 750 tokens per second and completes Humanity's Last Exam in 11 hours 11 minutes.","whyItMatters":"OpenAI is pairing its frontier model with Cerebras hardware to push GPT-5.6 Sol to 750 tokens per second, a speed tier that changes what real-time use of the model can look like.","webCardHtml":"\u003cp\u003eOpenAI already offered Fast mode, which runs GPT-5.6 Sol at up to 2.5 times its standard speed. Ultrafast goes further, claiming up to 14 times the speed. The company says the mode sustains state-of-the-art performance while producing 750 tokens per second.\u003c/p\u003e\u003cp\u003eOn the 2,500-question benchmark Humanity\u0026#39;s Last Exam, Ultrafast finished in 11 hours 11 minutes. OpenAI also cites its GDP-Val test, where Ultrafast handled practical work at 5.6 times the speed of standard mode.\u003c/p\u003e\u003cp\u003eThe preview is available to a limited set of users. OpenAI plans to expand access as compute becomes available, and interested users can join a waitlist.\u003c/p\u003e","blueskyPost":"OpenAI's Ultrafast mode for GPT-5.6 Sol runs 14x faster on Cerebras hardware, but its 750 tokens per second and 11-hour Humanity's Last Exam time suggest speed is measured on narrow benchmarks, not general use.","twitterPost":"OpenAI's Ultrafast mode for GPT-5.6 Sol hits 750 tokens per second, but 11 hours on Humanity's Last Exam shows the speed is benchmark-specific.","threadsPost":"OpenAI's Ultrafast mode for GPT-5.6 Sol is 14x faster using Cerebras inference, hitting 750 tokens per second. Yet it still takes 11 hours on Humanity's Last Exam, so the speed gain is real but context-dependent, not a universal upgrade.","newsletterBlurb":"OpenAI introduced Ultrafast, a preview service that runs GPT-5.6 Sol up to 14 times faster using Cerebras inference hardware. The company reports 750 tokens per second and an 11-hour completion of Humanity's Last Exam. Access starts with a limited preview and a waitlist.","attributionJson":"[{\"source\":\"GIGAZINE\",\"url\":\"https://gigazine.net/news/20260814-gpt-5-6-sol-ultrafast/\",\"title\":\"GPT-5.6 Solを最大14倍高速化する「Ultrafast」モードが登場\"}]","lintFlagsJson":null,"lintHits":0,"costUsd":0,"inputTokens":4964,"outputTokens":1180,"status":"published","repairAttempts":0,"nextRepairAt":null,"factsAttemptedAt":1786787395,"createdAt":"2026-08-15T09:32:22.000Z","publishedAt":"2026-08-15T09:36:44.000Z","updatedAt":"2026-08-15T09:32:22.000Z"},"cluster":{"id":"c_85bb1b913a1bef81dfee7c9b","canonicalTitle":"GPT-5.6 Solを最大14倍高速化する「Ultrafast」モードが登場","representativeArticleId":"a_40c18b4d883af7f0931946d3","sourceCount":1,"writtenSourceCount":1,"writeAttempts":0,"isSolo":true,"entitiesJson":"{\"anime_titles\":[],\"manga_titles\":[],\"work_titles\":[\"GPT-5.6 Sol\",\"Ultrafast\"],\"studios\":[\"OpenAI\",\"Cerebras\"],\"people\":[],\"type\":\"announcement\",\"domain\":\"other\",\"is_roundup\":false}","contentType":"news","status":"published","firstSeenAt":"2026-08-14T01:43:00.000Z","lastSeenAt":"2026-08-14T01:43:00.000Z","updatedAt":"2026-08-15T09:36:45.000Z"},"attribution":[{"source":"GIGAZINE","url":"https://gigazine.net/news/20260814-gpt-5-6-sol-ultrafast/","title":"GPT-5.6 Solを最大14倍高速化する「Ultrafast」モードが登場"}],"entities":{"anime_titles":[],"manga_titles":[],"work_titles":["GPT-5.6 Sol","Ultrafast"],"studios":["OpenAI","Cerebras"],"people":[],"type":"announcement","domain":"other","is_roundup":false},"keyFacts":null}