OpenAI announced that ChatGPT Images 2.5 is now available to ChatGPT, ChatGPT Work, and Codex users, covering desktop, mobile, and web platforms. The official announcement highlights faster generation, higher fidelity, consistent detail preservation across multi-round edits, and comment-style editing that modifies only specified areas. The API simultaneously provides Flare and Sunburst: the former leans towards rapid generation, the latter towards high-precision creative work; Sketch allows users to first draw a sketch within ChatGPT, then hand the intent to the model for completion.<\/p>\n
Sources:<\/p>\n
Visible tests from multiple bloggers pinpoint the upgrade to specific aspects: @ZHO_ZHO_ZHO believes material texture, feature preservation, and multi-style blending have significantly improved, with the noise artifacts from the previous generation reduced; @LufzzLiz used a simple “reference image, convert to a certain style” to complete a doodle character transformation. Limitations also emerged: ZHO_ZHO_ZHO noted that after approximately 10 consecutive generations, the task might be lost, while @94vanAI reported that portrait results are not universally better and distortions can still occur.<\/p>\n
Sources:<\/p>\n
OpenAI announced that Astra is now fully available to Plus, Pro, Business, and Enterprise users, and has been integrated into Codex and ChatGPT Work. User-side cases show it has been used for tasks requiring sustained planning: @derrickcchoi relayed that Astra completed Portal in 24 hours at a cost of $571; @uguraksay relayed that Higgsfield generated 13,038 annotations from 81 frames of warehouse surveillance footage. Both are visible cases from bloggers or product parties and are not equivalent to official, unified performance conclusions.<\/p>\n
Sources:<\/p>\n
OpenAI publicly stated that its internal models and multi-Agent system have produced a solution to the Navier\u2013Stokes Millennium Problem, with formalization in Lean completed by GPT-6 Astra; the official statement only explicitly describes it as “sharing a solution,” not treating community review as concluded. @dotey’s summary indicates the related system involved approximately 10,000 Agents and 2.7 million messages, with the solution emerging after about 88 hours, followed by about 17 hours for formalization. The significance and controversy of this event are high, but the Lean code still requires external mathematicians to run, inspect, and follow formal publication and long-term community review processes.<\/p>\n
Sources:<\/p>\n
@tianyi stated that after the official launch of DeepSeek V4.1 Flash and before the launch of V4.1 Pro, requests originally destined for V4 Pro are being routed to Flash and billed at Flash’s price, citing its superiority over V4 Pro in metrics like performance, cost, speed, and total processing time. This point in the target day’s data primarily comes from blogger accounts; @Gorden_Sun cautioned that model scale and world knowledge may still affect non-coding scenarios, so “faster and cheaper” does not equate to being stronger for all tasks. The daily report treats this as a visible signal of API changes, not an unverified official conclusion.<\/p>\n
Sources:<\/p>\n
@levelsio shared his workflow: replacing multiple SaaS services like weather, image processing, moderation, monitoring, customer service, and scraping with self-built services, estimating monthly savings of approximately $25,000; among these, the self-built scraper costs about $1\/month with a target success rate of around 90%, compared to the previous $249\/month for Scrapingbee. This case supports the judgment that small teams can already bring some standardized capabilities back in-house, but infrastructure like domains, email, storage, backups, servers, and models are still retained, indicating that “de-SaaS-ification” is not unconditional self-hosting.<\/p>\n
Sources:<\/p>\n
@vista8 announced registration information for the second GEO Conference: the event is scheduled for November 21\u201322, 2026, in Nanshan District, Shenzhen, and simultaneously organizes the GEOFlow, example sites, course materials, cited data research, and dataset paper repositories. The referenced @yaojingang project TokEMS has been iterated 140 times and is open-source, positioned as an integrated, self-hosted conference operations platform covering official websites, registration, ticketing, orders, invoices, notifications, and on-site check-ins. The valuable information here is not just the event announcement, but also includes reusable resource portals and software assets.<\/p>\n
Sources:<\/p>\n
@XCreators announced that eligible creators can now apply for Original Content Rewards in Creator Studio, while Revenue Sharing will end, with the final old payout on September 11th, and the first OCR payout for approved applicants scheduled for September 25th. Tests by @vista8 and @LufzzLiz show that the application requires selecting 10 representative original pieces from system-screened recent content, then waiting for review; LufzzLiz also advised choosing original content over ordinary citations where possible. This information is better understood as a platform rule transition and application pathway, rather than a guarantee of earnings results.<\/p>\n
Sources:<\/p>\n
Statistics: Timeline scan count=720 Matched blogger count=71 Matched tweet total=509 Weighted tweet score=409.25 Original tweet count=226 RT tweet count=85 Crawl attempt count=5 Boundary coverage status=max_count_reached_before_confident_boundary<\/p>\n","protected":false},"excerpt":{"rendered":"
OpenAI released an image generation upgrade and opened access to its agent model on the same day, also sharing its progress on solving a Millennium Prize math problem, marking new AI breakthroughs in creativity and reasoning.<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[5],"tags":[19],"class_list":["post-1704","post","type-post","status-publish","format-standard","hentry","category-brief","tag-x--ai-"],"_links":{"self":[{"href":"https:\/\/en.blog.liu-qi.cn\/index.php\/wp-json\/wp\/v2\/posts\/1704","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/en.blog.liu-qi.cn\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/en.blog.liu-qi.cn\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/en.blog.liu-qi.cn\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/en.blog.liu-qi.cn\/index.php\/wp-json\/wp\/v2\/comments?post=1704"}],"version-history":[{"count":1,"href":"https:\/\/en.blog.liu-qi.cn\/index.php\/wp-json\/wp\/v2\/posts\/1704\/revisions"}],"predecessor-version":[{"id":1705,"href":"https:\/\/en.blog.liu-qi.cn\/index.php\/wp-json\/wp\/v2\/posts\/1704\/revisions\/1705"}],"wp:attachment":[{"href":"https:\/\/en.blog.liu-qi.cn\/index.php\/wp-json\/wp\/v2\/media?parent=1704"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/en.blog.liu-qi.cn\/index.php\/wp-json\/wp\/v2\/categories?post=1704"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/en.blog.liu-qi.cn\/index.php\/wp-json\/wp\/v2\/tags?post=1704"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}