Cloudflare
Cloudflare frees 100 terabytes of memory in 1.1.1.1's DNS cache
Five Rust optimizations cut per-entry memory from 953 bytes to 420. The company reports insert throughput up 43% and lookup latency down 19%.
En son
Cohere
Cohere announces Parse 5, a model that turns documents into Markdown
The 2.3-billion-parameter vision language model converts PDF, presentation and image pages into structured text. The company's price is $1.50 per 1,000 pages.
Westlake University
Code World Model: code holds the world state, the video model draws it
Researchers from Westlake and NTU hand world evolution to a coding agent and visual rendering to a video model. Code and weights are released openly.
Tencent
Tencent quietly releases EVIE, a model that searches documents visually
The 8.5 GB open-source model searches the page itself rather than converting PDFs, tables and scanned files to text. It outranks models twice its size on ViDoRe and runs locally.
TIME
TIME publishes its 2026 TIME100 AI list
The magazine's fourth annual list of the most influential people in AI includes Sam Altman, Dario Amodei, Elon Musk, Mira Murati, Ilya Sutskever and Fei-Fei Li.
fal
fal's H3 Max tops the image-to-video leaderboard
Post-trained on the open-weights MiniMax H3, the model ranks first on Artificial Analysis's image-to-video board. It generates a 5-second 720p clip in under 3 seconds and is 50% off for a week.
Z.ai
Merge Gateway announces 90% off GLM 5.3 Flash through September
On the model gateway, input drops to $0.012 and output to $0.04 per million tokens. The company argues this puts cost per task far below rival models.
Anthropic
Claude Cowork gets a built-in browser that navigates sites and fills forms
Anthropic added a Chromium-based browser to its desktop app. It opens in the side panel when a task needs a website, and stays separate from the user's own browser, tabs and passwords.
Google's Gemini Omni 1.1 Flash extends video scenes up to 40 seconds
Google's new video generation model extends scenes in 10-second increments up to 40 seconds total, lets developers set first and last frames, and adds a draft mode starting at $0.03 per second.
Google lets purchased ebooks become sources in Gemini Notebook
Under an initiative called "Expert Intelligence", eligible Play Books purchases can be added to Gemini Notebook as sources, with answers grounded in the book and citing it directly. Over 100,000 titles qualify.
Google's Gemini 3.5 Transcribe turns speech into text in 85+ languages
The new speech recognition model strips filler words, follows self-corrections and formats text on its own. Google reports a word error rate of 4.0% for live streaming and 2.6% for recorded audio.
Anthropic
Anthropic opens 10,000 free and discounted Claude seats to scientists
Under a new plan for principal investigators at academic and nonprofit institutions, standard seats are free and seats with 5x usage limits cost $15 a month for one year.