All Stories — Page 34

Cerebras says disaggregation raised throughput using the same number of its systems without reducing token-generation speed. A separate General Compute agreement is scheduled to bring Cerebras inference capacity to customers in Q1 2027.

Founder Adam Pietrasiak says the model beats alternatives on speed and quality; Screen Studio's site says it runs locally and can upscale recordings by 6.5x.

Founder Juan Reyero is building a self-hostable communication layer that keeps messages available across sessions, while runtime integrations still determine whether agents wake up.

Co-founded by Harikrishnan Mulackal, Cantina built Apex Flash-1 with Yeta Labs and says it solved 40 of 60 held-out security tasks.

The Planet-built prototype launched on SpaceX's Transporter-18 mission on October 1st; Google says it has made contact and is beginning in-orbit tests.

The free app lets developers keep agents running, review code and pull-request status, and merge changes; Apple's listing dates version 1.0 to September 14th.

The Google Labs experiment turns selected family emails into shared briefs, calendars and tasks, with access limited to U.S. adults using personal accounts.

The June sample records 9.3 million model calls and 8.7 million tool calls, while leaving prompts, code, outputs and user identities out of the public files.

Reuters reported that OpenAI was searching roughly 50 petabytes of data; OpenAI's notices cover a range of behavior and do not establish that every recipient was breached.

Bloomberg traced the financing to deals begun in mid-2025; Hongjing's filings also show a separate sale-and-leaseback guarantee with the same lessor.

OpenAI says a July campaign generated 16,000 extraction-pattern requests over two days, but those figures measure attempts, not confirmed successes.

CEO Emmanuel Le Roux says the €80 million Angers expansion lifts monthly production to 12 racks, with room to reach 24 in 2027 if demand holds.

Two Form D notices filed October 1st report fully sold offerings of $8.28M and $8.68M by Prevalence Holdings, which the source material identifies as Precise.ai's parent. AdExchanger had reported a $7M seed round in June; the filings do not explain how that figure relates to the two offering totals.

Ai2 says its system can train larger sparse models with less overhead. Its trillion-parameter result measures system performance, not a completed model.

Anant Kale and Kunal Verma are extending AppZen's expense-audit business with models designed for receipts, policies and finance workflows; general availability is planned for December.

Infrastructure AI is seeking early-access users for GAOS. Its public materials identify no customers or pilots, disclose no funding, and report no measured results.

Modal is extending Erik Bernhardsson's pooled compute platform to distributed training and inference, with per-second billing and a single Python decorator.

Saronic CEO Dino Mavrookas is turning the autonomous-vessel bet into a $3.2 billion Brownsville shipyard, with Navy programs and a 10,000-job projection attached.

The preview model encodes document chunks together, and Perplexity says it leads on two retrieval benchmarks while using smaller vectors than Voyage's competing model.

Kodiak's October 1st anniversary post reported more than 40,000 cumulative hours and 3.5 million autonomous miles. Atlas owned and operated the 35-truck Permian fleet as of June 30th; IKEA's driver-out freight service is planned for later in 2026.

The September 28th launch pairs query-based passage extraction with company-run tests against Codex, Exa, Claude Code and Zilliz.

The global rollout pairs AI-generated outfit previews with saved favorites, while OpenAI warns that the images cannot confirm fit or size.

The open-source BootLoops project targets exact calculations, while Schwartz says human experts still have to decide which results matter.

Nebius said the open-weight model was live on Token Factory on September 28th; Alibaba's Qwen team highlighted the availability two days later.