All Stories — Page 57

Microsoft's Superintelligence team released specialized MAI models that claim to match GPT‑5.6 in Excel and beat GPT‑5.4 Mini in Copilot while using fewer tokens.

The August 1st rates also reduce request, storage and data-transfer charges, while database pricing stays flat and one legacy VM tier gets pricier.

Shensi Ding says a panel of cheaper models matched Claude Fable 5 on a 25-task research test at one-quarter of the cost.

Video enters limited early access first, while image generation, open weights and robotics access will follow in phases.

Poke will remain available while its founders join Cognition, which plans to apply the assistant's proactive messaging design to its coding agent.

The deal brings HRL's spin-qubit and quantum-sensing expertise into IBM's expanding quantum portfolio, with Boeing and GM staying as partners.

Louis Beaumont's 2024 project has 20,000 GitHub stars, but commercial use of its once-MIT code now requires a paid license.

The new text‑to‑speech model adds inline tag steering for emotions and breaths, and supports free‑form natural‑language prompts.

Chaofan Shou says Kimi K3 found 19 flaws in 90 minutes; public code documents an authenticated exploit against Redis's latest release.

Jason Marks is using contract manufacturing and a $41,520 mini pickup to avoid the factory spending that buried earlier EV entrants.

Founder Justin Haubrich is turning a $24.99 Mac utility into a consumer data-discovery tool that VaultSort says makes zero network calls.

Clement Delangue's open-model bet became incident-response infrastructure after hosted APIs rejected the attack logs.

Independent tests put it level with 3.5 Flash on intelligence, with half the task time and an 18% lower cost per task.

Chris Best is betting that reader-triggered Pangram scores and creator disclosures can preserve authorship without banning AI-assisted writing.

Alexander Borzunov and collaborators' 2022 research project trades cloud control for model access, while privacy, latency and incentives remain constraints.

Benjamin Davis and Philip Mims spent $99.59 on 650 persistent text-world runs across 13 models, then treated judge instability as the result.

Cognition builder Jared Zoneraich ran roughly five Ultra-mode agents per problem, with one counterexample checked in Lean.

The platform stores all activity as signed events on a user‑run relay, aiming to replace fragmented chat, code and CI tools.

A Reddit discussion shows heavy Claude users moving to Kimi's K3 model after pricing and openness concerns.

The researcher disclosed CVE‑2026‑3854, a critical RCE flaw, prompting the largest payout in GitHub's Vulnerability Reward Program to date.

Ayush Agarwal's team is offering developers a persistent canvas, shared browser and permission layer for agents that work alongside people in real time.

Jeff Cui shows that a frozen ViT backbone plus a tiny transformer can outperform a 7B‑parameter VLM in robot control.

The filing lists five investors and directors from a16z and Dell, but leaves the buyers, round label and valuation undisclosed.

Founder Matthew Wong introduces Ask Poth, a tool that unifies data from HubSpot, PostHog, Pylon and Granola into a single customer graph.