Find published research by company, product, platform or technology.
Showing 1–20 of 50 dossiers
The post-release evidence sharpens the original story. Qwen3.8-27B can retain useful agentic-coding performance at practical 4-bit sizes, but local model quality is not a property of the checkpoint alone: quantization, reasoning effort, context handling and the agent harness can materially change the result.
Safari 27 blocks third-party email-to-ID resolvers by domain. First-party personalisation remains allowed; the affected vendor list and measured advertising impact remain unverified.
WebKit's Safari MCP server is available in stable Safari 27.0 as well as Technology Preview. It runs locally, but captured page data passes to the connected agent; remote automation must be enabled explicitly.
SwarmLLM does not route whole prompts to separate machines; it pipelines one model across browser tabs. A MacBook and iPhone can jointly hold Qwen 3.8 27B even when neither device can hold the full 15GB quantized model alone, with no inference server in the loop.
Six of eight SVG sanitizers in a PortSwigger test had bypasses; developers must review SVG upload and rendering contexts, CSP and library versions.
JPEG XL now ships by default in Chrome 155 and Firefox 158; Safari supports still images, but Edge, installed-version lag and format tooling still warrant fallback delivery.
A major UI framework is moving from legacy Sass and JavaScript conventions to native CSS and ESM. Existing Bootstrap 5 projects need a deliberate migration plan, not a blind version bump.
WebMCP is no longer a Chrome-only browser experiment: Microsoft Edge now has its own active origin trial, while ChatGPT’s built-in browser and WordPress Playground show agent-client and platform implementation paths.
OpenAI's Ultrafast tier has moved beyond its GPT-5.6 preview: GPT-6.1 Sol access is broad and pricing is documented at 6× Standard.
The follow-up clarifies package upgrade behavior and provides a practical migration starting point for developers moving away from v8.
Astra's adoption question is no longer only model capability. Builders can now model its long-context economics and task-level efficiency, while enterprises get a more explicit control plane for computer use. The same release also raises the cyber-safety boundary: OpenAI says Astra is its first model to reach the Preparedness Framework's Critical cybersecurity capability threshold.
The PyTorch 2.15 RC changes what Apple-silicon developers can run locally: FP8, faster MPS decoding, new GPU operators and a video VAE memory fix. The reported 5–6.5x gain is not universal.
Connection Allowlists turn outbound browser networking into an explicit allowlist for Fetch and other web-platform APIs. The control is opt-in and currently Chromium-only, and strict policies can break legitimate dependencies such as FedCM identity-provider requests if teams omit required endpoints.
Cloudflare’s RUM measurement model now distinguishes hard navigations, native soft navigations and routing-API fallbacks. For React, Vue, Angular, Svelte and other client-routed sites, the immediate consequence is a metric discontinuity: pageviews and Core Web Vitals can shift without an underlying traffic change.
Developers can now train and evaluate adapters through custom code while Together runs the GPU-heavy operations. The flexibility comes with B200-only support, a $100 credit threshold and ongoing charges for idle ready resources.
The release is more interesting than another Qwen3.8 size point because Qwen is deliberately exposing the next architectural generation early. QSA sparse attention, gated residual streams and offloadable n-gram embeddings are now testable before the full Qwen4 family arrives.
Cloudflare Computer 0.7 makes agent code resumable and interruptible. The important catch: recovery replays code, and unrecorded external actions can be repeated even if they already succeeded.
Google Ads now documents positive-only AI review summaries inside Search ads in the US, UK, Australia and India. Advertisers cannot customise the text; source, opt-out and reporting controls remain unclear.
Copilot now offers model-reviewed action approvals between manual and Allow All. Unlike OS sandboxing, this probabilistic gate can approve unsafe operations; unattended CLI flags may leave unexpectedly broad permissions.
pg_ivm 1.17 fixes restore state, duplicated incremental-maintenance triggers and upgrade metadata handling, plus index inference and LATERAL crash bugs.