Find published research by company, product, platform or technology.
Showing 1–20 of 283 dossiers
CS-4 combines three WSE-3 Turbo wafers with Cerebras’ Nexus rack design. The practical shift is architectural: compute, power and I/O become modular, while Cerebras now says the same platform is intended to support CS-5 in 2027 and a 3D-memory CS-6 generation after that.
Groq 3 LPX is moving from architecture announcement to manufactured infrastructure. Artificial Analysis measured about 3,400 output tokens/s at both 10K and 100K context on an NVIDIA-hosted private endpoint, but the single-concurrency benchmark does not yet establish public-cloud price, multi-tenant throughput or end-to-end agent speed.
Jalapeño is working first-party silicon rather than a roadmap item, and OpenAI now says AI itself materially accelerated the design process. The distinction still matters: tape-out means the design was finalized for manufacturing; it does not mean fleet-scale production qualification or API deployment is complete.
DeepSeek V4.1 Flash supersedes the old Flash and Vision-Exp API lines with native multimodality, lower pricing and new architecture. Unlike those retired Flash aliases, the current DeepSeek API changelog and rate card still show V4 Pro as a distinct service.
PyTorch 2.15 is now generally available. FP8 and faster MoE decoding improve some Mac AI workloads, while the vendor/contributor speed claims remain configuration-specific rather than universal.
GitHub's multi-model coding orchestration is now GA, with admin-controlled model pools and per-model charges. Local routing remains experimental and cloud-dependent.
A detailed engineering case study links a 68% AI-review cost reduction to prompt-cache structure and effort tuning, with security-review trade-offs still unresolved.
Google reversed its earlier ad-free approach to Personal Intelligence in AI Mode. Connected-app-derived insights may influence ads, with testing scope and advertiser controls still unclear.
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.
Meta has made the privacy-versus-price trade explicit in its Model API: developers can choose standard pricing or a contributor model ID with steeply discounted inference in exchange for training-data permission. The choice matters for proprietary code, customer data and AI SaaS workloads.
This is a hard capability removal rather than a routine model migration. Products built on OpenAI’s video-generation API now need another provider or a redesigned video path because the official deprecation table offers no successor endpoint.
The Assistants API shutdown date has passed. OpenAI’s deprecation documentation lists August 26, 2026 as the removal date and directs developers to Responses and Conversations for replacement workloads.
Zabbix 8.0 LTS adds OTLP/gRPC ingestion through proxies, APM trace views, ClickHouse history and complex event processing, with meaningful migration and security considerations.
OpenBSD 8.0 operators should apply the first signed errata or syspatch updates promptly, especially where IPsec AES-GMAC and NSD are used.
OpenSSH 10.6p2 and Alpine backports address a critical musl/no-PAM password bypass; operators must restart sshd after upgrading.
Kimi K3.1 Pro's full open-weight release is now public. Its 2.8T-parameter MoE has 104B active per token and a 1M-token context; local hosting requires serious infrastructure and checking Moonshot's specific licence.
Agent-heavy monorepos can overwhelm conventional Git hosting. Pando retains Git client compatibility while relaxing repository-wide consistency; its performance numbers are OpenAI-reported.
A free open-source video editor gains a scriptable, headless editing surface: import and trim clips, apply effects, inspect contact sheets, export and hand projects back to a human. Auto-Optimize separately speeds up 4K editing without lowering final output quality.
Reddit's Q3 shareholder disclosures set a direction towards logged-in browsing, not an immediate worldwide login wall. Q4 non-US tests follow declining search referrals, with consequences for community discovery, advertising and access to public conversations.
OpenAI's new API credit grants can lower the cost of prototyping apps and agents, but eligibility, expiration, API limits and excluded Codex usage matter.