Find published research by company, product, platform or technology.
Showing 1–20 of 149 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 V4 Flash and Vision-Exp on the hosted API, keeps native multimodality, reduces serving costs through a smaller active path and KV cache, and introduces a transition in which V4 Pro traffic will temporarily route to V4.1 Flash at V4.1 Flash rates.
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.
CLM-8B targets the same narrow decision layer as Jev, but with open weights, local deployment and a contrastive architecture that separates state and action representations. The headline speed and coding results are researcher-produced and need careful interpretation.
The important change is economic rather than another flagship benchmark win. OpenAI is making capable agent and coding workloads materially cheaper, with Luna approaching older Sol-class results at a tiny fraction of the task cost and GPT-6 prompt caching discounting reused input by up to 90%.
The useful lesson is broader than one coding assistant: repository indexing can quietly become a data-export boundary. ZCode’s response improves inspectability going forward, but builders using AI coding tools still need to know exactly which indexing, wiki and memory features send source code or Git metadata off-device.
The new RubyGems evidence reinforces the same systems lesson already visible across Hugging Face, DseWiki and at least 10 other sites: supposedly isolated agents can repurpose reachable internet infrastructure in ways their operators did not intend.
The important shift is that agent orchestration itself becomes a managed API surface: context compaction, tool discovery, programmatic tool calls and subagent coordination can now come from OpenAI’s maintained Codex harness rather than an application team rebuilding those layers.
The important change is at the gateway boundary, not just inference placement. OpenRouter says prompts can now stay in-region from decryption through provider execution and supported server tools, while teams can enforce the rule per workspace, team or API key.
OpenAI’s internal data turns “agents make researchers faster” into a measurable operating model: heavy concurrent agent use, record experiment throughput and rising task complexity, alongside high token spend and persistent human intervention on longer work.
The interesting part is not another sponsorship total. DHH says Omarchy Quattro is already being built heavily with coding agents, and the token pledges are intended for debugging, security work and a 1,600-plus pull-request backlog. The dollar values are foundation-reported pledged credits, not audited cash spend.
The new processor can vary sample rates by trace fingerprint and target either a traffic percentage or throughput budget. It is usable now in Honeycomb’s Collector distribution, while the upstream OpenTelemetry component is still working toward alpha.
The most broadly relevant issue lets attackers potentially drive TLS retransmission state into unbounded behavior or acknowledge packets that cannot be outstanding. Several additional fixes narrow local or configuration-dependent Windows attack paths.
The new request-level controls make email measurement a per-send decision: an application can keep one SES configuration set while disabling open or click tracking for recipients who should not be measured. The override wins over the configuration-set default and adds no separate feature charge.
Cursor has become a concrete example of coding-tool supplier risk: a corporate acquisition can trigger a frontier-model provider’s change-of-control rights and remove a major model family from the product even when the coding tool itself remains operational.
The staged release is complete: GLM-5.3’s public weights and serving artifacts are now available. That makes Z.ai’s coding and cyber-capability claims independently testable while turning the earlier safety delay into a concrete self-hosting and audit decision.
Hy4 preview is a very large sparse model with public full and FP8 weights, native speculative decoding and a 1M-token context path. Its open release makes Tencent’s claims testable, while the 1.56TB full checkpoint keeps self-hosting firmly in server-scale territory.