Find published research by company, product, platform or technology.
Showing 1–20 of 349 dossiers
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.
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.
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.
Devin Fusion's persistent lead/sidekick architecture now has newer model choices and independently reported cost/quality comparisons, changing its practical deployment economics.
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 existing coverage-first mode keeps more query detail at the cost of possible series growth. Cost-first applies all reductions but only restores detail within the same series budget; it is a guardrail against metric-cardinality spikes, not a guarantee of a fixed bill.
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.
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.
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.
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.
A new buyer survey connects the shift from SaaS seats to agent usage with real budget anxiety: 46% report higher-than-expected agent costs and 45% say oversight eats more than a quarter of expected productivity gains. Preferences are not purchasing outcomes.
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 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 latest DRAM-driven increase forces makers, edge deployments and small teams to re-evaluate whether high-RAM single-board computers still offer the right economics.
Always-on RAM-heavy Fly Machines cost more from October, while bursty Sprites get much cheaper. CPU pricing for Machines and Managed Postgres plans stay unchanged.
PostgreSQL operators gain per-statement and per-transaction estimated-cost limits across versions 14–18, useful for runaway reports and ORMs. The guard is disabled by default and can be bypassed by users allowed to change planner cost parameters.
Buttondown’s 'Great Pruning' is a small-SaaS operations story about deleting architecture rather than adding it. The company removed duplicated or over-retained request and email-event data after changing how those workloads were processed.