Find published research by company, product, platform or technology.
Showing 21–40 of 248 dossiers
The useful part of Smaug Agentic is not another frontier-style benchmark claim. Abacus.AI is publishing a drop-in Kimi K3 derivative that targets a specific production failure mode in coding agents: long runs that burn the reasoning budget without converging. The weights and model card are public, but the training data is not disclosed and the benchmark gains remain vendor-produced.
The workflow shift is continuity rather than another model upgrade: one Kiro agent session can outlive the laptop that started it. Cloud configuration can also carry agent setup across environments, although enterprise governance is not identical between local and web/cloud surfaces.
GitHub Copilot can now turn Slack or Teams threads into collaborative cloud-agent sessions. Teammates can add context and steer the work in public, while repository permissions, agent budgets and optional extra PR approvals remain the main control boundaries.
Android Bench 2.0 moves coding-agent evaluation away from small repository fixes toward dependency upgrades, app builds, migrations and other jobs that can take a human engineer days. The results expose a much larger reliability gap than short-task benchmarks—and show that the agent harness can materially change cost and outcome.
The important failure is not another prompt injection. Plugin4Shell breaks the mechanism intended to guarantee that an AI-agent plugin is still the exact code a marketplace reviewed.
Google appears to have completed a talent-focused Mechanize deal: the startup still exists, but much of the team that builds coding-agent training environments and evaluations has moved into Google’s model-development work.
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.
GLM-5.3-Flash combines open weights, multimodal coding/agent capability and an 18B-active sparse architecture with a large anonymous pre-launch trial. Z.ai has already issued a chat-template correction for early downloads, showing that day-one self-hosted deployments need artifact-level validation as well as model benchmarking.
The Agent Host is becoming less like a background chat process and more like durable development infrastructure.
Docker’s new agent stack combines pay-as-you-go microVM sandboxes with an OCI-based Kit format for declaring what an agent can use. Cloud sessions cost from $0.07 to $1.12 an hour, and Docker says it plans to take the Kit specification toward CNCF neutral governance.
The useful part is not the 800,000-line headline. GitHub has published unusually detailed receipts for a production-scale agent-assisted migration: roughly $120,000 of token spend, 14.5 weeks of incremental releases, dozens of regressions, extensive compatibility tests and a workload-specific jump from 7.55 to 120 session lifecycles per second.
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.
Gemini 3.8 Flash keeps 3.7 Flash’s promotional per-token rate and Flash-tier latency, but early independent analysis suggests harder reasoning can increase tokens consumed per task. A separate 3.8 Flash Cyber model is available only through Google’s Fairwind defensive-security program.
The useful change is not another reporting dimension. GA properties can now discard events whose hostname is not approved, directly addressing Measurement-ID abuse and ghost traffic while reducing the maintenance burden of chasing new spam domains.
For agent and untrusted-code workloads, the useful change is not simply lower latency. Sandbox location becomes an explicit execution policy, so teams can align code execution with nearby data and avoid a resilience fallback quietly moving work outside an allowed region.
The migration risk is subtle: nothing breaks immediately, yet ERP, marketplace, POS and supplier integrations can become incomplete as soon as merchants start attaching multiple UPC, EAN, GTIN, ISBN or ASIN identifiers to one variant.
The interesting change is architectural rather than another storage feature: migration becomes a server-to-server transfer initiated through an S3-compatible PutObject or UploadPart call, with range and multipart support for large objects.
The change is not about where database rows live; Cloud SQL already has regional instance placement. It changes where API control traffic is processed, reducing dependence on global frontend infrastructure and making data-in-transit boundaries easier to align with sovereignty requirements.
The counting-rule change is no longer theoretical. Early post-cutover data suggests public views can materially outpace Engaged views, with the size of the gap varying by channel size, category and discovery surface.
The limits themselves were already documented; the material change is enforcement. Free-tier D1 workloads that previously relied on soft overage behavior now need query-cost awareness, indexes and a plan for temporary failures or paid migration.