Find published research by company, product, platform or technology.
Showing 141–160 of 196 dossiers
GLiNER2.5-Decide attacks the same bounded-decision layer as Jev and CLM from a much smaller encoder architecture. Its strongest benchmark claims are vendor-produced, but CPU deployment and constrained joint decoding make it a materially different option for software-facing AI decisions.
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%.
Muse packages persistent autonomous execution, credentials, payments, app access and memory into a mainstream consumer product. A September macOS hotfix now provides an early real-world lesson: agent containment has to protect not only the cloud runtime but also the local control path into the agent.
The important change is not simply that Claude can run several agents. Projects now owns decomposition, shared context, branch isolation and progress coordination across full Claude Code sessions, while the trade-offs become usage burn, cloud-only execution and ordinary merge conflicts when parallel work overlaps.
The dangerous detail is the delivery path: WordPress gives an unauthenticated commenter a moderation-preview URL for their own pending comment, and The Events Calendar can process attacker-controlled block markup from that preview before a moderator approves anything.
Jev’s launch claims were interesting; Vercel’s usage data is more useful. Nearly 13% of paid AI Gateway teams tried the typed decision model in its first day, while Jev also rose to a material share of gateway requests. That does not establish retention or production success, but it is unusually fast developer uptake for a model designed to make bounded software decisions rather than generate prose.
GitHub Actions now has enforceable actor and event rules before a workflow starts, plus a coming default block for a trigger that can expose repository secrets to untrusted fork code.
This is an identity-system failure rather than an application bug: a vulnerable Keycloak deployment can let an attacker turn the legitimate “forgot password” flow into full account takeover without credentials or victim interaction. Upgrade is the proper fix; disabling Forgot Password in every realm is Red Hat’s temporary mitigation.
The observe–test–release loop now has explicit economics: Free and Pro include 30,000 captured generations and 25 million system-initiated AI tokens per month; Pro overages start at $1.50 per 1,000 generations and $2 per million LLM Eval/Guard tokens, while ordinary telemetry is billed separately.
The practical change is bigger than another package-manager version. Homebrew can now tell operators whether vulnerabilities are actually outstanding in the formula revisions they installed, while its own recent advisories show why package-manager metadata, uninstall paths and build isolation deserve the same scrutiny as package contents.
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 material change is that model routing is no longer a single opaque optimization target. Developers can now tell Copilot whether to bias Auto toward lower cost, a middle ground or higher quality while GitHub still chooses a model prompt by prompt.
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.
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.
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.
This is a patch-and-hunt event rather than a routine Commerce security release. Exploitation began before the vendor fix existed, and Adobe plus independent responders recommend remediation that goes beyond installing the hotfix when compromise is suspected.
The new program creates a specialized Gmail deliverability path for verified political senders, but it is not an inbox guarantee: recipients can still mark mail as spam, block senders or unsubscribe, and non-compliant domains can be suspended or removed.
Android Studio’s agent layer has crossed an important boundary from preview features into the stable channel: domain-specific skills are preloaded and auto-selected, while Gemma 4 can execute tool-calling code tasks locally without sending source code to a cloud model.
X’s replacement creator program is now live enough to expose a new dependency: U.S. creators must route Original Content Rewards payouts through X Money, and one X Money account can connect to only one X account. Eligibility and qualified-impression rules remain unchanged.