For Devs

Perplexity has shipped hybrid compute for its Mac app, splitting a single Perplexity Computer task between frontier models in the cloud and a compact model running on the user's machine. Tasks start in the cloud for search, planning and reasoning, then hand sensitive steps down to the Mac without restarting or losing context. An on-device privacy gate decides what may cross: keep local, mask, refuse, or ask for consent. Perplexity also open-sourced the 0.6B classifier behind that gate, which records the highest character F1 (0.629) among twelve detectors evaluated. Hybrid compute is live now for Pro, Max and Enterprise on Apple silicon Macs with 24GB of unified memory.

Google released Gemini 3.8 Flash and Gemini 3.8 Flash Cyber on September 2, 2026. Both variants run on the same foundational intelligence, split by safety mitigations rather than model size. Gemini 3.8 Flash is generally available at $0.75 and $3.75 per 1M tokens, introductory through December 31, 2026. Flash Cyber reaches 47.2% pass@1 on CWE-Bench and is restricted to vetted defenders through the Fairwind Program. This article covers the benchmarks, the token-for-accuracy tradeoff, and what deploying either one actually requires.

Anthropic announced Enterprise Frontier Safeguards on September 1, 2026, an architecture that stores monitoring data in the customer's own cloud account rather than Anthropic's. Detection stays automated and Anthropic-run; custody, encryption keys, and flag review stay with the customer. Built with more than 100 enterprises and rolling out in phases this fall.

Vercel has open-sourced vgpu, the WebGPU library it built to ship the shaders on vercel.com. It treats .wgsl files as importable TypeScript modules, runs the same shader in the browser, in headless Node.js via Dawn, and in a deterministic CI mock, and ships a fullscreen effect in 25 KB gzipped.

Four falsifiable conditions for agentic coding replacing juniors, tested against METR, OpenAI, DORA and Stanford primary source evidence

Vercel and Ora launched Is Agentic, a free audit scoring website readiness for AI agents across 118 checks.

In this tutorial, we explore how to design production-grade safety for LLM-based applications using the NeMo Guardrails framework. We move beyond simple prompt filtering to implement a layered architecture, featuring deterministic PII redaction, retrieval filtering, output masking, and policy-based tool gating. By integrating stateful multi-turn evaluation and detailed activation tracing, we demonstrate how to build an auditable, secure, and cost-effective AI assistant capable of managing sensitive financial interactions while maintaining strict compliance standards

NVIDIA has released TensorRT Model Connect (TRTMC) in public preview, an Apache-2.0 project that takes a supported Hugging Face or local checkpoint to end-to-end TensorRT inference in two commands, with no intermediate ONNX export. The build emits a versioned .bundle artifact that runs through native C++ task APIs, so inference executes without PyTorch in the runtime path. NVIDIA's July 29, 2026 GB300 snapshot covers 105 release profiles across 76 model families.

Google has open-sourced SAM (Sovereign Agent Mesh) under Apache-2.0 — and it has nothing to do with Segment Anything. SAM is a zero-config, zero-trust P2P overlay that lets autonomous agents discover and call each other's MCP tools across cloud, on-prem, laptop and edge environments, without exposing a single internal endpoint to the public internet. Identity flows from OIDC into Biscuit capability tokens, so nodes authorize every request offline under a strict default-deny model.

Z.ai released GLM-5.3 on August 14, 2026. The model reuses the 743B GLM-5.2 base unchanged. Every reported gain comes from scaled post-training: more long-horizon task environments, more environment types, longer training. Terminal-Bench 3.0 moves from 4.6 to 28.3, and DeepSWE v1.1 from 46.2 to 66.9. Cybersecurity moved further than Z.ai says it planned, with CyberGym at 84.5% and ExploitBench more than doubling to 54.4%. Weights arrive in about two weeks.

Prime Intellect has open-sourced Prime Agent, a coding and research harness built on two abstractions: the Recursive Language Model, which turns sub-agent calls into functions inside a persistent IPython kernel, and the Continual Harness, which lets the agent edit its own prompts, skills, memory, and sub-agent specs mid-run. With Opus 5 it reports 95.5% RHAE Best@1 on ARC-AGI-3, above the reported human expert baseline of 95.4%.

Meta Superintelligence Labs has released Muse Code, a terminal coding agent in beta, powered by the new Muse Spark 1.2 model. Muse Code plans changes, writes code, and validates results across large repositories. Async background agents stay active for the whole session instead of spawning per task. A local append-only event log makes the runtime replay-exact and restart-safe after a crash. Muse Spark 1.2 was co-trained with the harness and trained on long-horizon, repository-scale work.

CopilotKit has published the Channels SDK, an MIT licensed library that runs an existing AG-UI agent inside Slack and Microsoft Teams. Version 0.5.0 ships five platform adapters and a documented runtime contract. This breakdown covers the verified deployment paths, the baseline requirements, and the one dependency that is easy to miss

Alibaba's Page Agent runs as client-side JavaScript inside the webpage. It reads the live DOM as text, then clicks and types from natural-language commands. No screenshots, no multimodal model, and no backend rewrite are required.

Perplexity has launched Brain, a self-improving memory system for its Computer agent. Instead of remembering the user, Brain remembers the agent's work — what worked, what failed, and what corrections got made. It builds a traceable context graph, reviews it overnight, and reports early gains in correctness, recall, and cost.

We break down Google Cloud's new Open Knowledge Format (OKF), an open spec that formalizes the LLM-wiki pattern. We explain how a bundle works: a directory of markdown files with YAML frontmatter, where each concept needs only a type field. We cover the three design principles, the reference tools Google shipped, and how OKF differs from RAG. We include a working Python consumer and an interactive bundle explorer you can embed.

Moonshot AI has open-sourced Kimi K2.7-Code under a Modified MIT license. It is a coding-focused, agentic model built on Kimi K2.6, with a 256K context window and roughly 30% lower reasoning-token usage. Moonshot reports gains over K2.6 on six benchmarks, including +21.8% on Kimi Code Bench v2. The model is available via the Kimi API and Kimi Code.

Google released the Colab CLI, letting developers and AI agents run local code on remote Colab GPU and TPU runtime

The EAGLE team, vLLM, and TorchSpec jointly release EAGLE 3.1 to fix speculative decoding instability in production.

Most web applications still have no structured way for an AI agent to register. auth.md proposes a fix: a Markdown file apps publish at their domain that tells agents which registration flows are supported, which scopes to request, and how to get credentials tied to a real user — without a human filling out a form.

Tencent has open-sourced TencentDB Agent Memory, a fully local memory system for AI agents released under the MIT license. The project pairs symbolic short-term memory, which offloads verbose tool logs into a compact Mermaid task canvas, with a 4-tier long-term memory pyramid (L0 Conversation → L1 Atom → L2 Scenario → L3 Persona). It ships as an OpenClaw plugin and a Hermes Docker image, runs on local SQLite + sqlite-vec by default, and uses hybrid BM25 + vector retrieval with RRF fusion. Tencent's own benchmarks report a 61.38% token reduction and 51.52% relative pass-rate gain on WideSearch with OpenClaw, alongside PersonaMem accuracy moving from 48% to 76%.

Perplexity has open-sourced Bumblebee, an internal security tool it uses to protect the developer systems behind its search product, Comet, and Computer. Bumblebee is a read-only inventory collector for macOS and Linux developer endpoints. It scans npm, PyPI, Go modules, MCP configs, editor extensions, and browser extensions — without invoking any package manager or running any code.

OpenAI launched a $4B+ Deployment Company and Anthropic closed a $1.5B joint venture with Blackstone and Goldman Sachs — both built around the Forward Deployed Engineer model Palantir pioneered. Here is what FDEs actually do, why standard SaaS fails for enterprise AI, and what skills early-career AI engineers need to break into this role.

Vercel Labs has released Zero, an experimental systems programming language designed so AI agents can read, repair, and ship native programs without requiring human interpretation of compiler output. The language emits JSON diagnostics with stable codes and typed repair metadata, enforces capability-based I/O at compile time, and compiles to sub-10 KiB native binaries.

Poetiq's Meta-System automatically constructed and optimized an inference harness for LiveCodeBench Pro using only Gemini 3.1 Pro — no fine-tuning, no model internals. The same harness, applied without modification to GPT 5.5 High, Kimi K2.6, Gemini 3.0 Flash, and four other models, improved every one of them.

Cline has extracted its internal agent harness into an open-source TypeScript SDK called @cline/sdk, the same runtime now powering its CLI and Kanban, with VS Code and JetBrains extensions being migrated. The SDK is structured as a four-layer stack — @cline/shared, @cline/llms, @cline/agents, and @cline/core — with native support for plugins, subagents, CRON scheduling, checkpointing, and MCP connectors. On Terminal Benchmark 2.0, Cline CLI scored 74.2% on claude-opus-4.7, compared to Anthropic's published 69.4% for Claude Code on the same model. Install via npm install @cline/sdk. Requires Node.js 22+.