NEWS

DeepSeek V4 Flash routed to Novita on Vercel AI Gateway is 90% off for Pro customers through August 11, dropping the effective rate to $0.014 input / $0.028 output per million tokens. Here is the verified before/after math, the provider-pinning setup, and what a 10x cheap agent loop means for routing decisions.

Liquid AI shipped LFM2.5-2.6B on August 4, 2026: a 2.6B open-weight model trained for agentic work inside real harnesses, decoding at 220 tokens/s on an M5 Max and 113 tokens/s on a Ryzen CPU. Here is how it was trained, what the benchmarks say, and how to run it.

Next.js 16.3 ships the biggest update since 16.0: opt-in Instant Navigations with partial prefetching, up to 90% less dev-server RAM, cached repeat builds up to 5.5x faster, native Node.js streams for SSR, and an AGENTS.md block that points coding agents at version-matched docs.

Databricks measured the same model through different coding harnesses and found cost per task varied more than 2x at identical quality. Pi's minimalism explains why: roughly 1k tokens of system prompt and 3x less context per turn.

The v0 API is now generally available: programmatic, headless access to v0's app-building agent. Send a prompt, get a running app with a live preview URL you can embed, then deploy to Vercel in one call. Here is what changed, how the sync/async/streaming model works, and how it fits in an agent loop.

On August 4 Cloudflare launched the Agent Development Lifecycle: agent traces with session replay, @cloudflare/ci for CI/CD as Workflows, and local OpenTelemetry. A software factory is no longer just an idea, it is a platform product.

Cloudflare launched a single endpoint that returns account usage and cost per product in a FOCUS-aligned shape. For teams whose agents provision infrastructure, the dashboard is no longer the only way to see what a month costs.

Cloudflare's new CI SDK runs pipelines as Workflows: TypeScript instead of YAML, cached sandbox steps, artifact-push triggers, and a healing agent that fixes failed builds. Here is how it works and what it means for platforms.

Day three of Agents Week brought Cloudflare Wallets: Account Wallets for humans and Virtual Wallets for agents, x402 stablecoin micropayments for APIs and content, and human-readable agent identity at handles like research.example.cloudflare.pay.

A Microsoft Research analysis of 3.2M users and 761M LLM calls shows coding agent traffic is 87% agent-initiated, burns KV cache at turn boundaries, and punishes every tool failure with up to 4x compute.

GitHub Enterprise Importer now supports self-serve GitLab to GitHub migrations in GA. gh gl2gh exports GitLab projects, transforms merge requests into pull requests, and stages archives in GitHub or your own blob storage. Here is what actually moves and what you rebuild.

Shieldstral is a 3B-parameter Apache 2.0 multimodal safety classifier that takes your moderation policy as a plain-language question at inference time, scores content 0-1 in a single forward pass, and runs on one 16GB GPU. It beats 12B-20B guard models on text safety and sets state of the art on multimodal benchmarks.

Cloudflare's Agents Week opens with @cloudflare/computer, an open-source agent runtime where an SQLite-backed workspace gives every agent a shared filesystem and lets the model pick between fast isolates and full Linux containers per task. The bet: containers for under 10% of agent work.

Cloudflare published the serving playbook behind Workers AI running Moonshot Kimi K2.6 and Zhipu GLM 5.2: FP8 KV caches double Kimi's resident context to 1.37M tokens, INT4 weights shrink GLM 5.2's checkpoint 40%, and a page-tagging integrity check protects the shared cache at under 1% overhead. The numbers show what actually matters when open frontier models run on GPU fleets.

Cloudflare announced inbound TCP connections and gRPC support for Workers and Containers as part of Agents Week: a connect() handler on Spectrum, full-duplex gRPC from Containers, and automatic gRPC to gRPC-web translation so Workers can serve gRPC APIs without a container. Private beta today.

Cloudflare's JavaScript-native RPC on Workers now works across languages: TypeScript Workers can call methods on Python Workers and vice versa, with live objects, functions, and streams crossing the boundary. Pyodide's FFI handles type conversion, so no schemas, no protobuf, and no serialization code are needed. Available now.

Octane is a new MIT-licensed UI framework from Inferno's creator that compiles React-style hooks, Suspense, and actions to direct DOM code. No virtual DOM, no rules of hooks, no hand-maintained dependency arrays. Here is what shipped, the benchmark grid, and what it means for teams and AI agents.

OpenAI published the engineering story behind GPT-Live, its third-generation voice system: a full-duplex model with no turn detector, Go replacing Python on the media path, seamless stateful handoffs, and WARP, a new WebRTC transport going through the IETF.

Alibaba released Qwen 3.8 Max on August 3, 2026 - a 2.4T-parameter MoE with 95B active per token, a 1M context window, and $2/$6 per million tokens on QwenCloud. It leads PaperBench at 93.0, and the weights open next week.

A final Digital Markets Act decision requires Alphabet to give third-party AI assistants the same Android access Gemini has: DSP wake words, ambient sensors, screen automation, on-device models, and fair background execution. Home Assistant's three-year fight over the 'Okay Nabu' wake word shows exactly what the ruling unlocks.