#Deployment/Engineering

AITencent Hunyuan extends classical critical-batch-size theory to online LLM reinforcement learning, where models generate their own training data. Across GRPO and PPO, learning-rate retuning preserves learning per response over a bounded range of batch sizes. On fixed hardware, larger batches raise PPO generation-stage throughput by up to 2.29×, and the best measured GRPO setup reaches the same validation target in 29% less time.

AIOpenClaw lets users keep public GitHub issues, PR discussions, commits, and diffs open beside their chat. Several references can be followed in tabs. The feature is read-only and public-only, so private content and edits remain on GitHub.

AIAfter a restart, OpenClaw can return unfinished conversations with their saved history, progress, and tool results. The agent checks earlier actions before continuing interrupted helpers, while chats the user stopped remain stopped.

AIAlex Wortega releases openjev, a family of Jev-style models at 0.8B, 2B, 4B, and 35B-A3B for reranking, grading, guardrails, and real-time game decisions. The models come with SGLang serving support, according to the post.

AIMeta Glasses can now function as FDA-cleared hearing aids, according to Mark Zuckerberg. He says they are medical grade, help with mild to moderate hearing loss, and cost a fraction of a typical hearing aid.

AILiquid AI released an experimental DSpark draft model for its LFM2.5-VL-3B vision-language model, delivering decoding throughput gains of up to 2.66× on GPUs and 3.13× on edge devices. The drafter adds about 280M parameters, an 8.9% increase in the deployed model's parameter count, and is available on Hugging Face with support in llama.cpp, SGLang, and MLX-VLM.

AIMeta is extending its Private Processing confidential computing infrastructure to AI glasses, running AI models inside confidential virtual machines so that even Meta cannot access user data. The system relies on hardware Trusted Execution Environments, with remote attestation checked by clients before any data is sent. Meta first introduced Private Processing in 2025 for WhatsApp and the Meta AI app.

AIvLLM now supports Gumbel-max watermarking, which embeds a keyed signal into generated text without changing the expected token distribution. Detection requires the secret key and tokenizer, and the signal accumulates over longer outputs. Benchmarks on Qwen3.5-27B with MTP-3 show throughput changes between -1.1% and +2.0% across batch sizes, with no consistent slowdown.

AIGoogle Developers reproduced Ai2's Olmo 3 7B from scratch in MaxText on Google Cloud TPUs, covering both the stage-1 pre-training run and the stage-2 mid-training anneal. The match was checked on held-out C4 loss, an 8-task accuracy suite, multi-domain perplexity, and token-level KL, not just the training loss curve. The post also describes a data-loader bug that made training loss look better than the reference while held-out metrics did not move.

Why it matters: The post documents how a faithful reproduction was verified on held-out metrics, including a data bug that training loss alone would have hidden.

AIAmp users can now share a runner with their workspace by starting it with --share, letting everyone spawn threads on that machine from ampcode.com. Shared runners appear under Shared Runners in the picker, and --amp-env gives them workspace and project Secrets & Env Vars but never personal ones. Amp warns that collaborators run code as the owner with their files and credentials, so sharing should be limited to trusted people, and workspace admins can disable runner sharing in Member Settings.

AIGoogle Cloud API Gateway now acts as a remote MCP server in Public Preview, making REST operations in an annotated OpenAPI 3.0.x or 3.1.x spec available as agent-ready MCP tools. Existing JWT or API-key authentication, quotas, and logging apply to MCP calls, so teams do not need a separate MCP server. Current limits include no support for OpenAPI 2.0, a maximum of 1,000 tools per gateway, and no MCP and model routing in the same API config.

Why it matters: The post shows how an existing OpenAPI spec becomes agent-callable MCP tools, with the same auth and quota policies applied, which helps teams avoid building a separate MCP server.

AIThe Amp macOS app now starts a runner automatically, so threads can run on your Mac without keeping amp --no-tui open in a terminal. Users add folders or projects under Runner in App Settings, then select "This Mac" when starting threads from ampcode.com, a phone, or Puck. A Keep This Mac Awake option prevents sleep while the runner is on and the Mac is plugged in, though the screen still turns off and locks.

AIClaude builds a model of a program's most complex parts, such as state machines or race-prone code, and searches that model for counterexamples that signal suspected bugs. It then reproduces those bugs and fixes them in the code. The post clarifies that the whole codebase is not formally verified, only the riskiest sections are modeled and checked.

AIGlobal health organizations including CEPI, WHO AFRO, and INRB Kinshasa are using Claude to accelerate their response to an outbreak of an unusual Ebola variant in the Democratic Republic of the Congo. The post links to a full Anthropic feature on the Ebola response for more detail.

AIClaude Code cloud sessions are now generally available, letting work continue while a laptop is closed. Existing Pro subscribers get a one-time $100 credit and Max subscribers get $250 to try them.

AISemiAnalysis published ClusterMAX 3.0, a rating of GPU cloud providers covering 77 neoclouds across 323 tracked providers. The report tests managed clusters through audit, performance, and reliability phases, and its results place Nebius alongside CoreWeave in the Platinum tier.

AIAnthropic's Claude Opus 5.5 and OpenAI's GPT-6 Sol are rolling out in Microsoft Copilot across Word, Excel, PowerPoint, Chat, Cowork, and Copilot Studio. The post says Work IQ supplies work context, so users can choose the model that fits each task.

AICursor has made Rollouts and Security Reviewer available today on Teams and Enterprise plans. The company is including Rollouts usage credits for the next 10 days so customers can try it on real changes. More details are available in Cursor's blog post.

AICursor's Security Reviewer now completes reviews in 3.8 minutes on average, down from 4.8 minutes, a 21% speedup. The update was released alongside Rollouts.

AICrusoe, a data center builder and cloud provider, raised $3.9 billion at a $30.9 billion valuation, led by Atreides Management, Valor Equity Partners, and Mubadala. Financials obtained by Newcomer show the company is on track for $2 billion in revenue this year.

AIAnthropic says it made claude.ai 3x faster in two weeks, and the same speedup applies to the Desktop app. A linked blog post explains how the team used Claude to measure, debug, and improve performance, and includes prompts and methods.

AICursor's Eric Zakariasson shared a prompt for improving an LLM agent harness to lower token cost per completed task without losing quality. The prompt covers the system prompt, tool definitions, cache layout, tool results, compaction, and subagents, and reports that one team's round of these changes cut overall token cost about 7%.

Why it matters: The prompt gives a concrete checklist for cutting agent token cost per completed task, with tested figures on cache layout, tool offloading, and compaction.

AIInferact says vLLM maintainers used megakernel optimization to reach 700 tokens per second per user on TPUv7 running Kimi K3. SemiAnalysis, which shared the work, reports this is 56% better performance than Nvidia's GB200 NVL72. Inferact links a full technical breakdown of the TPU megakernel work on its blog.

AIGitHub rebuilt the pull request view in the GitHub Copilot app to keep review fast on very large diffs, testing it on an open source pull request with 2,200 files, over a million changed lines, and more than 400 inline review comments. The core difficulty is that review comment heights can only be measured at render time, which breaks the fixed-geometry virtualization used for code-only diffs. GitHub split the document height into a deterministic code domain and a separately measured domain for comment blocks.

AIGoogle announces support for local AI models in the Antigravity SDK, per a linked developer blog post. The post itself offers no further details, so specific features, supported models, or limits cannot be confirmed from this source.

AIWrite-up and megakernel: 2/2

AIChatGPT Voice can now use plugins such as email, calendar, and Slack, and it can be powered by GPT-6 Astra, Sol, and Luna. It is also available in ChatGPT Work on web and mobile for creating docs, decks, sites, and spreadsheets by voice, rolling out globally in the latest app version.

Why it matters: The quoted OpenAI post names the new voice tool access, supported models, and Work integration, showing how voice now acts across workflows.

AIGoogle is shipping Google Beam units to customers in the U.S., Canada, U.K., France, Germany, and Japan, supported by 18 channel partners, with HP Dimension with Google Beam as the flagship hardware. Starting in October, users can book Beam at select Industrious locations in Atlanta, Chicago, New York City, and Palo Alto, and an internal eight-week Google study reported 50% more connection and 21% fewer follow-up meetings.

AIMicrosoft AI announced that its MAI-Image-2.6 image model can be tried on Microsoft Foundry and OpenRouter. The post provides links to both platforms but no further details on capabilities, pricing, or benchmarks.