Skip to content
#Deployment / engineering
October 27Tuesday
  1. Preferred Networks News60

    PFN releases PLaMo 3.1 Prime with Web search and Deep Research

    Preferred Networks began offering PLaMo 3.1 Prime on October 27, 2026, adding Web search and a Deep Research feature that autonomously plans, collects, analyzes and writes source-cited reports over several minutes to tens of minutes. Deep Research also runs on-premises and can connect to internal document search systems, while the company reports an average ~15% score improvement across eight hallucination-related benchmarks including AA Omniscience versus PLaMo 3.0 Prime. PFN also cites 1.7x faster processing via speculative decoding and a 90% reduction in output tokens in non-reasoning mode, and launched a free PLaMo API plan plus a PLaMo Chat campaign offering Deep Research twice daily until November 30, 2026.

    Why it matters: PFN's PLaMo 3.1 Prime adds autonomous Deep Research that runs on-premises, giving Japanese enterprises a locally deployable option for source-cited investigation work.

  2. Apple Newsroom58

    Apple Creator Studio adds AI assistants, MCP support, and MotionVFX in Final Cut Pro

    Apple introduced updates to Apple Creator Studio that bring an intelligent assistant built on Apple and third-party models to Keynote, Pages, and Numbers, plus Model Context Protocol support so external AI agents can read content and perform edits in those apps. Final Cut Pro for Mac now includes built-in MotionVFX and Content Hub stock video, Logic Pro gains a beta assistant and a Stem Splitter that separates drums, and Pixelmator Pro adds Generate Fill for natural-language image edits. The updates arrive on the App Store this November, free for existing subscribers, with new subscriptions at $12.99 per month or $129 per year.

October 26Monday
  1. Replit Blog38

    Replit: Every Fortune 500 Company Will Be a Self-Driving Company by 2030

    Replit predicts every Fortune 500 company will become a "Self-Driving Company" by the end of 2030, with agents executing and coordinating work while people set goals and remain accountable. Replit says its engineers nearly tripled code output in the first half of 2026 while review times, reversions and product incidents stayed flat. Replit for Business is being built as the foundation, and at SMFL a development process that previously took six people six months was completed by one person in one week.

  3. Mastra Blog40

    Mastra opens Knowledge, a self-hosted company brain for AI agents, to eight more design partners

    Mastra is opening its Knowledge service, a self-hosted company-wide context layer for AI agents, to eight more design partners on a first-come, first-served basis. Knowledge connects systems like Slack and Monday, keeps itself current, and supports separate knowledge bases per function with permissions at the individual, team and org level. Mastra says wiring a company-wide knowledge graph into its coding agents produced a noticeable improvement in shipped code quality.

October 24Saturday
  1. SemiAnalysis60

    SemiAnalysis tests Amazon Trainium3's native PyTorch training stack and Neuron Explorer profiler

    SemiAnalysis's engineering team tested Trainium3 racks on the TorchTitan and TorchNeuron stack and found the native PyTorch experience a major improvement over the old PyTorch/XLA approach, though bugs and long compile times remain. Training Qwen3-32B on 16 Trainium2 chips reached 14,156 tokens per second at 30.47% MFU in compiled mode, a 1.28x gain over eager mode, while the gpt-oss 20B recipe took 27 minutes and 17 seconds to compile. The team also rated Neuron Explorer as the industry's leading profiler, ahead of Nvidia's and AMD's tools, and reported Trainium3 passed 89 of 96 FlexAttention test cases.

  2. Google Research58

    Google Research outlines how its TB screening AI moved from benchmarks to certified clinical deployment

    Google Research described at IDWeek 2026 how its chest X-ray TB screening models progressed from a 2019 Radiology benchmark to a CE MDR Class IIb certified tool deployed by partner Nexus AI in about 40 healthcare facilities across six countries. More than 90 clinicians have evaluated over 25,000 chest radiographs, cutting turnaround from weeks to under 45 seconds, and the tool has also been deployed offline four kilometers underground at South Africa's Mponeng Gold Mine. Google has open-sourced chest X-ray embeddings and MedSigLIP through HAI-DEF and released MedGemma, which it says has passed 10 million downloads, while Indonesia's Ministry of Health is fine-tuning both models on domestic radiographs for a national screening program.

October 23Friday
  2. Amazon Science38

    The science behind TorchNeuron, Amazon's new PyTorch backend for AWS Trainium chips

    Amazon released TorchNeuron, an open-source PyTorch backend that translates PyTorch code into operations executable on AWS Trainium systems with minimal or no code changes. It adds an opt-in adaptive eager mode that groups and compiles operations for near-compiled performance, plus multi-level NEFF caching; a shared compilation service cut time to first training step by about a third in a 64-process LLM test. The Annapurna team tuned and benchmarked it using 841 representative models distilled from hundreds of thousands of open-source models.

  3. Modal Blog58

    How Modal and Sakana AI served a ~1T-parameter MoE model for the Sakana Namazu API

    Modal's forward deployed engineering team worked with Sakana AI to run Sakana Namazu, a ~1T-parameter MoE model tuned for Japanese business context, on Modal for its public API launch. The team applied selective NVFP4 quantization to routed experts only, keeping attention, the router and shared experts in BF16, and calibrated on Japanese-heavy conversations so all 23,040 experts saw representative traffic. A custom DFlash draft model trained on more than 30 billion raw token IDs raised speculative token acceptance by 25% and cut per-token decode latency by about 35%, while a second drafter trained on roughly 1M real requests lifted average acceptance length from 3.51 to 5.01 and cut p99 latency from 68s to 52s.

October 22Thursday
  1. Sakana AI Blog58

    Sakana AI launches Sakana Aquarium, an ontology-based AI agent platform for enterprise decisions

    Sakana AI announced Sakana Aquarium, a platform for building, running and operating AI agents that support organizational decision-making and execution on top of an ontology. The platform combines three parts: an Ontology that organizes knowledge, rules and data with their meanings and relationships while connecting existing sources without migration; Governance that sets access permissions and pre-execution approvals and logs everything from decision rationale to execution results; and Managed Agents that deploy workflows covering information lookup, AI judgment, function execution and human approval. Sakana AI says the data connections and ontology can be reused across agents, and that the platform grew out of its FDE-led work with financial institutions such as MUFG, SMBC Group and Daiwa Securities Group.

  2. Modal Blog71

    Modal releases Spindle, an open-source Tinker-compatible training server

    Modal released Spindle, an open-source, Tinker-compatible training server that runs in a customer's Modal workspace, letting teams customize the stack and pay for GPU time instead of tokens. Spindle supports full-parameter training and multi-LoRA across open models, and Modal reports that sharing one trainer across 8 clients cut cost per 1M output tokens by up to 80% versus Tinker on a DAPO-math GRPO job. Modal says Spindle is still in beta and positions it between its Dojo training library and raw cluster primitives.

    Why it matters: Modal's open-source Spindle server offers teams a Tinker-compatible alternative that trades per-token pricing for GPU time, with multi-LoRA sharing reported to cut costs by up to 80%.

  3. SemiAnalysis75

    The Huawei-DeepSeek Symbiosis: Ascend 950PR Benchmarks and DeepSeek's Ascend Kernel Support

    SemiAnalysis examines the Huawei-DeepSeek partnership, benchmarking an Ascend 950PR (28 cube cores, 128 GiB HBM) and reviewing DeepSeek's September 30 release of Ascend support for TileLang, DeepGEMM, DeepEP, TileKernels, FlashMLA and DeepSelect. The 950PR reaches 1.6384 TB/s peak HBM bandwidth and beats H100 bandwidth utilization at small message sizes, while its median achieved/attainable compute fraction is 41.5% versus H100's 46.8%; only TileLang is usable so far because the required CANN 9.2.0 was unreleased at writing. The authors argue the software stack is near the frontier even as Huawei hardware trails, and note that most optimizations target supernodes, with vLLM-Ascend still failing out of the box on x86 hosts.

  4. Google Developers Blog38

    Google Cloud's Project Chrysalis: A Governed, Self-Evolving AI Agent Architecture on Gemini and ADK

    Google Cloud researchers unveiled Project Chrysalis, a reference architecture for governed, self-evolving AI agents built on the Gemini platform and Agent Development Kit (ADK). In simulated Black Friday cache-stampede outages, it cut time-to-first-mitigation from 15–45 minutes to 1.8 seconds, reduced HTTP 5xx errors from 72.7%–80.0% to 0.0%, and dropped P99 latency from 519.8 ms to 18.2 ms. Every autonomous action is bounded by Model Armor, Cloud KMS, Binary Authorization, OPA Policy Controller and dual-custody human-in-the-loop sign-off.

  5. Google Developers Blog58

    Google Cloud API Gateway adds streaming support in Public Preview

    Google Cloud API Gateway now supports streaming in Public Preview, letting SSE, chunked HTTP, WebSocket and gRPC streaming endpoints sit behind the same gateway and policies as existing APIs. Streaming is enabled with a single --enable-streaming flag at gateway creation, requires no changes to the API spec, and keeps API key enforcement, JWT validation, service account auth and Cloud Logging and Monitoring working as before. Gateways also get a deterministic default hostname of GATEWAY_ID-PROJECT_NUMBER.REGION.gateway.dev, and streams can stay open for up to one hour.

October 21Wednesday
  2. Modal Blog58

    Modal releases Dojo, a library for production-grade LLM training

    Modal released Modal Dojo, a training library that provides building blocks, tuned default recipes and an observability dashboard for production-grade LLM post-training. It builds on open-source frameworks Miles and Slime, supports models from half a billion to multi-trillion parameters, and exposes an API and CLI that agents can drive to launch and debug runs. Modal says Axiom Bio used it to scale to 64k-context DeepSeek V4.1 Flash runs on 64 B300s, and Afterquery trains Kimi and DeepSeek at 256k context length.

  3. LiveKit Blog25

    How Acuity Health Uses LiveKit Voice Agents to Automate Patient Access

    Acuity Health rebuilt its patient-access voice agents on LiveKit after a low-code platform proved too rigid and opaque, and now 80% of calls are fully contained by AI agents. Agents run on LiveKit Inference with the Gemma 4 31B model, most bookings finish in under a minute, and 81% of booking calls end with a "thank you." LiveKit Simulations tests each change against patient profiles before release, and one prompt review cut roughly 24% of words while still passing the same scenarios.

  4. Google AI · DEV58

    Google Antigravity adds Claude Opus 5.5 and Sonnet 5.5 with mid-session model switching

    Anthropic's Claude Opus 5.5 and Claude Sonnet 5.5 are now officially available in Google Antigravity across Antigravity 2.0 Desktop, the CLI, and IDE extensions for VS Code, JetBrains, and Visual Studio. Opus 5.5 targets heavy reasoning and multi-file architectural refactoring, while Sonnet 5.5 serves as a faster, lower-cost option for day-to-day coding. Developers can switch models mid-session via the dropdown or /model claude-opus-5-5 in the CLI without losing context, terminal states, or workspace artifacts, and admins enable the models and set spend limits in Google Cloud Console under Gemini Enterprise.

  5. Google Developers Blog58

    Google Agents CLI 1.8.0 ships Extension System and first-party LangChain extension

    Google's Agents CLI now includes an Extension System, introduced in v1.5.0, that lets developers override or add commands without forking the tool, and v1.8.0 ships a first-party LangChain extension. The LangChain template scaffolds a project with no google-adk dependency that deploys to Cloud Run through the same code path as an ADK project, while six ADK-coupled commands such as playground, run and eval generate are replaced or cleanly refused. Extensions are distributed as directories in Git repositories with a lockfile pinning an exact commit SHA, and the system is marked experimental in v1.8.0.

October 20Tuesday
  1. Sakana AI Blog15

    Sakana AI Engineer Open House 2026: Company Outlines AI Sovereignty Strategy and Five Engineering Roles

    Sakana AI held its second Engineer Open House on October 5, 2026, covering five engineering roles including Applied Research Engineer, Forward Deployed Engineer, Software Engineer, Solution Engineer and Cybersecurity Engineer. CEO David Ha outlined an "AI sovereignty" strategy built on pre-training, post-training with Namazu and orchestration with Fugu, after expanding its portfolio to Sakana Chat, Marlin, Fugu and Translate in the past six months. The event also detailed work with Japanese financial institutions and defense projects, including research with the Ministry of Defense's Information Headquarters and the Defense Innovation Science and Technology Institute.

  2. Mistral AI82

    Mistral releases Voxtral 3 speech models and voice agent SDK

    Mistral released Voxtral 3, a family of speech-to-text and text-to-speech models plus a voice agent SDK built on LiveKit that streams listening, reasoning, tool calling and speaking in one pipeline. Voxtral Small Transcribe 3 covers 50+ languages with diarization and context biasing, Voxtral Mini TTS 3 supports 35+ languages and 50+ dialects, and the stack is available through Mistral's EU-hosted API or self-deployment. Pricing starts at $0.0035/min for batch transcription and $13 per 1M characters for TTS.

    Why it matters: Mistral's Voxtral 3 pairs multilingual speech models with a LiveKit-based voice agent SDK, giving teams a sovereign-deployable option for real-time voice agents.

  3. Zyphra Research58

    Zyphra Inference Wave 2 brings custom HIP MoE and sparse-attention kernels to MI355X serving

    Zyphra released the second wave of its Zyphra Optimization Engine (ZOE) for AMD Instinct MI355X GPUs, adding four custom HIP kernels for MoE and sparse attention into its production forks of vLLM and SGLang. 40x). 3 checkpoint stays within 3 points of FP8 on four reasoning benchmarks but decodes faster only at high concurrency.

  4. Waymo Blog58

    Waymo to begin laying groundwork for autonomous ride-hailing in Madrid region, first riders targeted for 2027

    Waymo announced it is laying the groundwork for its fully autonomous ride-hailing service in Spain, starting with the Region of Madrid, with the aim of welcoming its first riders in 2027. Electric Jaguar I-PACE vehicles will appear on regional streets in the coming days, beginning with manual driving before technical validation by locally trained autonomous specialists. Waymo cites over 20 million completed trips and data from more than 400 million fully autonomous kilometres showing 95% fewer crashes involving serious injury or worse than human drivers in cities where it operates.

  5. Google Developers Blog60

    Google 将 TorchTitan 接入 TorchTPU，并发布 TPU 模型优化手册

    Google released a TPU fork of Meta's TorchTitan that runs PyTorch training on Cloud TPUs through TorchTPU's native torch.device("tpu") backend, with TPU-native kernels and a step-by-step optimization playbook. On TPU v7x (Ironwood), Llama 3 405B holds about 57% MFU from 128 to 1,024 chips, Qwen3 235B reaches 32.42% MFU on one 1,024-chip slice, and FLUX.1 reaches 19.4% MFU on 1,024 chips at 93% weak-scaling efficiency. The fork is at https://github.com/google-pytorch/torchtitan/, and a Transformers modeling backend plus an agent onboarding skill extend the same trainer to other checkpoints.

    Why it matters: Google 与 Meta 合作把 TorchTitan 移植到 TPU，并给出从 FLUX.1 到 Qwen3 235B 的逐级调优路径与 MFU 数字，可作为 PyTorch 团队评估 TPU 训练可行性的参考。

October 19Monday
  1. Modal Blog58

    Modal benchmarks NVIDIA Vera CPU: up to 2x sandbox throughput under load

    Modal benchmarked NVIDIA's new Arm-compatible Vera CPU server and measured up to double the task throughput of an equivalent x86 GNR machine, with per-task latency equal to or sometimes lower. The results come from Agentless Agent Bench, 50 Terminal-Bench oracle solutions and verifiers stripped of network calls and run in concurrent Docker containers. Modal attributes the gain to Vera holding its pace as concurrent containers accumulate, which it links to high memory bandwidth, and says supporting Arm would require changes across its infrastructure and image build system.

  2. Modal Blog38

    How Legora runs 2,000 parallel agent evals on Modal Sandboxes

    Legora now runs each of its legal agent evals inside a separate Modal Sandbox, booting its full application pinned to a specific commit so every run reflects the exact product version. A typical experiment is 100 cases, 4 harness variants, and 5 trials each — 2,000 runs in parallel — with Modal Volumes snapshotting document, database, and permission state. Legora also moved from gVisor to Modal VM Sandboxes, removing its Docker networking and FUSE workarounds, and cut agent cost by 30% via smarter compaction and custom tool search.

October 16Friday
  1. Modal Blog36

    Modal explains how it built a fast filesystem stack for VM Sandboxes

    Modal detailed how it evolved its filesystem stack to back VM Sandboxes, routing guest requests through virtio-fs and virtiofsd onto its ImageFS lazy-loading filesystem. The first version booted fast but ran slow: recursive chown went from four seconds to over a minute, and pnpm and bun installs were about three times slower. Microbenchmarks showed chown at 9.87 s over virtio-fs versus 2.31 s on gVisor and 1.58 s on ImageFS.

  2. Replit Blog40

    Replit Brings Your Own Cloud (BYOC) to Its Platform

    Replit is introducing Bring Your Own Cloud (BYOC) options that let enterprises run the Replit platform and published apps inside their own Microsoft Azure, Google Cloud, or Amazon Web Services tenant. Full BYOC, in private preview, deploys Replit's cells and agent harness in the customer's cloud, while BYOC for Published Apps, in early access, hosts published projects there with Replit managing the platform. Image Exports instead give teams a Docker image, Terraform file, and SBOM to self-host Replit-built apps.

  3. OpenClaw Blog22

    How PJ Helps Small Businesses Adopt and Own OpenClaw

    PJ, an engineer-turned-pest-control owner in Ocala, Florida, now consults for small businesses—roofing, pool and automotive contractors among them—setting up OpenClaw and teaching them to run it themselves. His clients want a tool they own outright rather than feeding company data into rented software, and one client, 66-year-old admin Donna, uses her OpenClaw daily.

  4. LiveKit Blog38

    How to Build a Production-Ready AI Voice Agent

    LiveKit's guide distills lessons from solution architects and forward deployed engineers on taking AI voice agents from demo to production, organized around three questions: whether the agent is smart and natural enough, whether its performance is observable, and whether it scales reliably and cost-efficiently. It recommends measuring the first turn separately, since one company found roughly 3 out of 4 calls ended at the greeting step, and notes clients moving from GPT-4.1 to Gemma 4 31B on LiveKit Inference saved 300–500ms at p50.

October 15Thursday
  1. Anthropic · Claude.dev62

    How the Claude Code team does code review

    The Claude Code team describes its layered code review flow, in which most reading is done by machines and human attention is reserved for risk-based approval. From June to September, merged pull requests to the Claude Code codebase more than doubled while the share that had to be reverted fell by more than 40 percent. The process stacks verification before the PR, required CI, Claude review, security review, risk-tiered human approval and a merge queue with flags and fast rollback, and the team turned past incidents into a daily eval set for the reviewer.

    Why it matters: Anthropic's Claude Code team details how it moved most code reading to automated layers while merged PRs doubled and reverts fell over 40 percent, offering a transferable review model for agent-heavy teams.

  2. Modal Blog34

    Augment Your LiveKit Agent with Modal GPUs

    Modal explains how to run LiveKit Agents with open-weights or custom models on its serverless GPUs, using BitHuman's open-weights Expression Video Avatar voiced by Qwen3-TTS to clone Charles Frye. The setup exposes the avatar container's /launch endpoint via a Modal Server and uses Sticky Sessions to keep WebSocket clients on the same container, with GPU snapshots hiding cold starts.

  3. Modal Blog78

    Meta's Muse Spark Open weights released and available on Modal Auto Endpoints

    Meta released the weights for Muse Spark Open, and Modal now serves the model through its Auto Endpoints with per-token billing. The model targets agentic coding and long-horizon tasks, with native multimodal understanding, a 1M-token context window, multi-step planning, sequential tool use and failure recovery. Meta reports 76.4 on Terminal-Bench 2.1 with the Terminus 2 agent and 60 on DeepSWE 1.1 using its preferred harness at xhigh reasoning effort.

    Why it matters: Meta's open-weight Muse Spark Open is now served on Modal, giving teams a per-token way to try an agentic coding model with a 1M-token context window.

October 14Wednesday
  1. GitHub Blog · AI & ML38

    GitHub's secret detection model: finding more secrets with less noise

    GitHub trained a fine-tuned ModernBERT classifier to replace its previous LLM-based pipeline for generic secret detection, improving owner-weighted alert precision by 11.5% and cutting median post-push scan execution time by 86%. The two-stage pipeline uses regex candidate generation followed by context-aware classification, which also lowered serving cost by more than 98% versus remote LLM calls. The efficiency gain made it practical to extend generic secret detection to push protection, returning results before changes are pushed.

  2. Modal Blog71

    Modal crosses $1B annualized run rate and raises $750M Series D at $15.75B valuation

    Modal crossed $1B in annualized run rate and raised a $750M Series D led by Accel at a $15.75B post-money valuation, with all existing investors participating. The company says usage has tripled since its Series C five months ago and grown more than 15x year-over-year, and that its Sandboxes product alone is at a $300M run rate with close to 3 billion sandboxes spun up. Modal also describes rebuilding its own file system, container runtime, scheduler and networking on a capacity pool spanning 20+ clouds, and points to recent launches in multi-node training, Modal servers and generally available VMs for Sandboxes.

    Why it matters: Modal's post explains why AI workloads pushed it to rebuild its own file system, container runtime and scheduler across 20+ clouds, a concrete view of how the AI infrastructure stack is diverging from the cloud stack.

October 13Tuesday
  1. LiveKit Blog18

    Troubleshooting LiveKit Egress: Common Errors and Fixes

    LiveKit published a troubleshooting guide for Egress, covering common errors such as the 412 "start signal not received" from templates that never call START_RECORDING, 404 participant or track not found within 30 seconds, and 400 upload failures from invalid S3/GCP/Azure credentials. It also addresses layout, webhook, and self-hosted Egress issues, and notes the guidance applies to both the current StartEgress API and the deprecated source-specific APIs.

  2. Modal Blog58

    Modal tests mixed-precision, mixed-accelerator RL samplers on Qwen3.6-35B-A3B

    Modal trained Qwen3.6-35B-A3B on agentic coding with samplers spread across six accelerator types (B200, B300, H100, H200, A100, RTX PRO 6000) at NVFP4, FP8 and BF16 precision, feeding one BF16 trainer on B200. Plain GRPO collapsed on both mixed and homogeneous samplers, but adding masked importance sampling plus score centering, with or without top-p mask replay, trained stably to the end and reached SWE-bench Pro V2 pass@4 on par with homogeneous samplers, at roughly 15% lower sampler cost per update. The mixed samplers ran with an average sampler-trainer KL 4.7x that of the homogeneous setup, and the authors argue stability depends on keeping the mismatch bounded rather than low.