Skip to content
#Open-source ecosystem
October 24Saturday
  1. Epoch AI · Gradient Updates42

    Epoch Brief: AI Work Automation Test, SpireClimb Benchmark, and Gemini 4 Argon's Frontier Push

    Epoch AI tested six leading models on 11 realistic Epoch work tasks and found none could reliably meet its quality bar, with Claude Fable 5.1 and GPT-6 Astra roughly tied for the lead. Its new SpireClimb benchmark shows frontier models went from struggling at Slay the Spire to winning over 90% of base-difficulty runs in six months, with GPT-6 Astra at 98%. Gemini 4 Argon scored 166 on the Epoch Capabilities Index, tying Claude Fable 5.1 and trailing Claude Opus 5.5 and GPT-6 Astra at 169.

  2. 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.

October 23Friday
  1. 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. 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.

  2. Anthropic Newsroom71

    Anthropic donates Agent Skills standard to the Agentic AI Foundation

    Anthropic is donating Agent Skills, the open standard for giving AI agents specialized knowledge and workflows, to the Agentic AI Foundation (AAIF) under the Linux Foundation. More than 70 agent platforms including ChatGPT, Cursor, GitHub Copilot and VS Code have integrated the standard since it was published in December 2025, with 1 million skills published and nearly 280 million installed. The donation covers the Agent Skills name and logo, the GitHub organization and the agentskills.io domain, with engineers from Anthropic, Google, OpenAI and other companies maintaining it as individuals under Lead Maintainers David Soria Parra and Mahesh Murag.

    Why it matters: Anthropic hands its Agent Skills standard to the Linux Foundation's AAIF, the same neutral home as MCP, giving the community a formal say over the spec's roadmap.

October 21Wednesday
  2. Mistral AI48

    Mistral and NVIDIA Partner on AI for Engineering Workflows

    Mistral and NVIDIA are collaborating to build AI-native engineering workflows that pair Mistral Large 4 with NVIDIA CUDA-X libraries and Warp, aiming to speed simulation and enable closed-loop design. The partnership will also produce public benchmarks and open datasets for physics models and agentic AI in CAE and EDA, plus integration with NVIDIA Omniverse for 3D scene generation and SimReady assets. Mistral Large 4, known as Le Chonk, has roughly 1 trillion total parameters and 49 billion active parameters and was trained on 3,800 NVIDIA Grace Blackwell GPUs.

  3. OpenClaw Blog22

    OpenClaw Launches Focus Weeks, Starting With Control UI

    OpenClaw launched Focus Weeks, a program that each week targets one area of its roadmap with a Maintainer on point and invites the community to contribute code, bug reports, testing, documentation and accessibility feedback. The first Focus Week runs October 19–25, 2026 on the Control UI, with Vyctor Brzezowski as Maintainer. A select group of rotating Clawtributors also gets temporary guest access to a shared OpenClaw Team environment to test its multiplayer collaboration features.

October 18Sunday
  1. Recode China AI40

    China Proposes AI Export Controls, DeepSeek Bets on Robotics, Infinigence Raises $334M

    China's commerce and science ministries proposed adding AI models, agents, chip designs and computing clusters to the technology export-control catalogue, with API, app and SaaS access excluded from the draft rules. DeepSeek reportedly plans to lead a seed round in robotics-AI startup Winverse AI, while Infinigence AI closed a RMB2.3 billion Series C. DeepSeek also released V4.1-Flash-1017 with MIT-licensed weights and a 1M-token context window.

October 15Thursday
  1. 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.

  2. Meta AI Research Blog88

    Meta releases Muse Spark Open, an 816B open-weight multimodal model that runs on two Mac Studios

    Meta released the weights for Muse Spark Open, an open-weight natively multimodal model with 816B total parameters (42B active) and a 1M-token context window, positioned for coding and agentic work. It is the first model in its class to run on a desktop using two Apple M5 Ultra Mac Studios with the MLX framework, and it is available through API partners including Baseten, Fireworks AI, Groq, Modal, Nebius, OpenRouter and Together AI, with weights on Hugging Face, Amazon SageMaker AI, Google Cloud Model Garden and Microsoft Foundry Models. Meta says the model ships with auto-tunable Helion GPU kernels in SGLang, delivering 1.03-3.95x kernel-level speedups on NVIDIA and AMD GPUs and a 3% end-to-end speedup on AMD MI350.

    Why it matters: Meta releases open weights for an 816B-parameter multimodal model that runs on two Mac Studios, giving developers a local option for long-horizon coding and agentic tasks.

October 13Tuesday
  1. Recode China AI34

    DeepSeek Engineer's Viral Blog on AI Risk and Anthropic Draws Global Media Attention

    A blog by DeepSeek machine learning and kernel engineer Liu Shengyu went viral after he questioned whether frontier AI can now optimize its own low-level GPU code and compared Anthropic CEO Dario Amodei to Adolf Hitler. The post drew coverage from outlets including South China Morning Post and Business Insider, amplifying debate over who should control advanced AI.

  2. 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.

October 8Thursday
  1. Goodfire Research71

    Goodfire trains and deploys probe-based cyber misuse monitors on Kimi K3 and GLM 5.3

    3 and deployed them on a production inference stack, where a cheap probe filters turns and only suspicious exchanges reach an LLM judge. 5% benign-session interruption rate, costs $65 per million turns without judge reasoning and $185 with it, and adds no measurable throughput loss inside SGLang.

    Why it matters: Goodfire shows a probe-plus-judge cascade can monitor agentic cyber misuse at roughly 50x lower cost with no measurable throughput loss, a concrete recipe for serving open models safely.

October 1Thursday
  1. Modal Blog58

    Modal Clusters 正式可用，通过 @modal.clustered 一行代码获取多节点 RDMA 集群

    Modal announced that Modal Clusters are generally available to all workspaces, letting users request multi-node clusters through the single decorator @modal.clustered. The clusters provide InfiniBand RDMA communication at up to 6.4 Tbps, automatically configured for PyTorch and NCCL, with clusters starting within seconds and billed by the second. Modal also describes a new gang scheduler that places N nodes at once and RDMA support built into the gVisor runtime, with Decagon, 1x and Runway cited as users for trillion-parameter fine-tuning, world model pre-training and multi-node video inference.

September 28Monday
September 25Friday
  1. OpenClaw Blog74

    Microsoft Autopilot is built on OpenClaw, with contributions flowing both ways

    Microsoft introduced Autopilot, a persistent personal agent built on OpenClaw, as part of its new Copilot experience, with a private preview expanding to first customers at the end of this month. Autopilot is the new name for Microsoft Scout, announced earlier this year, and Omar Shahine's team says it is working with OpenClaw and the OpenClaw Foundation on an enterprise-grade runtime. OpenClaw highlights upstream contributions from Microsoft engineers, including policy conformance checks, a native Windows companion, and reliability fixes for background jobs, restarts and duplicate replies.

    Why it matters: Microsoft's Autopilot is built on OpenClaw, and the upstream contributions listed here show how a shared agent runtime improves across platforms.

September 22Tuesday
  1. OpenClaw Blog60

    OpenClaw adds opt-in decision models and a shared Plugin SDK interface

    OpenClaw introduced support for a separately configured decision model that its core and plugins can call through api.runtime.decisions.evaluate in the Plugin SDK, returning a choice, score or probability instead of a chat reply. The feature is opt-in and does not replace the chat model, and the provider packages are still awaiting a supporting release. The TypeSafe adapter supports hosted Jev and a local System One server such as Jared Palmer's Kev, and the evaluation tool is moving into OpenClaw core as decision_evaluate.

    Why it matters: OpenClaw adds an opt-in decision model layer that returns typed answers from application code, with a shared Plugin SDK interface so plugins can use it without per-provider integrations.

September 21Monday
  1. LiveKit Blog38

    Comparing Open Source Frameworks for Realtime Voice and Video AI Agents: LiveKit Agents vs Pipecat

    This guide compares LiveKit Agents and Pipecat, the two open source frameworks it says matter most for building production voice and video AI agents, across architecture, adoption, developer experience and production readiness. LiveKit Agents runs on LiveKit's open source WebRTC and SIP infrastructure and wires the pipeline for you, while Pipecat assumes no transport and has developers assemble the pipeline stage by stage. It also covers where TEN Framework, Dograh, Vision Agents and Vocode fit in.

September 16Wednesday
  1. Mistral AI55

    Mistral partners with Mozilla to power Firefox Smart Window in France and North America

    Mistral announced a partnership with Mozilla under which its models now power Firefox Smart Window (beta), Mozilla's AI browsing assistant, for users in France and North America, with the UK and Germany expected later this year. Mistral says Smart Window conversations are not saved on Mozilla's servers by default and that partners agree to zero data retention, and it frames the deal as bringing open-weight, regionally fine-tuned models to consumer browsing.

September 14Monday
  1. Modal Blog40

    Modal Product Updates: Sandbox Sidecars, Kimi K3 and GLM 5.3 Support, Refreshed Dashboard

    Modal added day-zero support for four open-weight models — Kimi K3, Qwen3.8-2.4T-A95B, GLM-5.3, and GLM-5.3-Flash — on its Auto Endpoints, with Kimi K3 running at 460 tokens per second on agentic workloads via a custom DFlash speculator. Sandbox Sidecars entered public alpha, VM Sandboxes reached public beta, and the broad-region pricing multiplier dropped from 1.5x to 1.15x. The dashboard was refreshed with a light mode beta, and Python SDK 1.5.5 plus JavaScript and Go SDK 0.10.0 shipped.

August 18Tuesday
  1. LiveKit Blog40

    LiveKit: multi-participant robotics beyond 1:1 connections

    LiveKit is extending its SFU architecture beyond single peer-to-peer links so robot deployments can run multi-participant sessions, where a Room lets each camera, model, operator, or service join as its own participant. LiveKit describes three architectures: human-in-the-loop data recording with robot, policy, operator, and recorder in one session; VLM-guided skill orchestration where a vision-language model selects skills at roughly 1 Hz while separate policy participants control the robot at tens of hertz; and policy escalation, where a reward model at 30 Hz or higher flags stuck or unsafe states and the robot revokes authority from the specialist policy.

August 13Thursday
  1. MiniMax Blog71

    MiniMax releases Music 3.0, an open-weights music generation model for full songs up to five minutes

    MiniMax introduced Music 3.0, an open-weights music generation model that composes, arranges and produces a complete song of up to five minutes from a creative concept and optional lyrics in a single generation. The pipeline pairs an 8B Global LLM initialized from Qwen3.5-8B with a 0.6B Local LLM, fuses their continuous hidden states to condition a 2.4B flow-matching module and a 123M Flow-VAE, and uses eight-layer RVQ plus Structured Captions that describe emotion, instrumentation and vocal delivery over time. MiniMax says this yields more accurate interpretation of creative intent, fuller arrangements and clearer, more natural vocals.

    Why it matters: MiniMax's architecture breakdown shows how an 8B global LLM paired with a 0.6B local model and flow-matching stack targets the long-song coherence problem.

August 4Tuesday
  1. Modal Blog34

    Modal moves serverless Functions to a regional I/O plane, cutting p50 latency by ~80ms

    Modal has migrated all Functions to a new geographically distributed I/O plane, dropping end-to-end Function Call latency by roughly 80ms at p50. Previously every Function's inputs and outputs routed through us-east I/O servers, adding up to a few 100ms per call; the new plane is available in ≥4 regions and selectable via the routing_region flag. The server layer was rewritten in Go, with non-critical work moved off the hot path.

July 27Monday
  1. Modal Blog75

    Moonshot's Kimi K3 now available on Modal with day-zero vLLM support

    Modal made Moonshot's Kimi K3 available on its platform on release day, offering the 2.8 trillion parameter multimodal model with a 1M token context window through a token-priced Shared API and dedicated Auto Endpoints. Modal says a custom DFlash speculator tuned to K3's architecture raises interactivity from 100 to 460 tokens per second and throughput from 800k to 1.5 million TPM per GPU on agentic workloads. The model uses a mixture-of-experts design with 16 of 896 experts active per token, and Moonshot contributed a new prefix-caching implementation to vLLM ahead of the release.

    Why it matters: Modal details how it brought Moonshot's 2.8T-parameter Kimi K3 to its platform on release day, including a custom DFlash speculator that lifts interactivity from 100 to 460 tokens per second.

June 23Tuesday
  1. Modal Blog60

    Modal launches Auto Endpoints for self-serve, OpenAI API-compatible LLM inference

    Modal launched Auto Endpoints, a self-serve service that deploys production-grade LLM inference as an OpenAI API-compatible endpoint backed by a Modal App the user can see and control. Each supported model ships with a starting deployment tuned from Modal's own benchmarking, using engines like SGLang and the DFlash speculative decoding drafter from Z Lab, and users can inspect GPU choices, engine flags and inference metrics such as TTFT, ITL and speculative decoding acceptance length. Modal also released Modal Servers from beta for regionalized routing with 5ms overhead, and said it generates endpoint deployments with an internal agentic system still monitored by human engineers.

    Why it matters: Modal's Auto Endpoints expose the serving stack, engine metrics and benchmark recipes that managed inference providers keep hidden, giving teams a self-serve path to owning open-model inference.

June 19Friday
  1. Modal Blog60

    Modal releases DFlash speculators for Qwen 3.5 and 3.6, claiming 5-20% extra speedup

    Modal released a set of DFlash draft models for the Qwen series on Hugging Face, including Qwen 3.6 35B-A3B-DFlash and Qwen 3.5 variants from 4B to 122B-A10B, plus a DFlash speculator for Qwen 3.5 397B-A17B built with SGLang. The company says the new drafters add 5-20% speedup over existing DFlash baselines and can run Qwen 3.5 122B-A10B at over 1000 tok/s at concurrency 1 on a B200 node, versus 250 tok/s without speculation. The post argues speculative decoding delivers 2x-3x gains where kernel optimization yields small percentage points, and presents a roofline model for estimating speedup from acceptance length.

    Why it matters: Modal explains why it bets on speculative decoding and releases DFlash draft models for the Qwen series, with measured speedups and a roofline model for estimating gains.

June 11Thursday
  1. Modal Blog34

    Modal Makes FlashAttention-4 Faster for LLM Inference

    Modal contributed optimizations to the FlashAttention-4 kernel to speed up decode-heavy LLM inference workloads, where memory bandwidth is the bottleneck. The changes include porting the "split KV" parallelism technique to FA4 and replacing TMA cp.async.bulk loads with cp.async for irregular memory accesses. An FP8 input PR (merged April 17, 2026) delivers up to 1.16x throughput over the bf16 baseline and enables smaller KV caches.

June 1Monday
  1. MiniMax Blog87

    MiniMax M3 released with 1M context, frontier coding and native multimodality

    MiniMax released M3, an open-weight model it says is the first to combine frontier coding and agentic performance, a 1M-token context window, and native image and video input. M3 uses MiniMax's new MSA sparse attention architecture, which the company reports cuts per-token compute at 1M context to 1/20 of the previous generation and speeds prefill by more than 9x and decoding by more than 15x. It scores 59.0% on SWE-Bench Pro, 66.0% on Terminal-Bench 2.1 and 74.2% on MCP Atlas, and is available through MiniMax Code, the Token Plan and the API, with weights and a technical report promised within 10 days.

    Why it matters: MiniMax M3 combines a 1M-token context, frontier coding and agentic scores, and native image and video input in one open-weight model, with MSA cutting per-token compute at long context.

May 4Monday
February 14Saturday
  1. Charlie O'Neill38

    The Michael Scott Paper Company of AI: Why Open-Source Specialisation Beats Frontier Labs on the Intelligence Ceiling

    Opus 4.6 and GPT-5.3-Codex both shipped last week, but the decisive contest is not raw capability. The argument holds that most economically valuable work sits in the specialisation layer, where a sub-80B model fine-tuned on a client's own data outperforms GPT-5.2 on that client's workflow — faster, cheaper, and without sending proprietary data to a third party. Open-source needs only to match the intelligence ceiling, not the frontier, and the labs' capital-dependency clock is the weaker position.

February 11Wednesday
  1. Modal Blog78

    Modal 上线 GLM-5 免费端点并发布自部署与前端接入指南

    Modal 推出 GLM-5 的免费 OpenAI API 兼容端点，可用至 4 月底，限单个并发请求，同时发布在单节点 8 张 B200 上用 SGLang 部署的可复现代码。

    Why it matters: Modal 为 Z.ai 的 GLM-5 提供免费 OpenAI 兼容端点，并给出在 OpenCode、OpenClaw、Claude Code 和 Vercel AI SDK 中接入的具体配置。

December 2, 2025Tuesday
  1. Modal Blog42

    Modal adds Day 0 support for Mistral 3 with GPU snapshotting for 10x faster cold starts

    Modal now offers Day 0 support for Mistral 3, letting developers deploy the open model family on serverless GPUs without managing infrastructure. On the 3B Ministral 3 model, Modal's GPU snapshotting feature cut median cold start time from about 118s to about 12s, nearly a 10x reduction. Ministral 3 comes in 3B, 8B, and 14B sizes and performs competitively with the Qwen 3-VL series on benchmarks.

November 18, 2025Tuesday
  1. Modal Blog36

    Modal Blog: Host Overhead Is Killing Your Inference Efficiency

    Modal's blog explains how host overhead—gaps where the GPU sits idle waiting for the CPU to prepare work—shows up as low GPU kernel utilization in AI inference. At 50% host overhead, GPU inference costs are 2x what they could be. The post details fixes used in production open source inference engines, including constructing tensors on-device instead of transferring them and reducing unnecessary CUDA kernel launches.

November 4, 2025Tuesday
  1. Modal Blog58

    Modal, Pipecat and open models achieve one-second voice-to-voice latency

    Modal's Ben Shababo walks through building a conversational voice bot on Modal with the open source Pipecat framework and open weights models, reaching a median voice-to-voice latency of about one second when the client and containers are geographically close. The pipeline uses NVIDIA parakeet-tdt-0.6b-v3 for STT, Qwen3-4B-Instruct-2507 served with vLLM, KokoroTTS for speech output, and ChromaDB with all-minilm-l6-v2 for RAG over Modal's docs. Modal Tunnels bypass the input plane for low-latency WebSocket and HTTP links to the STT, TTS and vLLM services, while a FunctionCall tied to each conversation session recovers autoscaling, and pinning services to nearby regions cuts network latency.

October 30, 2025Thursday
  1. Modal Blog38

    Modal and Datalab: Deploy Marker and Surya Document Intelligence in Under 5 Minutes

    Modal has partnered with Datalab to let developers deploy the Marker document-parsing model and Surya OCR toolkit on Modal GPUs in under five minutes. Marker is a sub-billion-parameter model supporting over 90 languages, and running it on Modal reaches about 2.2 pages per second per container, roughly 10x the 0.22 pages per second on an M4 Mac using Apple MPS. Datalab's tools stay free for research, personal use, and startups under $2M in funding or revenue, with commercial licensing available.

July 23, 2025Wednesday
  1. Modal Blog52

    Modal benchmarks open ASR models for batch transcription at 100x speed and cost gains

    Modal implemented a batch transcription service on its platform using NVIDIA's parakeet-tdt-0.6b-v2 and canary-1b-flash, reporting over 100x faster or 100x cheaper end-to-end performance than a popular proprietary ASR API while matching its error rate on roughly a week of ESB benchmark audio. The team attributes the gains to shuffling files across request batches, sorting audio by duration within GPU batches, parallel downloads, and empirically chosen GPU and worker configurations. The post walks through the architecture and optimization steps so others can build similar services.

July 11, 2025Friday
  1. Modal Blog31

    Jamsocket is joining Modal

    Jamsocket, the backend platform behind the open-source orchestrator Plane, Y-Sweet and ForeverVM, is joining Modal, with co-founders Paul Butler and Taylor Baldwin joining the Modal team. Jamsocket products will keep operating normally while core functionality is migrated to Modal, which the founders say lets AI-centric customers deploy GPU inference, code sandboxes and real-time backends from one platform.

June 9, 2025Monday
  1. Modal Blog40

    Modal 1.0 Released: More Stable Python SDK with Explicit Configuration

    Modal has launched version 1.0 of its Python SDK client, a milestone the company says brings a new level of maturity and stability to its serverless cloud compute platform, with far fewer breaking changes expected going forward. The release removes "magic" behaviors such as automatic mounting of local packages, splits core functionality across separate decorators like the new @modal.concurrent, and consolidates local file handling under the modal.Image class. Modal also plans SDKs for other languages and is exploring how the platform should serve AI-driven agent workflows.