Google Research released TimesFM-3, a 330 million parameter time series foundation model that performs zero-shot multivariate forecasting in a single forward pass, accepting multiple targets and covar…
MarkTechPost published a benchmark on August 30, 2026, comparing time-to-first-token (TTFT) latency across inference APIs for voice and realtime agents, covering LLM, speech-to-text, text-to-speech, a…
Google Cloud AI Research, with Washington University in St. Louis and UNC Chapel Hill, released EnvHarness, an Apache-2.0 layer that turns static agent benchmarks into adaptive training environments b…
Anthropic has opened a research preview of the Model Hardware Standard (MHS), a shared driver specification that lets AI agents discover and safely operate physical devices, reducing instrument integr…
MirroS released Code-as-World, a paradigm that represents physical scenes as executable MuJoCo programs recovered from real videos via an agentic loop, and open-sourced it under Apache 2.0 with two ch…
NVIDIA's Earth2Studio tutorial demonstrates a custom batched ensemble weather forecasting workflow, using the FCN prognostic model and GFS initial conditions to generate 8-member ensembles with wind-p…
Google released Gemini Omni 1.1 Flash, a production update to its native multimodal video generation and editing model, enabling scene extension up to 40 seconds, first/last frame control, and 4K upsc…
Hugging Face's Pollen Robotics team opened pre-orders on August 27, 2026, for Microduck, a $399 open-source 25 cm bipedal robot whose movements are neural policies trained in a physics simulator and e…
Z.ai released GLM-5.3-Flash, a 320B-parameter multimodal MoE model with 18B active parameters, and Alibaba's Qwen team released Qwen3.8-Flash-Next, a 125B model with 6B active parameters, within a day…
Vercel has open-sourced vgpu, a TypeScript WebGPU library for AI agent shaders, under an MIT license and published to npm at v0.3.1. The library treats .wgsl files as importable modules, runs the same…
Google AI released Gemini 3.5 Transcribe, a speech-to-text model with two endpoints: a streaming endpoint offering sub-second transcription without speaker diarization or word timestamps, and a batch …
Cohere released Parse (parse-v5.0), a 2.3B-parameter vision language model that converts PDFs, slides, and images into Markdown with HTML tables, bounding boxes, and image descriptions. The model, ava…
A new open-source benchmark from ComputeSDK measuring Time to Interactive across five agent sandbox platforms found Daytona fastest at a 0.27s median but with only a 37% success rate, while Vercel San…
Anthropic's claude-protein-binder-design dataset, containing 1,440 AI-designed miniprotein binders tested against 16 targets, was used in a tutorial to evaluate AI protein design performance, finding …
Z.ai released GLM-5.3-Flash, a 320B-total-parameter mixture-of-experts model with 18B active parameters, a 1,048,576-token context window, and native image and video input, under an MIT license with w…
Alibaba's Qwen team released Qwen3.8-Flash-Next, an open-weight multimodal Mixture-of-Experts model with 125B backbone parameters and 6B active per token, previewing the Qwen4 architecture. The model …
A new analysis by an industry observer argues that agentic coding will not replace junior engineers because three of four necessary conditions are unmet, citing METR's finding that frontier AI systems…
IBM released Granite 4.2, a family of open reasoning language models in 3B, 8B, and 30B parameter sizes, available under Apache 2.0. The models are decoder-only dense transformers pre-trained on rough…
Liquid AI released Pipette, an open-source benchmarking platform for foundation models on edge devices, developed with Artificial Analysis as an independent methodology validator. Pipette measures on-…
Perplexity has released Portable Computer, a local-first build of its agentic Computer platform that runs on NVIDIA DGX Spark and RTX GPUs with 24 GB of VRAM, packaging the local model, inference engi…