With $23M in funding, Luminal is the latest startup to put Nvidia's CUDA moat in its crosshairs - SiliconANGLE

Swedish investor Norrsken and Paris-based VC Singular co-led the round

UFORCE's NEMESIS drones have completed 300,000 combat flights, and its MAGURA naval vessels have contributed to the destruction of Russian warships.

Iusful raccoglie 2 milioni di euro per una piattaforma LegalTech che unisce AI e avvocati. La startup italiana vuole innovare l'assistenza legale delle PMI.

Telltale Bot tells website owners whether a visit comes from a person, from an AI agent acting for a person, or from a bot. It identifies; it never blocks.

Mubit sits between your agent and Neon AI Gateway, learns from every call, and moves routine steps to a small model tuned on your own traffic. In the example workload, that cut gpt-5.5 tokens by 73%.

Nectar's new AI agent automates influencer marketing, helping brands discover creators, manage outreach and partnerships, and track campaign results.

A Blog post by IBM Granite on Hugging Face

Meet Edge0-145B, an open-source 145B model for local coding, reasoning, and agent tasks, with no cloud-model token fees for local inference.

Supervised fine-tuning (SFT) on expert trajectories is a standard paradigm for training LLM agents. However, it suffers from a well-known training-inference distribution mismatch, calling for target-aware data synthesis strategies that ground the training data in the target model's native state distribution. To bridge this gap, emerging approaches incorporate the expert either to guide mixed rollouts alongside the target model, or to provide dense step-wise supervision. Nonetheless, existing methods apply expert assistance blind to the target model's inherent behavior: they either disrupt the rollout using sub-optimal random intervention heuristics, or they enforce indiscriminate uniform supervision that unnecessarily overwrites the model's already-correct actions. We introduce JudgeMix, a judge-guided data synthesis framework that dynamically interleaves the target model and expert based on step-wise semantic evaluations. JudgeMix queries expert guidance precisely when the target model struggles, ensuring high trajectory success rates while keeping the state distribution aligned with the target model. It additionally uses the judge evaluation to reweight expert supervision on different steps, preserving model's inherent valid behavior while correcting its errors. On two agentic tool-use benchmarks, AppWorld and AutomationBench, JudgeMix demonstrates consistent superior performance to existing methods, with improvements up to 10 pass@1 points.

Looped Transformers increase effective depth by repeatedly applying a shared block, offering a parameter-efficient approach to language modeling. However, most existing works directly reuse each loop's output as the next's input, which can lead to hidden-state scale growth and perturbation amplification as recurrence deepens. We propose Rectified Looped Transformer (RLT), which redesigns feedback between loops using an interpolation schedule inspired by flow-based modeling. At each loop, RLT forms the next input by linearly interpolating between the current state and its prediction. Such interpolation distributes state changes across loops, encouraging a smoother trajectory from the initial input to the final output. Theoretical analysis proves that, under appropriate assumptions, RLT keeps state magnitudes uniformly bounded across loops and reduces the worst-case perturbation amplification. We evaluate RLT on language modeling corpora and downstream benchmarks. At matched training and inference depths, RLT consistently achieves lower perplexity than a standard looped Transformer, while exhibiting greater robustness to input perturbations and variance in inference depth. Consistent with the theoretical results, latent trajectory analysis shows that RLT maintains more stable hidden-state norms across loops, which also facilitates extrapolation-based decoding strategies. Notably, RLT achieves these gains without requiring additional trainable parameters or extra recurrent block evaluations, thus providing a simple yet more stable feedback mechanism for looped Transformers.

AI agent tooling provider LangChain secures $300M investment - SiliconANGLE

Day 1 at #SGLangSummit going strong! Just now on the Main Stage:
• Ying Sheng (@ying11231) announces SGLang v1.0: faster inference, better memory use, & flexible scaling for prod
• Banghua Zhu (@BanghuaZ) introduces Miles v0.2: Production RL across more hardware and workloads

Accel led the round for SE3 which creates software for both autonomous drones and to command swarms of drones across a battlefield.

This is the investment firm's largest fund to date.

Portare l’automazione nelle macchine agricole è l’obiettivo di Anostra, startup selezionata per la quarta edizione di FoodSeed.

L’azienda romana Translated presenta un nuovo modello linguistico basato su 45 lingue e culture. Grazie al supercomputer italiano Leonardo, il software da 0,6 …

WebGPU-native LLM inference for the browser. Contribute to erdemozkan/kiln development by creating an account on GitHub.

Rather than giving consumers yet another photo-editing tool, Phota Labs aims to use advances in AI to automate the editing process so it’s easier for anyone to turn everyday moments into better photos. 

The model is available through Qianfan and free on OpenRouter. Baidu pairs it with two specialist models, while its speed and efficiency claims lack numerical results in the announcement.

A Blog post by Salesforce Sovereign Intelligence on Hugging Face

Low-bit quantization reduces the memory footprint and inference cost of large language models (LLMs), yet accurate FP4 quantization remains challenging. Practical formats such as MXFP4, NVFP4, and HiF4 pair a 4-bit floating-point payload with block-wise or hierarchical scale codes, making accuracy highly sensitive to how each local distribution is scaled. Default maximum-absolute-value scaling preserves dynamic range, but can spend scarce FP4 levels on rarely occupied outlier regions and leave dense tensor values poorly resolved. Existing scale-search methods minimize local element-wise reconstruction error, optimizing tensor fidelity rather than layer-output fidelity. We propose SoScale, a second-order weight scale-code search method for hierarchical FP4 quantization. SoScale searches hardware-valid scale codes under a Hessian-aware layer reconstruction objective and couples this search with GPTQ-style error recovery, selecting scales by their impact on layer outputs. Across Qwen3 models, SoScale consistently reduces downstream quantization degradation under MXFP4, NVFP4, and HiF4.

Multimodal diffusion large language models (dLLMs) generate text through iterative masked denoising with bidirectional attention, offering a compelling alternative to autoregressive vision-language modeling. However, each denoising step requires a full forward pass over long multimodal sequences in which visual tokens dominate the sequence length, and a fixed step budget further compounds this quadratic per-step cost. Existing inference acceleration approaches for multimodal dLLMs characterize attention redundancy implicitly, motivating designs from isolated attention visualizations rather than from measurement. We close this gap with DRAMA, a training-free acceleration framework grounded in a systematic dissection of attention redundancy in multimodal dLLMs. By probing attention tensors along spatial and temporal axes, we reveal two persistent redundancy structures: (i) attention mass consistently concentrates on a sparse subset of visual tokens, and (ii) attention distributions rapidly stabilize across denoising steps after an initial formation phase. These findings motivate two complementary mechanisms: Spatial-Redundancy-Aware Visual Token Pruning (SVTP) and Sufficiency-Guided Adaptive Stopping (SGAS). Together, DRAMA achieves an average 18.67 times inference speedup across six vision-language benchmarks without training, architectural modification, or attention caching, while maintaining competitive accuracy. Our results establish a strong, measurement-grounded acceleration baseline for multimodal dLLM inference.

Predicting every observation in a high-rate stream requires a model to finish before the next arrival. A large model may offer better predictions but miss that deadline. We propose RESYNC: the large model runs asynchronously, while a fast path updates its latest exact but stale latent state using the observations received since that state was computed and applies its unchanged prediction head. The updater learns from the frozen model's representations without task labels. Across one real and two synthetic streams and five architecture families, RESYNC beats the best compact or distilled alternative with a paired 95% interval above zero in all twelve dataset--architecture pairs with a material teacher advantage at anchor ages of 4--12 observations, and in eleven pairs at older tested ages up to 32. The share of the teacher's gain preserved remains complete on the real stream but is lower and generally declines with anchor age on the synthetic streams. In live replay with 1 ms arrivals, the synchronous large models cannot deliver any predictions on time while RESYNC serves at least 99.4% of arrivals within the interval, with 0.35--0.62 ms median latency on one CPU thread: around the same speed as or faster than the compact model, and 7.9--17.7$\times$ faster than synchronous large-model inference. RESYNC targets settings with background compute available but a strict per-observation response deadline.

Muse's AI can now place outbound calls to U.S. businesses — but it's not free.