A focused feed of models, agents, research and open-source releases for people building with AI.
Anthropic introduced self-assessment , where Claude grades the safety of each conversation right after responding. Flags 78% of harmful conversations and 94% of high-priority ones at 1% false-positive rate on Fable 5.
Parallel launched Parallel for Finance , a universal index upgrade applied across all its APIs with no new subscription. On FinSearchComp T3, Parallel Advanced hits 57.4% accuracy at up to 2x lower cost than Exa and Perplexity.
LongChang tests whether LLMs can notice and explain altered physics with no task or hints given. Only GPT-6-Astra discovered hidden laws, scoring 16.7% while 11 other frontier models scored 0%.
New paper proves first universal approximation theorems for data-driven smoothing and forecasting of dynamical systems. Authors: Calvello, Carlson, Kovachki, Manta, Stuart; arXiv 2603.20359 .
CATT benchmark fine-tunes Qwen3.5-4B on fictitious facts for unambiguous attribution ground truth: project page . Redundancy and lexical distractors consistently break gradient-based attribution; TRAK falls to zero recall with distractors.
Kyutai shows Neural Thickets gains come from logit diversity, not weight-space experts. Logit PSV adds one random bias vector over the vocabulary, model stays frozen. Beats full-weight PSV on 13 of 15 reasoning benchmarks across four tested LLMs.
KAIST's L3G-World generates explorable 3D worlds at 9 FPS from a single image. Each 3D Gaussian stores a VAE latent, rendering directly into the diffusion model's conditioning space. Memory grows with scene extent, not stream length, keeping GPU use flat during long exploration.
CMU-led team unifies pretraining, SFT, and RL into one maximum-likelihood objective called unified training. RL rollouts reframed as likelihood fitting to model-generated targets via MaxRL extension . Single trainer mixes static text and RL generations; staged training becomes a special case.
Prime Intellect released Prime Mega MoE , adding a backward kernel to DeepSeek's forward-only Mega MoE. Up to 3.11x faster MoE kernels versus a PyTorch baseline, 1.34x faster end-to-end training. Qwen3-30B-A3B steps 48% faster and Qwen3-235B-A22B 55% faster than TorchTitan.
Pika released Pika H3 Flex , an open-weight video+audio model post-trained on MiniMax H3. Generates 15-second clips with soundtrack in as little as 7 seconds using a custom 6-step sampling schedule.
Multi-account switching lands on Claude iOS and Android , no more sign-out dance between work and personal. Voice mode now works with Apple CarPlay on iPhone for hands-free driving conversations.
Jev and winnow:e4b matched the reference on all 12 Claude Code fixes, Claude in a plain message matched 9, and a filename rule matched 3. Only one of the 12 fixes left a bug behind, and all four pickers caught it, so the test measures fit and says nothing yet about catching more bugs.
NVIDIA's LucidLyra turns one image into a 3D Gaussian scene with geometry, semantics, and instances. A frozen camera-controlled video diffusion model supplies multi-view latents to a single Gaussian decoder.
IFM delivered K2 Horizon 360: 8 artifact categories and 30+ repos covering every LLM training stage, see artifact directory . 380 checkpoints across six models from pre-training through RL, letting you trace when capabilities emerged.
Artificial Analysis launched the Progress Index v1 , a single capability scale for every LLM since 2022. Uses item response theory to compare models tested on different benchmarks over time. Frontier rose ~68 points in 2025 and 56 so far in 2026, roughly double 2023's pace.
Gemini now lets you generate, preview, and edit files like PDFs, docs, slides and spreadsheets inside the chat . Highlight any passage and ask Gemini to rewrite text or revise visuals without leaving the preview.
Alibaba's Wan team released Wan-Animate-2 , an end-to-end character animation model with native ComfyUI support . No pose extraction or skeleton preprocessing: the driving video goes straight into a redesigned Diffusion Transformer.
Phonic 1 (High) debuts at #1 on the Speech to Speech Index at 87.1, edging out Grok Voice 3.0. Leads τ-Voice agentic benchmark at 79.9%, a 5.4 point gap over GPT-Live-1 (Astra, medium). Scores 98.9% on Big Bench Audio reasoning, third behind StepAudio 3 and Qwen Audio 3.0.
FrostWolfAI released Spark-X2.6-4B-abliterated-GGUF , a 4B uncensored local model in 14 quantization formats. Refusals drop from 100/100 to 2/100 at just 0.0177 KL divergence from the base model. Retains 1M-token context, controllable chain-of-thought reasoning, and native tool calling intact.
Cohere closed the initial tranche of a Series E targeting up to CAD $7B at a CAD $35.7B valuation, the largest private round in Canadian history . Schwarz Group leads, with NVIDIA, Inovia, Bell Ventures, Bullhound, Prosperity7, and the Government of Canada joining.
Zed announced DeltaDB , a Git-compatible version control system for agent context and conversations. Every human or agent edit becomes a delta in a causal DAG, linking code to the conversation that produced it.
New paper shows flow robot policies produce near-Gaussian per-step action distributions, with under 2% Wasserstein error. Gaussian Action Readout replaces the flow policy's final layer with a Gaussian head trained on 0.02% of states.
Fish Audio released Drama 3 , a 35B MoE TTS model with 3B active parameters, open weights on Hugging Face. Ranks #1 open-weights on both Artificial Analysis Speech Arenas and #1 overall on Pronunciation Robustness at 94.6%.
Mistral released the open weights for its 1-trillion parameter Mistral Large 4 model, now leading open-weight coding benchmarks with multimodal input and cyberdefense tuning.
No stories match these filters.
