Generative AI news and analysis on Web Pulse: 15653 curated articles tracking the latest Generative AI developments, tools, and research, updated continuously from vetted sources.
A developer launched Ceptile, an AI search platform that retrieves and ranks web pages and uses them to generate answers with links back to the underlying sources, according to a Show HN post. Ceptile also includes Notes…
OpenAI released GPT-6 Luna (US) on Amazon on 2026-09-22, according to the model listing on models.dev, with a 1,050k-token context window, 128k max output, and text plus image input. The model carries no announced retire…
Moonshot AI released Kimi K3 (Global) on Amazon on 2026-07-16, a text-and-image model with a 1049k context window and 131k max output, available as global.moonshotai.kimi-k3 on Amazon Bedrock. No retirement date has been…
OpenAI released GPT-6 Luna on 2026-09-22, a model that accepts text, image and PDF input with a 1,050k-token context window and 128k maximum output, according to the model listing on models.dev. The listing states no ret…
A new paper, "StableVQ: Practical Guidelines for Stable Vector-Quantized Tokenizer Training," addresses training stability in vector-quantized visual tokenizers, which underpin autoregressive and masked image generation …
Tabs CEO argues companies should deploy AI on verifiable, mundane tasks rather than creative work, citing a 2026 Boston Consulting Group survey finding 42% of regular frontline AI users save at least a full workday each …
A developer argues that conventional RAG architectures create a dangerous chasm between the data lakehouse and the AI stack, forcing teams to duplicate data and reimplement security in a separate vector database. The pos…
A new arXiv survey (2609.25017v1) concludes that effective deepfake governance requires defense-in-depth integrating forensic detection, verifiable provenance, and institutional accountability. The survey covers generati…
Researchers introduced ImIR, an image-instruction tuning method that adapts a single Qwen-Image-Edit model to six image restoration tasks using one low-rank adapter trained in about three hours on one GPU. ImIR replaces …
Researchers introduced RULER (Instance-aware Rubric Rewards for Reinforcement Learning), a method that converts each natural-language instruction into an instance-aware rubric of six items spanning semantic, visual, and …
Researchers proposed Point Diffusion Mamba (PDM), a method that combines diffusion models with state-space modeling for single-view 3D reconstruction under data-scarce conditions, according to an arXiv paper (arXiv:2609.…
A new arXiv paper, 2609.25134v1, presents a probabilistic framework for large language models, describing them as probability measures over token sequences specified by autoregressive conditional distributions. The paper…
A September 21, 2026 arXiv paper submitted to the Computer Science > Machine Learning category identifies "sequential reappearance," a failure mode in diffusion data-point unlearning where an instance initially judged fo…
A 2026 analysis argues that generative AI remains the default for content production while agentic AI should be reserved for multi-step, verifiable workflows such as bug fixing and data reconciliation. The piece cites An…
BreezeBlue's Breeze TTS 2 has become the highest-rated open-weight text-to-speech model on the Artificial Analysis Provider Voice Arena, scoring Elo 1215 and beating ElevenLabs' Eleven v3 at 1175 while hosting at $34 per…
A developer has published a 2026 guide detailing how real-time AI video avatars can be integrated into Zoom, Microsoft Teams, Discord, and streaming platforms, covering latency trade-offs between 2D and 3D avatars, face-…
Higgsfield AI posted samurai-game demos on September 22nd labeled Claude Opus 5.5 and GPT-6 Astra, using the two rival models to showcase its own game-building and asset-generation tools rather than to establish which mo…
Simon Willison announced a Birds of a Feather session on agentic engineering in San Francisco on October 14th, co-hosted with Jesse Vincent, focused on informal show-and-tell among builders and experimenters. The session…
A developer building a custom voice for an AI agent worked around the limited two-pole output of the VoxCPM text-to-speech model by pitch-shifting a feminine-pole render down four semitones with a single ffmpeg rubberban…
OpenAI released GPT-6 Luna, the fast, cost-efficient model in its GPT-6 series positioned below GPT-6 Sol, priced at $0.10-$0.20 per million input tokens and $0.50-$0.75 per million output tokens, with cache reads at $0.…