AI Research news and analysis on Web Pulse: 10137 curated articles tracking the latest AI Research developments, tools, and research, updated continuously from vetted sources.
DeepSeek open-sourced DSpark, a set of inference optimizations that achieve 60–85% faster generation on large language models, releasing the paper and code on GitHub.
Six former Google employees told Business Insider they left the company due to FOMO during the AI boom, uncertain job security from layoffs, and the desire for greater equity upside at startups or AI companies like OpenA…
China's government advisers are calling for measures to address a two-speed economy where AI and high-tech manufacturing boom while consumer spending lags, as fixed-asset investment fell 4.1% in early 2026 and the proper…
Despite AI accelerating code generation, software projects are not moving faster due to new bottlenecks like code review and cognitive debt. The Triple Debt Model—technical, cognitive, and intent debt—explains why unders…
OpenAI's ChatGPT Pro Community Substack published a profile on June 24, 2026 featuring economist Tyler Cowen sharing 10 real use cases for ChatGPT Pro, including travel planning, art analysis, and chess game analysis. Th…
The US government approved Anthropic's controlled release of its Claude Mythos AI model to approximately 200 vetted organizations, including government entities, under Project Glasswing. The model demonstrated the abilit…
VideoDB and Langfuse have released an open-source framework for evaluating multimodal vision-language models (VLMs) in video use cases, emphasizing task-specific benchmarks over generic model comparisons. The approach gu…
Liquid AI released LFM 2.5-230M, a 230-million-parameter model that outperforms larger models on data extraction benchmarks, achieving 22.51 on CaseReportBench versus 13.83 for Qwen3.5-0.8B and 2.28 for Gemma 3 1B. The m…
IBM Research unveiled its NanoStack architecture, a sub-1nm 3D transistor stacking technology that achieves a 50% performance improvement and 70% power reduction over its 2nm design. However, the transition to production…
ByteDance and Renmin University researchers released iLLaDA, an 8B diffusion language model that matches Qwen2.5 on base benchmarks but lags after fine-tuning. The model, trained from scratch on 12 trillion tokens, outpe…
Major cryptocurrencies fell 7-10% over the week as investor capital rotated from crypto into AI-related equities, with Dogecoin and Hyperliquid's HYPE token leading losses. The selloff was exacerbated by a tech-stock rou…
Z.ai released GLM-5.2, an open-weight AI model that performs within one percentage point of Anthropic's Opus 4.8 on agentic benchmarks while costing roughly one-fifth as much, according to CNBC. The company positions the…
Qualcomm CEO Cristiano Amon unveiled a five-year plan to challenge Nvidia's dominance in AI chips for data centers, projecting over $15 billion in annual AI component sales by fiscal 2029. The company also announced the …
A new proposal outlines a runtime-enforced checklist mechanism to prevent AI agents from guessing or filling gaps in user instructions, aiming to reduce errors and accidents as voice-controlled agents and APIs expand. Th…
NVIDIA tested its open-weight Nemotron family of models on real-world coding agent tasks and found a capability floor below which models cannot drive an agent loop at all. The smallest variant, Nano 12B, produced no resu…
Micron CEO Sanjay Mehrotra predicts the DRAM shortage will persist until at least 2028, with supply unable to catch up with AI-driven demand, despite new fab investments. The crisis is expected to keep memory prices elev…
A new study measuring AI adherence to software architecture rules found that even advanced models like Opus ignore them 60% of the time, raising concerns about reliability in structured development environments.
A Hacker News user proposes a blueprint for achieving human-like thinking in machines using current LLM technology, suggesting a system of random multi-domain image/video generation combined with specialized LLMs that an…
A new analysis of 80 million careers by SignalFire reveals that software engineers are AI's most resilient job category, now making up 55% of new hires at major tech companies, up from 46% in 2019. However, entry-level e…
A developer building an AI memory system called MemStrata found that it comprehensively beats RAG on mutating code content, as detailed in a paper on arXiv. The developer emphasizes the importance of trusting data over h…