Updated daily177 verified stories
New frontier, open-weight, multimodal, reasoning, and specialist AI models—plus the evidence behind their capability claims.
Every story links to the strongest available evidence: original sources when available, otherwise clearly attributed reporting.
What happened, why it matters, and what to watch — no jargon tax.
When the signal is thin, we publish nothing rather than padding the feed.
A growing stream of verified perspectives for people who need to understand AI without chasing hype.
The llama.cpp project released version 0.3.0 with support for the dots3-note vision-and-audio model, new DeepSeek 4 tensor-split handling, GLM-4.5-Air multi-token prediction, and updates across its inference backends and web interface.
The South China Morning Post reports that open-weight models reached 62% of token volume on Vercel’s AI Gateway, led by DeepSeek-V4-Flash and other Chinese models. The reported shift suggests developers are prioritizing lower-cost models for token-intensive production workloads, although the underlying platform…
A new arXiv benchmark evaluates chemistry-focused large language models across varied instructions, molecular representations and eight task categories, reporting substantial prompt sensitivity, representation dependence and uneven performance.
A preprint introduces Evidence-State Reliability, a measure for testing whether intermediate evidence remains usable as language-model pipelines process degraded inputs. In one controlled evaluation using GLM-5.2, parser-valid outputs persisted while stage success declined.
Fortune reports, citing payment-processor data from Ramp, that Claude Fable 5 represented 11% of corporate spending on Anthropic models two months after release, but that share has since stalled. The report says the pattern could pressure frontier AI companies to compete more aggressively on price, although Ramp’s…
A new arXiv paper identifies a gap in LLM unlearning: an agent may stop recalling information from its weights but recover it through web search, retrieval or database tools. The authors propose a two-stage method to reduce both forms of recovery while preserving legitimate tool use.
A preprint reports that language models can appear to forget targeted information under ordinary tests while still recovering it under strategic adversarial prompts.
India Today reports that Nvidia’s agreement with AI startup Poolside would give Nvidia access to Poolside’s technology and bring more than 100 employees into its Nemotron open-weight model effort. The report also says Nvidia will invest $1 billion in Poolside at a $12 billion pre-money valuation, while rising…
A new arXiv paper presents LURE, a zero-data self-play method in which one language model sets task difficulty while another solves verifiable reasoning challenges. The authors report stronger out-of-distribution zero-shot accuracy than trained baselines across nine held-out benchmarks, but the abstract does not…
An arXiv preprint introduces NC-GRPO, a reinforcement-learning method that perturbs a vision-language model’s hidden representation rather than its input image. The authors report improved out-of-domain mathematical reasoning and hallucination robustness on Qwen2.5-VL-7B, while noting tradeoffs between…
An arXiv paper introduces TASSO, a method for adapting vision-language models to new tasks while reducing the loss of earlier abilities and zero-shot performance. The authors report improvements over existing methods in CLIP-based continual-learning benchmarks, but the source provides no numerical results.
A new arXiv preprint introduces ConvergeFlow, a flow-based language model designed to end at valid token embeddings without a cross-entropy-trained decoder. The authors prove convergence under stated regularity conditions and report competitive results on OpenWebText, but the abstract does not provide benchmark…
One useful briefing each week
Get the week’s verified AI news, original data, useful tools, learning picks, and fresh AI jobs.
Hiring an AI professional or launching a useful AI product? Put it in front of people who came here to learn and act.
Post an AI job Submit an AI tool