Claude-4.5-Sonnet
PulseAugur coverage of Claude-4.5-Sonnet — every cluster mentioning Claude-4.5-Sonnet across labs, papers, and developer communities, ranked by signal.
5 day(s) with sentiment data
-
LLMs exhibit CARLAR influence, hallucinating cheque amounts
A new research paper published on arXiv details a phenomenon called CARLAR influence, where multimodal large language models (LLMs) incorrectly extract the same amount for both courtesy and legal fields on cheques, even…
-
GPU VRAM Sizing Error for MoE Models Detailed by Logic42
A technical analysis by Logic42 highlights a common miscalculation in sizing GPU clusters for large Mixture-of-Experts (MoE) models like DeepSeek-V4.1-Flash and Moonshot Kimi K2.5. The error involves underestimating VRA…
-
Cursor Pro user slams AI support after quota depleted in 2 days
A Cursor Pro user reported exhausting their monthly quota of "Other Models" within two days of purchasing an annual plan for $192. The user experienced significant usage from Claude models, with one Opus run consuming 7…
-
AI coding agents advance with new optimization, cost control, and safety research
Multiple research papers explore advancements in AI coding agents, focusing on improving their efficiency, reliability, and safety. Self-HPO introduces a recursive framework to optimize both the agent's policy and its i…
-
Darwin-180B-RSI model tops legal benchmarks without domain training · 1 source tracked
VIDRAFT's Darwin-180B-RSI, a 180-billion-parameter large language model, has achieved top performance on Swiss legal reasoning benchmarks, including LEXam and LEXam-hard. Notably, the model accomplished this without any…
-
AI model tops Swiss law exam benchmark without legal training
An open-weight 180B model named Darwin-180B-RSI, developed by the Korean startup VIDRAFT, has achieved top rankings on legal reasoning benchmarks, despite never being trained on legal data. The model utilizes a novel ap…
-
Video-DeepResearch 35B model outperforms GPT-5 on video reasoning benchmark
The open-weight Video-DeepResearch 35B model has achieved a score of 64% on a new video reasoning benchmark, surpassing GPT-5's score of 52.5%. This model also outperformed Claude 4.5 Sonnet, though the research team gr…
-
Video-DeepResearch agent sets new SOTA on video QA benchmarks
Researchers have developed Video-DeepResearch (Video-DR), a multimodal agent capable of processing continuous video streams for complex research tasks. This new framework addresses modality bias and parametric knowledge…
-
Apple M5 Ultra chip to feature 768GB memory, cost $13,000
Apple's upcoming M5 Ultra chip is rumored to feature between 512GB and 768GB of unified memory and is expected to cost around $13,000. Despite the high price, the chip's memory capacity may decrease as RAM production in…
-
Amazon Nova Act uses AI for scalable UX testing
Amazon has introduced Amazon Nova Act, a multimodal foundation model designed to enhance user experience (UX) testing for web interfaces. Unlike traditional automation tools that rely on brittle, hard-coded scripts, Nov…
-
AWS GraphRAG accelerates drug research by 87% using unified knowledge graph
Amazon Web Services has developed a GraphRAG framework that significantly accelerates drug research and development by integrating disparate proprietary databases into a unified knowledge graph. This system, utilizing A…
-
Anthropic's Claude 4.5 Sonnet allegedly surfaces online
A user on Reddit shared a screenshot from Twitter, allegedly showing a new model from Anthropic named "Claude 4.5 Sonnet." The post expresses surprise or alarm, indicated by the "Oh god" title. The source of the informa…
-
Developers seek affordable AI coding agents beyond GitHub Copilot
Developers are seeking cost-effective AI coding agents as GitHub Copilot's pricing becomes prohibitive. Users are evaluating platforms like Cursor, which offers a flat $20 monthly fee, and are concerned about potential …
-
Zyphra's ZAYA1-8B model shows strong reasoning on AMD hardware
Zyphra has released ZAYA1-8B, an Apache 2.0 licensed Mixture-of-Experts reasoning model with 8.4 billion total parameters and approximately 760 million active parameters. Notably, the model was trained entirely on AMD I…
-
Cursor IDE users hit with surprise bills from hidden Anthropic model usage
Users of the Cursor IDE are reporting unexpected and significant billing charges due to the platform's cloud agent feature. It appears that Cursor automatically utilizes Anthropic models, including versions of Claude Op…
-
Cursor IDE users hit with unexpected bills from automatic Anthropic model use
Users of the Cursor IDE are reporting unexpected and high billing charges due to the platform's cloud agents automatically utilizing Anthropic models. One user experienced a significant charge after the cloud agent cons…
-
New ACCORD framework boosts LLM agent task completion by 20%
Researchers have introduced ACCORD, a new framework designed to improve the performance of language agents by enabling them to better ground their actions in observed environmental context. ACCORD addresses the issue of…
-
New benchmark UXBench highlights MLLM limitations in UI reasoning
Researchers have introduced UXBench, a new benchmark designed to evaluate how well multimodal large language models (MLLMs) can reason about user experience (UX) based on UI screenshots. The benchmark includes 2,000 VQA…
-
LoRA fine-tuning for telecom AI shows validation loss disconnect
Researchers explored parameter-efficient fine-tuning (PEFT) using LoRA configurations on the Qwen2.5-3B model for telecommunications customer support. They developed a synthetic data generation method and evaluated 16 L…
-
Fudan, Tongyi Lab unveil ToolCUA for agents choosing between GUI and tools
Researchers from Fudan University and Tongyi Lab have developed ToolCUA, a new training paradigm for agents that can effectively utilize both graphical user interface (GUI) operations and tool calls. Experiments reveale…