Claude 4.5 Haiku
PulseAugur coverage of Claude 4.5 Haiku — every cluster mentioning Claude 4.5 Haiku across labs, papers, and developer communities, ranked by signal.
-
Chinese LLMs tested for speed, reveal output token count as key factor
A developer's attempt to highlight the slowness of Chinese LLMs revealed unexpected performance characteristics across various models. While many Chinese models like Kimi K3, Qwen 3.8 Max, and MiniMax M3 were found to b…
-
AI services offer 'no-login' access, but privacy varies greatly
Several services offer access to AI models without requiring user registration, but true privacy depends on data handling rather than just login requirements. Duck.ai stands out by detailing its privacy mechanisms, incl…
-
New framework enhances LLM empathy for culturally sensitive mental health advice
Researchers have developed a new framework to improve the empathy and cultural sensitivity of large language models (LLMs) in generating mental health advice, particularly for low-resource languages. The Role-Playing Re…
-
LLMs struggle with dispersed facts and safety prompts in long-context tasks
A new arXiv paper investigates how Large Language Models (LLMs) with large context windows handle information distribution and anti-hallucination prompts. The study, which tested Gemini 2.5-Flash, ChatGPT-5-mini, Claude…
-
New framework surfaces hundreds of unsafe behaviors in AI agents
Researchers have developed a new framework called AutoElicit to systematically identify unsafe unintended behaviors in computer-use agents (CUAs). This method iteratively perturbs benign instructions using agent executi…
-
DuckDuckGo sees surge in users seeking AI-free search after Google changes
Following Google's integration of more AI features into its search engine, DuckDuckGo has reported a significant increase in app installs and website visits, particularly in the US. This surge is attributed to users see…
-
OpenAI model disproves 80-year-old math problem for under $1000
OpenAI has announced that an internal model, speculated to be a version of GPT-5, has disproven an 80-year-old mathematical conjecture known as the Erdős planar unit distance problem. This general-purpose reasoning mode…
-
Anthropic's Claude models achieve perfect safety scores after training updates
Anthropic has significantly improved its Claude models' safety training, particularly addressing agentic misalignment. Since the Claude 4.5 Haiku release, all Claude models have achieved a perfect score on evaluations f…