Google AI Edge Foresight – offline, private meeting transcripts
Google AI Edge Foresight enables offline, private transcription of meeting content. It aims to enhance data privacy and accessibility for users. (developers.google.com)
Claude Haiku 5.5 is a new small model optimized for high-volume, cost-sensitive and speed-sensitive tasks - summaries, compaction, database queries, classification, subagents for coding, live support, and browser-based work. It delivers large reductions in price (up to ~90% lower for requests under 100k tokens and substantially lower on average) and is Anthropic’s fastest Haiku-class model. Benchmarks show big gains over Haiku 4.5 across knowledge-work, computer-use, visual-reasoning and multidisciplinary-reasoning suites, and an adjustable effort setting lets users trade cost for capability. Early customer evaluations report lower latency (Asana: ~30% latency reduction), higher task accuracy (HubSpot CRM suite: 92.8%), and improved query performance versus Haiku 4.5 (AlphaSense: 0.84 vs 0.76), while larger models (Sonnet/Opus 5.5) remain superior for complex agentic coding.
Deployment and ecosystem updates include lower cache-read pricing for Sonnet 5.5 (50% cut), new monthly API credits for Max and Team subscribers, and beta support for computer- and browser-use in the Python and TypeScript SDKs. Safety and alignment are improved relative to Haiku 4.5, with stricter cybersecurity safeguards (allowing defensive but blocking offensive techniques) and the same biology restrictions used across Opus/Sonnet. Haiku 5.5 is generally positioned as a fast, inexpensive subagent and high-throughput workhorse available now on major cloud providers and the Claude Platform.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.
Google AI Edge Foresight enables offline, private transcription of meeting content. It aims to enhance data privacy and accessibility for users. (developers.google.com)
Meta and Microsoft are reducing employee use of Anthropic’s Claude AI, shifting focus to their own AI tools. Microsoft is cutting internal AI spending, while Meta is replacing Claude with its proprietary models for internal use. (rswebsols.com)
Qwen 3.8-27B, a free local language model, replaces paid cloud AI services like ChatGPT and Claude for a user’s daily tasks. It provides similar performance at no cost, removing the need for subscriptions. (xda-developers.com)
Asking multiple AI models for their top opinions reveals that Chinese models tend to align more with the mainstream consensus, while others show more contrarian responses. The study scores models based on their agreement with group consensus, highlighting regional differences in AI responses. (magicnumbers.io)
Claude Haiku 5.5 is the most capable small model from Anthropic, offering a 75% reduction in running costs compared to its predecessor. It is optimized for high-volume, cost-sensitive tasks like summaries, classification, and customer support, with improvements in alignment and efficiency. (twitter.com)
OpenAI is developing GPT-6 with the capability to generate responses using text, visuals, and interactive elements. The model aims to create an intelligent user interface accessible to everyone. (openai.com)
Today's best Hacker News stories, summarized and screenshotted, one email a day.