Xiaohongshu AllSpark Open-Sources Search Agent Model Iris, 35B and 397B Versions Lead at Same Level
Xiaohongshu AllSpark team releases and open-sources Search Agent model Iris, with weights and evaluation code already public, data and training recipes to follow.
Model Releases & Updates
Xiaohongshu AllSpark Open-Sources Search Agent Model Iris, 35B and 397B Versions Lead at Same Level
Xiaohongshu AllSpark team releases and open-sources Search Agent model Iris, with weights and evaluation code already public, data and training recipes to follow.
SiliconFlow Launches Open-Source Model Hy4 preview, 770B Total Parameters, 1M Context
SiliconFlow announces the launch of open-source model Hy4 preview on their platform. The model has 770B total parameters, 49B active per token, supports 1M context, and uses Apache 2.0 license, targeting coding, analysis, research and complex real-world work. Users can integrate it with Claude Code, Codex, Cursor and other existing tools; pricing shows $0.834 per 1M tokens input, $2.501 output, $0.042 cached.
Product Releases & Updates
Apple Releases New Generation Apple Intelligence, Siri AI Officially Launches as Beta
Apple releases new generation Apple Intelligence, launching a fully redesigned Siri AI with personal context understanding, screen awareness, system-level app操作 and cross-device conversation, available today as English beta with 2027 system update, expanding to French, Japanese, Korean, Portuguese and Spanish next month.
Industry News
Anthropic Plans Nasdaq Listing, Targeting $2 Trillion Valuation with Consecutive Second Quarter Profit
Anthropic told investors it will achieve consecutive second quarter profit, but this is based on adjusted metrics excluding stock-based compensation; according to Financial Times, gross margin exceeds 80%, not yet accounting for revenue sharing with partners like Amazon and model training costs.
Research Papers
DeepSeek-V4.1-Flash (Max) Enters Agent Arena Open-Source Model Ranking at #3
DeepSeek-V4.1-Flash (Max) enters Agent Arena open-source model ranking at #3, with +4.87% net improvement and $0.07 median cost per task, reshaping the Pareto frontier. Its cost is 68% lower than #2 ranked Hy4 preview while only 0.09 percentage points behind in performance; ranked #12 overall, #4 in Confirmed Success signal (+13.75%).
Tips & Insights
Is Big Tech's AI Development Slowdown a Safety Consensus or a Cartel
Sam Altman, Dario Amodei, Demis Hassabis and Elon Musk broadly agreed last weekend to slow down AI development, proposing third-party audits, domestic lab regulation and a global slowdown agreement, with critics calling it a "cartel" to suppress competitors and the open-source movement.
GPT-5.6 Luna vs GPT-6 Astra: Is a $1.20 Model Good Enough for Code Review
Entelligence compared GPT-5.6 Luna and GPT-6 Astra for code review on 50 public benchmark PRs using the same prompts, with Luna finding 69 verified bugs (Astra: 92) at total cost $0.20 vs $5.66, and accuracy 74% vs 96%.
Tomer Tunguz Analyzes Five Schools of Thought Behind Amodei's Frontier Pause Proposal
Tomer Tunguz analyzes Dario Amodei's call to slow frontier AI development, sorting through five schools of thought: interpretability, labor, economic, geopolitical and regulatory capture, pointing out that none have provided specific pace.
How Anthropic Re-architected Test Impact Analysis Service to Handle CI Pressure from Agentic Coding
Anthropic engineers deliver 8x more code per quarter compared to 2021-2025 average, with 80% written by Claude, and CI tasks grew 25x in six months.
【Hacker News Top Posts】
Keywords: AI OR GPT OR LLM OR Claude OR OpenAI OR "machine learning" OR "large language model" OR AGI OR "generative AI" OR AI agent OR RAG OR diffusion OR transformer
Data Source: hnrss.org | Filtered from Hacker News
1. OpenAI's malicious bot swarm attacked RubyGems
Article URL: https://www.theregister.com/security/2026/09/14/openais-malicious-bot-swarm-attacked-rubygems/5296356 Comments URL: https://news.ycombinator.com/item?id=49705979 Points: 1 # Comments: 0
2. Claude for Financial Advisors
Article URL: https://claude.com/blog/claude-for-financial-advisors Comments URL: https://news.ycombinator.com/item?id=49705955 Points: 1 # Comments: 1
3. Sloppy Joe's – The era of AI signage
Article URL: https://luke.zip/posts/sloppy-joes/ Comments URL: https://news.ycombinator.com/item?id=49705904 Points: 2 # Comments: 0
4. China says AI CEOs' call for a slowdown is 'fear mongering'
Article URL: https://www.cnbc.com/2026/09/14/china-ai-slowdown-us-tech-ceos.html Comments URL: https://news.ycombinator.com/item?id=49705877 Points: 3 # Comments: 0
5. Personal Statement on AI Risk
Article URL: https://docs.google.com/document/u/1/d/e/2PACX-1vQNl3SEX5IyA6d9qHjjFZN-qzGRZNFI6b63g-yu1Fy-ZYkVfCWm7i9WXRXw63m6yDB_auDuPLyQ7jBm/pub Comments URL: https://news.ycombinator.com/item?id=49705873 Points: 1 # Comments: 0
Originally published on WeChat Official Account 「比特财商」。
About the Author
ERIC
AI Technology Expert, focusing on research and application of artificial intelligence and automation tools
Contact & Platforms