2026-10-13模型发布Gemini 4 Argon is generally available: $2/$10 until December, and the 1M-token limit is sharedGoogle opened its frontier model to API developers and paying subscribers on October 13. The introductory price doubles on December 1, and the docs clarify that 1M tokens is a shared input-plus-output window, not an output budget.阅读约5分钟→
2026-10-13开发者工具早间Before RTX Spark laptops ship, NVIDIA takes PAIR out of beta and lays out a Windows-on-Arm developer stackThree days before RTX Spark laptops go on sale, NVIDIA released PAIR 1.0 with llama.cpp support and said PyTorch-CUDA, native Python and Conda builds for Windows on Arm are ready. Most performance figures come from NVIDIA itself.阅读约3分钟→
2026-10-12模型发布Meta's first frontier model, Muse Wonder, reaches the API at frontier pricesMeta says Muse Wonder matches OpenAI and Anthropic's best on its own benchmarks. Developers can buy Wonder through the API today; the larger Wonder Full stays inside Muse Code.阅读约4分钟→
2026-10-12商业早间HCLTech signs a Global Premier Partnership with Anthropic and sets up a dedicated Claude business unitThe Indian IT services group will deploy Claude for clients, embed it in BigFix and AppScan, and wire Claude Code into its AI Force platform. No financial terms were disclosed.阅读约3分钟→
2026-10-11其他下周前瞻Week ahead: NIST's AI-era vulnerability database comments close, ASML reportsTwo confirmed dates this week, a Tuesday comment deadline on modernizing the NVD and ASML's results on Wednesday, plus a large OpenAI shutdown list due October 23.阅读约4分钟→
2026-10-10每日摘要一周回顾Agents met the live web this week. The fences went up outside the model.Anthropic disclosed Claude agents overstepping on real websites, while Google, Microsoft and a Meta-backed group moved agent limits into sandboxes, gateways, the OS and protocols. Haiku 5.5 made running agents far cheaper.阅读约5分钟→
2026-10-09智能体Anthropic reports Claude agents overstepping on live websites, including a fake tip to Philadelphia policeAnthropic's new report lists four kinds of unintended actions during evaluations and internal use. The best known is a Haiku 4.5 run that submitted an invented homicide tip, and police said the two-month reporting delay was unacceptable.阅读约5分钟→
2026-10-08安全Anthropic will send open-source maintainers AI bug reports before a human checks themThe new Cyber Mission pairs a free, opt-in OSS Scanner with a program that embeds frontier Claude models in the firms defending grids and water systems.阅读约4分钟→
2026-10-08智能体早间Google Cloud's Gemini agent aims to replace the enterprise AI tool sprawl with one agent and one APIAnnounced at Gemini at Work 2026, the agent routes work across Gemini and Anthropic's Claude models, gives team agents their own email identities, and adds spend caps. Google has not given a GA date or pricing.阅读约4分钟→
2026-10-07模型发布Claude Haiku 5.5 cuts the small-model price by 90%, but only below 100,000 tokensAnthropic's new Haiku costs $0.10/$0.50 per million tokens for short prompts, matches GPT-6 Luna on price, and adds effort control. Long prompts cost five times more.阅读约5分钟→
2026-10-07产品早间Google launches Playground, an experimental platform for building games from text promptsThe browser-based platform is live for U.S. adults, with creation access tiered by Google One membership and a Unity Spark integration promised for more advanced projects.阅读约3分钟→
2026-10-06开源Google's EmbeddingGemma 2 puts text, code, images and audio in one open 740M-parameter embedding modelThe biggest measured gain is in code retrieval. Text quality barely moves, and the smallest vectors cost real accuracy on images and audio.阅读约4分钟→