AITOP WeeklyIssue 112292 stories

Gemini 4 Carbon opens, Claude Fable 5.2 tops index, Qwen4 previews

Singapore·9 Nov – 15 Nov·Ten minutes

The US and China swapped flagships almost simultaneously this week. Google opened Gemini 4 Carbon to paid APIs and Google AI Ultra subscribers on 10 November — its third new model in six weeks, with Ultra tiers from US$99.99 to US$199.99 per month. Anthropic shipped Claude Fable 5.2 and Mythos 5.2, cutting API pricing to US$8/40 per million tokens and claiming the top of the Artificial Analysis Intelligence Index with 65. On the Chinese side, Alibaba opened Qwen4-Max preview on its Token Plan, while Meituan open-sourced the 1.6-trillion-parameter LongCat-2.5 under an MIT licence. Meanwhile, a former OpenAI researcher claimed weight-fingerprinting evidence that Chinese labs trained on Nvidia Blackwell chips despite export controls. For developers in Singapore, prices are falling fast and the model-selection window just reopened.

01

Five things this week

Enough to retell
1
Anthropic 发布 Claude Fable 5.2,登顶 Artificial Analysis 智能指数
Fable 5.2 以 65 分登顶智能指数,比 Gemini 4 Carbon 高 5 分。
Artificial Analysis
2
Google 开放 Gemini 4 Carbon:付费 API 今日可用,Ultra 订阅每月 99.99 美元起
Gemini 4 Carbon 开放付费 API,Ultra 订阅每月 99.99 美元起。
宝玉
3
Anthropic IPO 路演披露多芯片战略:Claude 已跑通四大芯片平台
Anthropic IPO 路演披露 Claude 已跑通四大芯片平台。
Simon Willison’s Weblog
4
Sierra Releases Personal Agent Protocol Poppy 1.0 with Shopify and Meta Muse
Poppy turns agent-to-business access from screen-scraping into a discovery, identity and authorisation layer, meaning companies can gate agents without blocking them entirely. Developers building agents that act on user accounts should review the OAuth scoping flow and the extension model, since shopping is delegated to UCP rather than reinvented. The Apache 2.0 licensing and draft status make early adoption low-cost but subject to change before final 1.0.
shao__meng
5
Neuro-VL vision-language model trained on one million neuroimaging studies
Training on real clinical archives lets a specialized model beat general frontier systems on both capability and resistance to leading clinical prompts — a notable result for medical AI evaluation methodology. Radiology teams, hospital AI buyers and health AI developers should examine its benchmark design, especially the longitudinal comparison task, when assessing clinical imaging products.
arXiv: OpenAI
02

US vs China

The same week, both sides
DimensionUnited StatesChinaWhat it means for ASEAN
New modelsGoogle opened Gemini 4 Carbon; Anthropic released Claude Fable 5.2 and Mythos 5.2Alibaba opened Qwen4-Max preview; Meituan open-sourced LongCat-2.5Flagships arrived together; worth waiting a round for reviews
PricesFable 5.2 API cut to US$8/40 per million tokens; Gemini 4 Hydrogen about 30% of Sonnet 5.5's costMeituan LongCat-2.5 open-sourced under MIT licence, no commercial authorisation neededUS cuts API prices; Chinese open weights allow self-hosting
BenchmarksFable 5.2 tops Artificial Analysis Intelligence Index at 65GLM-5.3 scores 28.3 on Terminal-Bench 3.0, highest among open weightsBoth sides pushing open-weight records on coding benchmarks
Policy—Former OpenAI researcher claims evidence of Blackwell-trained Chinese modelsExport-control dispute may affect regional cloud procurement
03

ASEAN this week

More →
04

Policy this week

More →
05

China to ASEAN

More →
06

Builders this week

More →
2292Stories
333China
18ASEAN
252Sources
AITOP · generated by the editorial system