AI 新动静

[AI 新动静 @aixindongjing]

@vercel_dev @GoogleAI Vercel 表示，Google 的 Gemini Decision 1 已通过模型 ID google/gemini-decision-1 在其 AI Gateway 上提供，可返回带概率的结构化决策，适用于分类、路由和评分。

[AI 新动静 @aixindongjing]

Gemini Decision 1 现已在 Vercel 的 AI Gateway 上线

Vercel 表示，Google 的 Gemini Decision 1 已通过模型 ID google/gemini-decision-1 在其 AI Gateway 上提供，可从文本、图像、文档、视频和音频返回带概率的结构化决策，建议用于分类、路由和评分。

[Vercel Developers @vercel_dev]

Gemini Decision 1 from @GoogleAI is now available on AI Gateway.

Get structured decisions with probabilities from text, images, docs, video, and audio. Use it to classify, route, and score.

𝚐𝚘𝚘𝚐𝚕𝚎/𝚐𝚎𝚖𝚒𝚗𝚒-𝚍𝚎𝚌𝚒𝚜𝚒𝚘𝚗-𝟷

https://t.co/uZV38eRExo

[AI 新动静 @aixindongjing]

Gemini Decision 1 登陆 Vercel AI Gateway

模型 ID 为 `google/gemini-decision-1`。Vercel 称它可从文本、图像、文档、视频和音频生成带概率的结构化决策，适用于分类、路由和评分。

#Google #模型 #基础设施

[AI 新动静 @aixindongjing]

@OpenRouter OpenRouter 表示，Google 的 Gemini Decision 1 已上线，型号为 google/gemini-decision-1，标价为每 100 万输入 token 0.10 美元，输出免费。

[AI 新动静 @aixindongjing]

Google 的 Gemini Decision 1 首发即登陆 OpenRouter

OpenRouter 表示，Gemini Decision 1 是其所谓的 Google 首款决策模型，型号为 google/gemini-decision-1，会对是非题、选择题和评分问题给出概率。标价为每 100 万输入 token 0.10 美元，输出免费，上下文长度 262K token。

[OpenRouter @OpenRouter]

Google's very first decision model is here, and it's on OpenRouter from day one.

Gemini Decision 1 answers yes/no, multiple choice, and score questions with a probability, for $0.10/M input and free output.

https://t.co/x95JfrtOEf

[AI 新动静 @aixindongjing]

@ArtificialAnlys @GoogleDeepMind Artificial Analysis 称，Google 的 Gemini Decision 1 在 Decision Index v2.1 上得分 66，落后 Drex Max v1.1 和 GLiDE v1.2 的 73 分。

[AI 新动静 @aixindongjing]

Artificial Analysis：Gemini Decision 1 在 Decision Index 上得分 66，落后领先模型 7 分

Google 的 Gemini Decision 1 在 Decision Index v2.1 上得分 66，落后得分均为 73 的 Drex Max v1.1 和 GLiDE v1.2。每 100 万项任务成本 345 美元，约为 Drex Max v1.1 的三倍。

[Artificial Analysis @ArtificialAnlys]

Google has released its first decision model, Gemini Decision 1, scoring 66 on the Artificial Analysis Decision Index and leading in knowledge and reasoning benchmarks

Gemini Decision 1 from @GoogleDeepMind scores 66 on the Artificial Analysis Decision Index v2.1, in line with Elephant 9B (@baselabs) and Jev 1.14 (@typesafeai), and 7 points behind leaders Drex Max v1.1 (@NaceAI) and GLiDE v1.2 (@fastinoAI), which both score 73. It is priced at $0.10 per 1M input tokens and costs $345 per 1M Tasks across the Index, around three times higher than Drex Max v1.1 ($103) due to its higher pricing

Key results for Gemini Decision 1:

➤ Not yet frontier on Decision Index: Gemini Decision 1 scores 66 on the Artificial Analysis Decision Index. It leads knowledge and reasoning benchmarks, scoring 66 on AA-Omniscience-Decision and 49 on HLE-MultiChoice. It trails on other evaluations, including our new proprietary classification and extraction benchmark AA-Paperclip, where it answers all 15 questions correctly on 33% of documents vs. 63% for Drex Max v1.1

➤ More than 3x the Cost per Task of leading models: Gemini Decision 1 costs $345 per 1M Tasks, compared with $103 for Drex Max v1.1, $156 for Jev 1.14. This difference is driven by its token price of $0.10 per 1M input tokens vs. $0.03 for Drex Max v1.1. Gemini Decision 1 has a similar Cost per 1M Tasks to GPT-6 Luna (Decisions, $339)

➤ Strong calibration for reasoning tasks: We measure calibration with the Brier score, the average squared gap between the probability a model assigns to its chosen answer and whether that answer was correct (lower is better). Gemini Decision 1 has a Brier score of 0.18 across the Index, between Drex Max v1.1 (0.16) and Jev 1.14 (0.19). Calibration varies by task. with its Brier score of 0.20 on HLE-MultiChoice among the lowest of all models in the Index, while its 0.14 on When2Call compares with 0.04 for Drex Max v1.1

[AI 新动静 @aixindongjing]

@jeffreygwang @OpenAI OpenAI 发布研究，探讨无 CoT 能力与 CoT 可控性这两个思维链可监控性代理指标如何随预训练、强化学习和蒸馏等训练过程变化。

[AI 新动静 @aixindongjing]

OpenAI 研究思维链可监控性在模型训练过程中的变化

预训练中，基础模型困惑度降低，无 CoT 能力稳步提升；在一款 Astra 级模型中，仅用长度限制和最终答案评分做强化学习，就使 CoT 可控性大幅提升。

[Jeffrey Wang @jeffreygwang]

(1/10) Today, we’re sharing new research from @OpenAI: How does CoT monitorability (no-CoT capability and CoT controllability) evolve over the model training lifecycle?

We emerged with unsurprising but sobering findings: as we scale toward more intelligent systems, models are becoming harder to monitor through their CoT. 🧵

[AI 新动静 @aixindongjing]

Google 推出 Gemini Decisions API 和 Gemini Decision 1

Google 推出了 Gemini Decisions API，这是 Gemini API 中一个用于快速进行封闭式决策的端点，由新模型 Gemini Decision 1（gemini-decision-1）驱动。该 API 可通过 Google AI Studio 中的 Gemini API 使用。

#Google #模型 #基础设施

[AI 新动静 @aixindongjing]

@arena @AnthropicAI Arena 称 Claude Fable 5.2（Max）在 Agent Arena 中夺得第 1 名，净提升 +30.4%，约为第 2 名 Claude Opus 5.5 的 1.6 倍。

[AI 新动静 @aixindongjing]

Arena 将 Claude Fable 5.2 列为 Agent Arena 第 1 名

Arena 称，Claude Fable 5.2（Max）在 Agent Arena 中夺得第 1 名，净提升 +30.4%，约为第 2 名 Claude Opus 5.5 的 1.6 倍，基于 4.7k 次真实世界智能体会话。Fable 5.2 在 Code、Chat 和 Work 类别中也排名第 1。

[Arena.ai @arena]

Big news: Claude Fable 5.2 (Max) by @AnthropicAI takes #1 in Agent Arena with an unprecedented lead: +30.4% net improvement, about 1.6x the #2 model Opus-5.5, across 4.7k real-world agentic sessions from our global community. It's also #1 in Code, Chat and Work.

A major leap over Claude Fable 5.1 (Max)
- Rank: #5 → #1, Net improvement: +14.8% → +30.4%
- Median cost per task: $7.7 → $7.3 (6% lower). The API pricing $10/$50 → $8/$40 per million input/output tokens (20% lower)

vs. The Frontiers
- #1 Claude Fable 5.1 (Max): +30.4% | $7.3/task
- #2 Claude Opus 5.5 (Max):+19.5%) | $5.5/task
- #3 Gemini 4 Carbon (High):+17.0%) | $1.5/task
- #6 GPT-6-Astra (Max): +14.5% | $6.0/task

Breaking down performance across signals:
Fable 5.2 improves over 5.1 across all four reported signals, with particularly large gains in user satisfaction and steerability.
- #1 Praise vs. Complaint , +73%, biggest jump vs +41% the next model.
- #1 Steerability: +33%
- #1 Confirmed Success: +22%
- #3 Bash Reliability: +14%

#1 across all major categories
- #1 Code: +34%
- #1 Work: +31%
- #1 Chat: +27%
- Versus Fable 5.1: Code #6 → #1 (+19.8 %), Work #4 → #1 (+15.8 %), Chat #6 → #1 (+12.0%).

Huge congrats to the @AnthropicAI team on this incredible release!

[AI 新动静 @aixindongjing]

Arena 将 Claude Fable 5.2（Max）列为 Agent Arena 第 1 名

Arena称，该模型净提升+30.4%，约为第2名 Claude Opus 5.5 的1.6倍；排名基于全球社区4.7k次真实世界智能体会话。它也位列 Code、Chat 和 Work 类别第1名。

#Anthropic #模型 #编程

[AI 新动静 @aixindongjing]

@OpenRouter OpenRouter 推出处于 alpha 阶段的 Extractions API，可从文本、图像和 PDF 中按所述字段提取信息并以结构化数据返回。

[AI 新动静 @aixindongjing]

OpenRouter 推出 Extractions API，现处于 alpha 阶段

开发者只需描述所需字段，例如电子邮件中的姓名或收据上的总金额，该 API 就会从文本、图像和 PDF 中提取这些信息，并将每项内容以结构化数据的形式返回。

[OpenRouter @OpenRouter]

The Extractions API is now live in alpha on OpenRouter.

Use it to pull specific details out of text, images, and PDFs by describing the fields you need, like the names in an email or the total on a receipt, and get each one back as structured data.

Read the guide: https://t.co/PUmN7Glclt

[AI 新动静 @aixindongjing]

@antigravity Google Antigravity 宣布浏览器使用功能正在逐步推出，打通研究、编写代码和实时验证的完整流程。

[AI 新动静 @aixindongjing]

Antigravity 为其智能体推出浏览器使用功能

Google Antigravity 宣布浏览器使用功能正在逐步推出，打通研究、编写代码和实时验证的完整流程。用户可选择智能体登录哪些服务，下载会暂停直到用户批准，智能体还可用浏览器同时执行多项任务。

[Google Antigravity @antigravity]

Browser use has been rolling out in Antigravity, closing the loop between researching, writing code, and verifying it live.

You choose what to log into, downloads pause until you approve them, and the agent can use the browser to run multiple tasks at once. Give it a try today.

[AI 新动静 @aixindongjing]

OpenRouter 下架三款 ByteDance Seed 模型

OpenRouter 已不再列出 Seed-2.0-Code、Seed 1.6 和 Seed 1.6 Flash；三款模型的条目到期日均为 2026 年 11 月 11 日，上下文长度均为 262,144 个 token。

#OpenRouter #模型

[AI 新动静 @aixindongjing]

OpenRouter 为 7 款 OpenAI 模型标注到期日

目录现将 GPT-5、GPT-5 Mini、GPT-5 Nano、GPT-5 Pro、GPT-5 Image、o3 和 o3 Pro 的到期日列为 2026 年 12 月 11 日。此前这些条目未标注到期日；目录未说明原因或届时会发生什么。

#OpenRouter #OpenAI #模型

[AI 新动静 @aixindongjing]

VulnCheck：七个月来，Anthropic 漏洞账本显示，37,966 项发现中有 217 项已修复

销售漏洞情报的安全公司 VulnCheck 发布了对 Anthropic Project Glasswing 的 11 月更新。以下是 VulnCheck 的数据及其解读，并非 Anthropic 的说法。

#Anthropic #模型 #编程