AITOPAI热报
ModelsMulti-source

Google 发布首个决策模型 Gemini Decision 1，上线首日即登陆 OpenRouter

OpenRouter@OpenRouterAI
Post· 3 sources
View the post
Why we picked it

Google 出了个专做选择题的模型，给个 yes/no、255 选 1 或打分题，直接回你概率。输入 $0.10/M、输出免费，做内容审核、自动打标这种批量判断任务能省不少钱。做流水线的开发者可以试试。

Google 推出旗下首个决策模型 Gemini Decision 1，现已通过 OpenRouter 提供服务，定价为输入 $0.10/M tokens，输出免费。该模型支持文本、图片、音频、视频及 PDF 输入。

Image · OpenRouter
Body

技术上，Gemini Decision 1 专为结构化决策任务设计：面对是非题、最多 255 个选项的多选题或评分题，它不输出长文本，而是直接给出答案并附带概率。模型上下文窗口为 262K tokens，可通过 @GoogleAIStudio 调用。

这类专用模型瞄准的是分类、审核、打标等高频判断场景，与通用对话模型形成分工，$0.10/M 的输入定价也明显低于主流通用模型的成本区间，适合大批量流水线任务接入。

Original · 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. openrouter.ai/google/gemini-… Models on OpenRouter @ModelsOnOR Gemini Decision 1 from @GoogleAIStudio is live on OpenRouter! Google's first decision model. Hand it text, images, audio, video, or a PDF plus typed questions, and it answers yes/no, picks from up to 255 options, or scores, each with a probability. 262K context. 🔗 View Quoted Tweet📊 1 ⚡

Industry Coordinates & Source Verification

This report is categorized under AITOP's "Models" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows and US-China foundation model benchmarks and commercialization comparison. Key entities and technical topics under continuous tracking include Gemini Decision 1, Google, OpenRouter, 推理模型.

Source attribution: curated from OpenRouter (@OpenRouterAI). Cross-source corroboration: 3 independent sources have covered this development (including Artificial Analysis, TestingCatalog), confirming industry-wide signal strength.

#Gemini Decision 1#Google#OpenRouter#推理模型

Keep reading

Recent stories with the same tag or section
ModelsUS vs China76°02:20
Chinese contentGoogle 发布首个决策模型 Gemini Decision 1，Decision Index 得分 66

Google 第一次做决策模型，知识推理类基准领先，但做文档分类抽取时准确率明显不如 Drex Max v1.1，价格还是人家的三倍多。如果你在选决策类模型做路由或抽取，这份基准数据直接帮你算清性价比，采购前值得对照看。

#Gemini Decision 1#Google DeepMind#Artificial Analysis#Drex Max
Event
Official accountArtificial Analysis@ArtificialAnlysSource
ModelsUS vs ChinaMulti-source83°02:41
Chinese contentClaude Fable 5.2 (Max) 登顶 Agent Arena，净改进领先第二名 30.4%

Anthropic 这版把 Agent Arena 的分数直接拉开了一倍多差距，Code 和 Work 也都冲到第一，API 还降价 20%。做智能体和编程工作流的朋友可以试试，尤其在意可控性的场景；预算紧的话 GPT-6.1 Terra 每任务不到 1 美元，性价比更合适。

#Claude#AnthropicAI#Agent Arena#GPT-6.1
Event
lmarena.ai@lmarena_aiSource
ModelsUS vs ChinaSelected02:20
Google DeepMind releases AlphaGenome 2 for drug target discovery

Benchmark performance nearly doubled against its predecessor, which matters more for drug discovery teams than raw model size comparisons. Genomics researchers and biotech teams can use variant-to-cell-type predictions to prioritise which gene targets to validate experimentally. Compared with frontier language model releases dominating headlines, this shows AI value concentrating in scientific tooling, a direction enterprise R&D leaders evaluating AI investments should watch closely.

#AlphaGenome#Google DeepMind#drug discovery#genomics
kimmonismus@kimmonismusSource
ModelsBuildersSelected02:11
ElevenLabs ships Scribe v3, Eleven v4 and Eleven Realtime models

The non-obvious point is bundling: one vendor now covers the whole voice-agent pipeline, which changes procurement and integration choices for teams that previously stitched together separate transcription and TTS providers. Developers building call centres, voice assistants or conversational products can benchmark all three layers against alternatives like OpenAI's Realtime API or Deepgram. Pricing and regional availability are not stated in the announcement, so check the platform docs before committing.

#ElevenLabs#Scribe v3#Eleven v4#Eleven Realtime
Official accountElevenLabs@elevenlabsioSource
ModelsBuildersSelected01:48
Google DeepMind releases AlphaGenome 2 DNA variant model

Cell-type-specific variant prediction is what genetics labs have long needed to move from genome-wide association hits to mechanistic explanations, so the practical value lies in what the API actually returns for a given variant and tissue. Genomics researchers, biotech teams and computational biologists can test it against existing variant-effect tools like Ensembl VEP or Sei on their own datasets. Availability is limited to academic research for now, which commercial users in Singapore's biomedical sector and elsewhere will need to track.

#AlphaGenome#Google DeepMind#genomics#AI safety
Official accountDemis Hassabis@demishassabisSource