VLLM语义路由器团队拥有发布决定3.0，一个多模式决定模型家族。 第3.0号决定的多式联运决定模型读作文本、JSON和图像，然后回答有关它们的问题。 有5个尺寸，从0.8B到27B，都低于Apache-2.0. 对于开发者来说，这是在不解析生成的文本的情况下快速分类，路由和门请求的一种方法。
TL；数据
- 大小：5个型号：d3-立特(0.85B),d3-nano(2.21B),d3-mini(4.54B),d3-flash(8.39B),d3(26.09B). 上下文长度：未披露。
- 运行：在1个AMD Instinct MI325X GPU上测得的耐久性。 BF16重量；没有列出官方的定量变体。
- 性能：27B d3既领导卡片中的文字表和视觉表，又领导内部评价。
- 最佳：97.7 在KIE(CORD+FUNSD)文档提取(d3)上。
- 最差：34.1在R-Bench-M(d3)上。
- 底线：
- 最佳：9B型现在胜过上一代的27B型。
- 最糟糕的是：根据现场数据自报得分。
第3.0号决定是什么？
第3.0号决定是一套来自vLLM语义路由器的开放式决定模型，它返回概率而不是文本。 您通过状态( 文本或 JSON)，可选图像，并命名问题。 每个问题都是一个选择，一个是/否，或者一个比分。 模型在一个呼叫中回答所有问题，并返回每个答案的概率。
这是路由器和护栏的 " 系统 " 模式。 与其刺激LLM并解析其输出，不如通过校准得分来达到阈值。
D3怎样工作？
每个型号都是Qwen基座的微调。 27B d3型机车建于Qwen3.8-27B上。 较小的4个在0.8B、2B、4B和9B的Qwen3.5检查站上。
每个尺寸都带有一个视觉编码器：d3升0.10B,d3-纳诺和d3-mini0.33B,d3-flash和d30.46B. 请求接受几个PNG,JPEG或WebP图像，每个图像的读数最高可达1.6兆像素。 每个问题都能看到所有图像。
每个问题都会通过输入得到自己的前进。 安装目标变压器==5.17.0，可选闪光线-注意包加快线性注意层。 装入需要信任 remote code=True.
第3.0号决定在基准测试上的表现如何？
通过模型卡，研究团队声明Jev决定指数0.3.1，这个板对TypeSafe的Jev决定系统的公开复制进行排名。 d3数字是小组的内部评价。
在文字上，d3分64.1分，领先于Perplexity Counter v1.1(62.8)和Jev(60.1). 一项警告：第27B号火炬决定张贴了较高的公共套期评分(65.1对64.9)。 在视觉板上，d3对Perplexity决策v1.1的成绩为71.6对70.6.
规模故事是最强的部分。 d3-flash(9B)分59.0分，高于第2.0号决定的27B分55.9分。 在公共套房方面，第2.0号决定的收益从+8.0(27B)到+15.1(0.8B)不等。
班级领导不统一，不过：
- d3-mini(4B)和d3-flash(9B)同时领导其文字和视觉表。
- d3-lite(0.8B)在文本上领先，但在视觉上排名第3位(41.1，落后于JPT-0.8B at 42.1)。
- d3-nano(2B)领导它的视觉表，但在文本(35.8 vs 40.1)上跟踪LiquidAI d1-3B.
在公共视觉任务方面，d3在InfographyVQA和Mind2Web上评分为97.3分，两者都是近似重建。 在MMU-Pro的愿景(46.2)和仇恨的Memes节制(44.6)上，它较弱。
模特儿们有多快？
下面的中间点是每次1AMD Instinct MI325X的1个请求，每张卡：
- d3-nano:21.7ms文本，92.0ms带有图像
- d3升：23.5ms文本，60.7ms带有图像
- d3-mini:28.5ms文本，115.0ms带有图像
- d3-flash:29.7ms文本，171.8ms带有图像
- d3:84ms文本，342ms带有图像
所有型号都接受了AMD Instinct MI325X GPU的培训。
d3与其他决定模型有何比较？
关键外卖
- 5 Apache-2.0决定模型，0.8B至27B，均带有图像输入。
- d3(27B)超过其文本(64.1)和视觉(71.6)表，关于内部电val。
- 9B d3-flash(59.0)击败第2.0号决定的27B(55.9).
- 0.8B型和2B型并不是每个板都领先。
- 中间文本延迟运行21.7至84ms在1 MI325X。
请检查access-date=中的日期值 (帮助) Hugging Face收藏，d3型卡和vLLM语义路由器 GitHub repo. 所有功劳都归功于这个项目的研究人员。 并加入我们的通讯。 等等你在电报里吗现在你也可以加入我们的电报。
米哈尔·萨特(Michal Sutter)是一名数据科学专业人员，拥有帕多瓦大学的数据科学硕士学位。 在统计分析、机器学习和数据工程方面有一个坚实的基础，Michal在将复杂的数据集转化为可操作的见解方面表现突出。
本文由 AI图鉴自动抓取并翻译，版权归原作者所有；如需查看原文请点下方链接。
The vLLM Semantic Router team has released Decision 3.0, a family of multimodal decision models. Decision 3.0 multimodal decision models read text, JSON and images, then answer typed questions about them. There are 5 sizes, from 0.8B to 27B, all under Apache-2.0. For developers, this is a fast way to classify, route and gate requests without parsing generated text.
TL;DR
- Size: 5 models: d3-lite (0.85B), d3-nano (2.21B), d3-mini (4.54B), d3-flash (8.39B), d3 (26.09B). Context length: not disclosed.
- Runs on: latency measured on 1 AMD Instinct MI325X GPU. BF16 weights; no official quantized variants listed.
- Performance: the 27B d3 leads both the text and vision tables in its card, but on internal evaluation.
- Best: 97.7 on KIE (CORD+FUNSD) document extraction (d3).
- Worst: 34.1 on R-Bench-M (d3).
- Bottom line:
- Best: a 9B model now beats last generation’s 27B.
- Worst: scores are self-reported against live board data.
What is Decision 3.0?
Decision 3.0 is a set of open decision models from vLLM Semantic Router that return probabilities instead of text. You pass a state (text or JSON), optional images, and named questions. Each question is a Choice, a Yes/No, or a Score on a scale. The model answers all questions in one call and returns a probability for every answer.
This is the ‘system one’ pattern for routers and guardrails. Instead of prompting an LLM and parsing its output, you get calibrated scores you can threshold.
How does d3 work?
Each model is a fine-tune of a Qwen base. The 27B d3 is built on Qwen3.8-27B. The smaller 4 build on Qwen3.5 checkpoints at 0.8B, 2B, 4B and 9B.
Every size carries a vision encoder: 0.10B in d3-lite, 0.33B in d3-nano and d3-mini, and 0.46B in d3-flash and d3. Requests accept several PNG, JPEG or WebP images, each read at up to 1.6 megapixels. Every question sees all images.
Each question gets its own forward pass over the input. Install targets transformers==5.17.0, and the optional flash-linear-attention package speeds up the linear-attention layers. Loading needs trust_remote_code=True.
How does Decision 3.0 perform on benchmarks?
Through the model card, the research team state the Jev Decision Index 0.3.1, a board that ranks open reproductions of TypeSafe’s Jev decision system. d3 numbers are the team’s internal evaluation.
On text, d3 scores 64.1, ahead of Perplexity Decider v1.1 (62.8) and Jev (60.1). One caveat: Torchcast Decision 27B posts a higher public-suite score (65.1 vs 64.9). On the vision board, d3 scores 71.6 against 70.6 for Perplexity Decider v1.1.
The size story is the strongest part. d3-flash (9B) scores 59.0, above Decision 2.0’s 27B at 55.9. On the public suite, gains over Decision 2.0 range from +8.0 (27B) to +15.1 (0.8B).
Class leadership is not uniform, though:
- d3-mini (4B) and d3-flash (9B) lead both their text and vision tables.
- d3-lite (0.8B) leads on text but ranks 3rd on vision (41.1, behind JPT-0.8B at 42.1).
- d3-nano (2B) leads its vision table but trails LiquidAI d1-3B on text (35.8 vs 40.1).
On public vision tasks, d3 scores 97.3 on InfographicVQA and 89.9 on Mind2Web, both approximate rebuilds. It is weaker on MMMU-Pro vision (46.2) and Hateful Memes moderation (44.6).
How fast are the models?
Medians below are for 1 request at a time on 1 AMD Instinct MI325X, per each card:
- d3-nano: 21.7 ms text, 92.0 ms with an image
- d3-lite: 23.5 ms text, 60.7 ms with an image
- d3-mini: 28.5 ms text, 115.0 ms with an image
- d3-flash: 29.7 ms text, 171.8 ms with an image
- d3: 84 ms text, 342 ms with an image
All models were trained on AMD Instinct MI325X GPUs.
How does d3 compare with other decision models?
Key Takeaways
- 5 Apache-2.0 decision models, 0.8B to 27B, all with image input.
- d3 (27B) tops its text (64.1) and vision (71.6) tables, on internal evals.
- The 9B d3-flash (59.0) beats Decision 2.0’s 27B (55.9).
- The 0.8B and 2B models do not lead every board.
- Median text latency runs 21.7 to 84 ms on 1 MI325X.
Check out the Hugging Face collection, the d3 model card and the vLLM Semantic Router GitHub repo. All credit goes to the researcher of this project. Also, feel free to follow us on Twitter and don’t forget to join our 150k+ML SubReddit and Subscribe to our Newsletter. Wait! are you on telegram? now you can join us on telegram as well.
Michal Sutter is a data science professional with a Master of Science in Data Science from the University of Padova. With a solid foundation in statistical analysis, machine learning, and data engineering, Michal excels at transforming complex datasets into actionable insights.
Auto-fetched and translated by AI Atlas. All rights belong to the original author.