Clover-35B-A3B-Preview：让智能，多一种生长方式
非常之事，何得循旧。
今天，我们正式推出DiffuSpace Clover（三叶）的第一个版本 35B-A3B-Preview。
2025 年 4 月，香港大学自然语言处理实验室（HKU NLP）推出Dream 7B（https://hkunlp.github.io/blog/2025/dream/），至今Dream系列模型累计在Hugging Face下载量270万+次。从Dream出发，扩散智能 DiffuSpace 成为团队成员新的起点，但开源普惠的初衷一直没有改变。
如果说 Dream 7B 证明了扩散语言模型能够与自回归模型正面交锋，那么 Clover 要回答的是一个更根本的问题：当参数走向更大规模，扩散语言模型能否依然保持竞争力，同时守住全局生成的速度优势？
Clover-35B-A3B-Preview 给出了肯定答案：在更大规模上，扩散语言模型甚至能比同尺寸自回归模型略胜一筹，同时保持更快生成速度。高速生成与高性能，可以共存于同一模型。
Clover-35B-A3B-Preview 是一个采用混合专家架构（MoE）的扩散语言模型，总参数 350 亿、激活参数 30 亿，支持思考模式。在多项基准测试中，它的性能超过同规模自回归模型 Qwen3.6-35B-A3B，且生成速度显著更快。
Clover-35B-A3B-Preview 在智能体编程与工具使用任务中表现尤为突出。在本次对比的六项 Agent 评测中，它均取得了所列模型中的最高分。其中，SWE-bench Verified 达到 73.6，Terminal-Bench 2.1 达到 51.7，τ³ Overall 达到 65.3，超过了 同为扩散语言模型的 LLaDA2.2 flash、Mercury 2.5，以及同尺寸的自回归模型 Qwen3.6-35B-A3B 和 Qwen3.5-35B-A3B。
Clover-35B-A3B-Preview 的通用能力同样扎实。在指令遵循、知识、逻辑推理和长文本理解等多项评测中，其成绩达到或超过同规模自回归模型，GPQA Diamond、ZebraLogic 和 AA-LCR 分别达到 83.8、95.8 和 72.0。
在推理效率方面，Clover-35B-A3B-Preview 在单张 NVIDIA GB300 GPU 上，以 BF16 精度、batch size = 1 进行测试，在内部Agent 场景数据中实现了 680 tokens/s 的整体生成速度。 即使与采用 DFlash2 加速的同规模 Qwen 模型相比，Clover 仍展现出更优的性能与推理速度的平衡。此外，在单张 NVIDIA H200 GPU 上，Clover 在 Agent 任务中达到 515 tokens/s，高于 DFlash2 方案的 365 tokens/s。以上结果都体现了扩散架构在复杂 Agent 场景下的速度优势。
在 Clover-35B-A3B-Preview 的基础上，我们将继续推进扩散语言模型的规模化研究，提升模型能力与推理速度，并在更广泛的真实任务中验证和改进模型。
Clover-35B-A3B-Preview 现已开源，采用 Apache 许可。模型权重与相关代码已在 Hugging Face 和 GitHub 发布，支持 SGLang 和 llama.cpp 推理。开发者可下载部署、构建应用，并在 Clover 的基础上开展进一步研究。我们期待与社区一起，拓展扩散语言模型的能力边界。
相关链接：
HuggingFace：https://huggingface.co/Diffuspace/Clover-35B-A3B-Preview
GitHub：https://github.com/DiffuSpace/Clover-35B-A3B-Preview
ModelScope：https://modelscope.cn/models/Diffuspace/Clover-35B-A3B-Preview
申请在线体验：https://chat.diffuspace.ai/
扩散，向更大的可能
支撑这次跃迁的，不只是参数规模，还有我们围绕块扩散（Block Diffusion）对门控 Delta 网络（Gated DeltaNet，GDN）的结构性重构。
我们重新设计了双向门控 Delta 网络（Bidirectional GDN），让历史记忆与块内去噪各自采用适合的计算方式：已生成的上下文通过 GDN 的递归状态（recurrent state）高效压缩，当前块内的 token 则通过双向 Softmax 注意力（bidirectional softmax attention）进行细粒度交互。关键之处在于，我们把历史记忆的读出作为一个独立的信息单元（information unit），与块内 token 放在同一次 Softmax 中共同参与信息选择，让模型自己学习何时依赖历史、何时利用当前块。我们还引入可学习的块大小校准（learnable block-size calibration），补偿块内 token 增多时对历史权重的天然稀释。在局部特征提取上，我们为右侧上下文设置独立、零初始化的卷积分支（zero-initialized convolution branch），让模型从预训练的因果卷积（causal convolution）行为出发，直接学习左右方向各自的特征。
由此，紧凑的长程记忆、具有选择性的块内交互与双向局部感知被结合起来，使 GDN 的结构更贴合块级并行去噪（block-level parallel denoising）。Clover-35B-A3B-Preview 的高速与高性能，正来自这些结构层面的针对性设计。
Agentic coding
Clover-35B-A3B-Preview 在 Agentic Coding 能力上不止是代码补全，它可以将自然语言需求直接转化为完整、可交互的应用，也能在真实浏览器环境中理解页面并连续完成多步操作。
在智能体中，速度本身就是一种核心能力。智能体需要在决策、执行、反馈与调整之间持续循环，每一次生成延迟都会沿着任务链路不断累积。Clover 更快的生成速度缩短了这条反馈回路：开发者可以更早看到应用预览，智能体也能更及时地根据执行结果调整策略、推进下一步。当每一步都更快，智能体就能在同样的时间里探索更多路径，并更早抵达结果。
端侧应用
端侧要的不是“慢思考”，而是快而全局的判断。
在端侧部署方面，Clover 已成功在仅配备 16GB 显存的 NVIDIA RTX 5070 Ti 上实现 INT4 量化部署。在 MATH-500 测试中，Clover 的生成速度达到 101.8 tokens/s，相比同规模 MiniCPM、Qwen 等模型的对照测试结果（48.5 tokens/s）表现出显著的效率优势。团队将持续优化推理引擎与端侧部署方案，进一步释放扩散语言模型在资源受限设备上的性能潜力。
性能上，与 MiniCPM-V 4.7（35B-A3B）、LFM2.5（24B-A2B）、Nemotron 3.5 Lightning（30B-A3B）和 Gemma 4（26B-A4B）等代表性轻量级、高效推理或端侧部署模型的对比中，Clover-35B-A3B-Preview 在九项评测中的六项取得最高分，优势覆盖知识推理、工具使用与智能体编程。其中，Terminal-Bench 2.1 达到 51.7，SWE-bench Verified 达到 73.6，分别比同为 35B-A3B 的 MiniCPM-V 4.7 高出 15.7 和 28.0 分；BFCL-v4 和 τ³ Banking 分别达到 67.9 和 14.4；GPQA Diamond 和 MMLU-Pro 分别达到 83.8 和 84.8。
Demo展示
路径规划
在此案例中用户要查询“从匹兹堡 CMU 到最近的 Wendy’s 需要开车多久”，但地图网站的搜索不带位置信号，直接搜索店名会跳到缅因州 Portland。Clover 识别出结果偏离目标，逐步加入城市、街区和邮编改写检索，并在页面卡住后两次重置、重填路线，最终找到目标门店，查得驾车距离 2.0 公里、车程约 3 分钟。
本次对照中，Clover-35B-A3B-Preview 用 28 步、169.9 秒完成任务；Qwen3.6 Flash 用满 30 步、耗时 578.3 秒，仍停留在距起点 347 公里、车程约 4.5 小时的门店，未通过任务评判。
WebMCP：连续完成六项业务任务
通过网页提供的工具，模型连续完成开票、课程注册、金额汇总、事件状态与客户信息查询等任务。演示将多个业务页面放在同一视图中，对照工具调用、页面操作和完成结果，两方均通过所选的六项任务。
在完整的 49 项任务、每项 3 次尝试的评测中，Clover 的 Agent 执行耗时中位数为 1.52 秒，Gemini3.6 Flash 为 8.86 秒。
创建私有项目并邀请成员
根据一条指令，模型需要创建名为 planner 的私有项目，并添加两名指定成员。这要求它连续完成项目创建、可见性设置、成员查找与邀请，始终保持对用户目标的把握。
在相同的智能体框架、提示词、工具和浏览器配置下，本次运行中 Clover-35B-A3B-Preview 用 13 步、169.9 秒完成任务，Qwen3.6 Flash 用 21 步、392.3 秒完成。
GenUI：用一个 HTML 文件构建交互式桌面
该案例中用户要求用单个 HTML 文件模拟一个桌面环境，支持窗口拖动，并能打开多个小应用。Clover 根据这条需求，生成了包含程序坞、窗口系统以及记事本、计算器、时钟、天气和设置等应用的模拟桌面。
在演示中，用户可以打开应用、输入文本、操作计算器并调整设置。这一任务同时考察了界面布局、交互逻辑与多个功能模块的代码组织能力，展示了从自然语言需求到可交互应用的生成过程。
无人机调度：复合报文的并行解析
在城市低空无人机调度演示中，模型通常需要将复合空管报文解析为禁飞区解除、航点修改、任务分配、禁飞区半径调整和紧急调度等多项任务。Clover在0.97秒后同时完成处理呈现五项正确结果 。Qwen在3.78S展示结果，其中将UAV5的测绘任务误解析为医疗任务。这一案例直观展示了扩散并行生成在复合指令解析中的速度以及任务意图理解的优势。
生态伙伴
SigmaZ 是我们在 GenUI 方向的独家生态伙伴。他们把 Clover 的能力带进教育场景，让知识不再只是静态图文，而是可操作、可互动的画面。目前，SigmaZ 已上线教育场景内测 Demo。接下来，我们正着手进行联合后训练，计划把 token 吞吐量提升到 2000 tokens 每秒，把实时视觉编码应用到更加通用的实时交互场景。
3D 科普：走进细胞的能量工厂
围绕“细胞如何产生能量”，模型构建了一个可交互的 3D 科普页面。页面以剖面形式展示动物细胞。用户可以点击细胞器拉近查看说明，也能切换结构标签、剖面和旋转视角，从细胞整体逐步深入到局部。该案例把知识讲解、三维场景与交互逻辑放在同一个页面里，让“葡萄糖如何变成细胞能量”不再只是文字描述，而是可以通过观察和操作来理解的过程。
交互式天文科普：让日月食的成因看得见
围绕“为什么日月食并不每月发生”，模型构建了一个随时间变化的太阳—地球—月球 3D 演示页面，将天体位置、月球轨道倾角与月相变化放进同一个场景，让用户直观看到日月食所需的特殊排列条件。
用户可以拖动时间轴，观察三者的相对运动，也可以切换轨道、标签与阴影锥的显示，或跳转到日食发生时刻，查看月球阴影如何落到地球上。页面还标注了距离压缩与尺寸调整，兼顾演示的可读性和科学概念的准确表达。
决策模型
我们同时推出 Clover-35B-A3B-Preview 的决策模型，面向快速、结构化的判断任务，将在后续发布。
模型支持两种作答方式：
独立模式：逐题调用，每次只回答一个问题，适合孤立或常规的快速判断；
联合模式：将同一份材料下的多个问题放在一起，在同一次去噪过程中并行生成答案（每次最多 8 个）。
和TypeSafe Jev或其他开源决策模型不同的是，相比于单点处理，联合模式的优势在于“全局统筹”：它不仅大幅减少了多次交互的时间，更能让多个判断在生成过程中相互参照、避免逻辑冲突。一次处理多个关联决策，不仅更快，还更准。
在包含 500 组、2420 道关联问题的内部验证中，联合模式答对了 95.0% 的问题，并在 80.4% 的问题组中全部答对。相比之下，同一模型独立作答时，两项成绩分别为 89.8% 和 62.4%；TypeSafe 的决策模型 Jev 在相同题目的历史评测中分别为 63.6% 和 15.8%。在本次验证中，把关联问题放在一起处理，同时提高了单题准确率与整组答案的正确率。
同一份合同，五个关联判断
同一份合同中的“通知是否及时”和“保证金是否应保留”存在规则上的依赖。逐题独立判断时，模型可能一边认为通知超时，一边又选择只有通知及时才允许的“保留保证金”。联合模式让这两个判断一起生成，有助于减少前后矛盾，也减少了重复读取材料和逐题调用的开销。
在该验证案例中，联合模式的 5 项判断全部正确，独立模式答对 4 项，Jev 答对 2 项。演示对照了通知时限、保证金保留等判断，展示联合模式如何处理相互依赖的规则。
中国象棋：根据棋局连续决策
决策模型读取当前棋盘状态，选择下一步，并输出结构化落子动作。在这局与TypeSafe Jev 的对弈中，DiffuSpace 的决策模型执黑获胜。
下一步
接下来，我们将持续推进扩散语言模型的规模化，并拓展多模态能力；同时提升长程任务中的稳定性，优化推理效率，推进量化与端侧适配。
同时，我们也在开发智能体运行框架（harness）与编程客户端（desktop），将需求输入、工具调用、任务执行和结果查看连接起来。相关工具仍在开发中，将在近期上线。
这些工作离不开开源社区打下的基础。感谢 Qwen 开放的模型架构与训练生态，让 Clover-35B-A3B-Preview 的研究有了可靠起点；感谢 SGLang 社区在块扩散推理路径上的工程支持；也感谢 llama.cpp 社区开放的适配接口与协作流程。
欢迎大家前往 diffuspace.ai 申请内测体验 Clover-35B-A3B-Preview。
同时，开发者可在 Hugging Face 与 ModelScope 下载模型权重、构建应用，或在 GitHub 提交 Issue、参与适配。来自真实任务的反馈，将帮助我们持续优化。我们期待与开源社区一起共建，也相信扩散模型会因此变得更好。
我们也欢迎各类场景的生态合作方加入，一起把扩散语言模型带进真实任务，共同验证、共同打磨。
Clover（三叶草）代表幸运，也承载着我们对扩散模型的期待。
今天，Clover 破土而出，向智能的旷野蔓延生长。
Clover-35B-A3B-Preview: A New Way for Intelligence
Extraordinary things are never achieved by following conventional paths.
Today, we are officially releasing Clover-35B-A3B-Preview, the first version of DiffuSpace Clover (三叶).
In April 2025, the NLP Group at The University of Hong Kong released Dream 7B. To date, the Dream family of models has accumulated more than 2.7 million downloads on Hugging Face. Building on Dream, members of the team began a new chapter at DiffuSpace, while remaining committed to making AI broadly accessible through open source.
If Dream 7B proved that diffusion language models can go toe-to-toe with autoregressive models, Clover sets out to answer a more fundamental question: When scaled up, can diffusion language models maintain their competitiveness while preserving the speed advantage of global generation?
Clover-35B-A3B-Preview provides a clear answer: yes. At a larger scale, diffusion language models can even edge out similarly sized autoregressive models while delivering significantly faster generation. Speed and performance can coexist in a single model.
Clover-35B-A3B-Preview is a Mixture-of-Experts (MoE) diffusion language model with 35 billion total parameters and 3 billion activated parameters, with support for a thinking mode. Across multiple benchmarks, it outperforms the similarly sized autoregressive model Qwen3.6-35B-A3B, while generating significantly faster.
Clover-35B-A3B-Preview particularly excels in agentic coding and tool-use tasks. Across all six agent benchmarks in our comparison, it achieves the highest scores among the evaluated models. Notably, it scores 73.6 on SWE-bench Verified, 51.7 on Terminal-Bench 2.1, and 65.3 on τ³ Overall, outperforming fellow diffusion language models LLaDA2.2 flash and Mercury 2.5, as well as similarly sized autoregressive models Qwen3.6-35B-A3B and Qwen3.5-35B-A3B.
Clover-35B-A3B-Preview also delivers strong general-purpose capabilities. Across benchmarks covering instruction following, knowledge, logical reasoning, and long-context understanding, it matches or exceeds the performance of similarly sized autoregressive models, achieving 83.8 on GPQA Diamond, 95.8 on ZebraLogic, and 72.0 on AA-LCR.
In terms of inference efficiency, Clover-35B-A3B-Preview achieves an overall generation speed of 680 tokens/s on internal agent workloads, tested on a single NVIDIA GB300 GPU with BF16 precision and a batch size of 1. Even compared with a similarly sized Qwen model accelerated by DFlash2, Clover offers a better balance between performance and inference speed. On a single NVIDIA H200 GPU, Clover also reaches 515 tokens/s on agent tasks, compared with 365 tokens/s for the DFlash2 setup. These results demonstrate the speed advantage of the diffusion architecture in complex agent workloads.
Building on Clover-35B-A3B-Preview, we will continue to advance the scaling of diffusion language models, improve model capabilities and inference speed, and validate and refine their performance across a broader range of real-world tasks.
Clover-35B-A3B-Preview is now open source under an Apache license. Model weights and related code are available on Hugging Face and GitHub, with inference support for SGLang and llama.cpp. Developers can download and deploy the model, build applications, and conduct further research on top of Clover. We look forward to working with the community to push the boundaries of what diffusion language models can achieve.
Related Links:
HuggingFace：https://huggingface.co/Diffuspace/Clover-35B-A3B-Preview
GitHub：https://github.com/DiffuSpace/Clover-35B-A3B-Preview
ModelScope：https://modelscope.cn/models/Diffuspace/Clover-35B-A3B-Preview
Request Online Access：https://chat.diffuspace.ai/
Diffusion: Pushing Toward Greater Possibilities
The leap forward is driven by more than parameter scaling. It also comes from our architectural redesign of the Gated DeltaNet (GDN) for Block Diffusion.
We redesigned the Bidirectional Gated DeltaNet to handle historical memory and intra-block denoising using different computational mechanisms. Previously generated context is efficiently compressed into GDN's recurrent state, while tokens within the current block interact through fine-grained bidirectional softmax attention.
Crucially, we treat the readout from historical memory as a separate information unit that participates in the same softmax operation as the tokens within the current block. This allows the model to learn when to rely on historical context and when to draw on information within the current block. We also introduce learnable block-size calibration to compensate for the natural dilution of attention to historical memory as the number of tokens within a block increases. For local feature extraction, we add a separate, zero-initialized convolution branch for the right-side context. Starting from the pretrained causal convolution behavior, the model can directly learn distinct features from both left and right contexts.
Together, these designs integrate compact long-range memory, selective intra-block interactions, and bidirectional local context modeling, making the GDN architecture better suited to block-level parallel denoising.
The speed and performance of Clover-35B-A3B-Preview stem directly from these targeted architectural innovations.
Agentic Coding
Clover-35B-A3B-Preview goes beyond code completion. It can translate natural-language requirements directly into complete, interactive applications, while also understanding webpages and executing multi-step operations in real browser environments.
For AI agents, speed is itself a core capability. Agents operate through continuous cycles of decision-making, execution, feedback, and adjustment. Every generation delay accumulates across the task execution pipeline.
Clover's faster generation speeds shorten this feedback loop. Developers can see application previews sooner, while agents can respond to execution results more quickly, adapt their strategies, and move on to the next step.
When every step becomes faster, agents can explore more possible paths within the same amount of time and reach results sooner.
On-Device Applications
Edge computing demands rapid, holistic judgment rather than "slow thinking."
For on-device deployment, Clover has been successfully deployed with INT4 quantization on an NVIDIA RTX 5070 Ti with just 16GB of VRAM. On MATH-500, Clover achieves a generation speed of 101.8 tokens/s, demonstrating a substantial efficiency advantage over the 48.5 tokens/s recorded in comparison tests with similarly sized MiniCPM and Qwen models. We will continue optimizing the inference engine and on-device deployment setup to further unlock the performance of diffusion language models on resource-constrained devices.
We compared Clover-35B-A3B-Preview against representative lightweight, inference-efficient, and on-device-deployable models, including MiniCPM-V 4.7 (35B-A3B), LFM2.5 (24B-A2B), Nemotron 3.5 Lightning (30B-A3B), and Gemma 4 (26B-A4B).
Clover achieved the highest scores on six out of nine benchmarks, demonstrating advantages across knowledge reasoning, tool use, and agentic coding.
Notably, Clover scores 51.7 on Terminal-Bench 2.1 and 73.6 on SWE-bench Verified, outperforming MiniCPM-V 4.7, which shares the same 35B-A3B parameter configuration, by 15.7 and 28.0 points, respectively.
For tool use, Clover achieves 67.9 on BFCL-v4 and 14.4 on τ³ Banking. For knowledge reasoning, it scores 83.8 on GPQA Diamond and 84.8 on MMLU-Pro, leading all other models included in the comparison on these benchmarks.
Demo showcases
Route Planning
In this example, the user asks"I am at CMU Pittsburgh，how long it takes to drive to the neareat Wendy's?"
However, mapping websites lack spatial location signals, meaning a direct search for the store name incorrectly defaults to Portland, Maine. Clover identifies that the result deviates from the target, progressively refining the search by adding city, neighborhood, and zip code keywords. When the page stalls, it resets and refills the route twice, eventually locating the target store and successfully retrieving a driving distance of 2.0 km and a travel time of roughly 3 minutes.
In this comparison, Clover-35B-A3B-Preview completed the task in 28 steps over 169.9 seconds. Qwen3.6 Flash exhausted all 30 steps and took 578.3 seconds, yet remained stuck at a store 347 km away (approximately 4.5 hours of driving time from the starting point), failing the evaluation criteria.
WebMCP: Completing Six Business Tasks in Sequence
Using tools exposed through webpages, the model completes a sequence of business tasks, including invoice creation, course registration, amount aggregation, event status checks, and customer information retrieval.
The demonstration places multiple business webpages within a single view, comparing tool calls, webpage interactions, and task outcomes. Both models successfully complete the six selected tasks.
In the full evaluation covering 49 tasks, with three attempts per task, Clover achieves a median agent execution time of 1.52 seconds, compared with 8.86 seconds for Gemini3.6 Flash.
Creating a Private Project and Inviting Members
Based on a single prompt, the model is required to create a private project named "planner" and add two designated members. This requires sequential execution of project creation, visibility settings, member searching, and invitations, while consistently maintaining alignment with the user's objective.
Under identical agent frameworks, prompts, tools, and browser configurations, Clover-35B-A3B-Preview completed the task in 13 steps and 169.9 seconds, whereas Qwen3.6 Flash took 21 steps and 392.3 seconds.
GenUI: Building an Interactive Desktop in a Single HTML File
In this example, the user asks the model to create a simulated desktop environment using a single HTML file, complete with draggable windows and multiple functional applications.
Clover generates an interactive desktop featuring a dock, a window management system, and applications including a notepad, calculator, clock, weather app, and settings panel.
In the demo, users can open apps, input text, operate the calculator, and adjust settings. This task rigorously tests interface layout, interactive logic, and code organization across multiple functional modules, showcasing the generation pipeline from natural language requirements to fully interactive applications.
UAV Dispatch: Parsing Compound ATC Messages in Parallel
In this urban low-altitude UAV dispatch demo, the model parses a compound air traffic control (ATC) message into multiple tasks, including lifting no-fly restrictions, modifying waypoints, assigning missions, adjusting no-fly zone radii, and emergency dispatch. Clover produces all five correct results together in 0.97 seconds. Qwen presents its results after 3.78 seconds, but incorrectly interprets UAV5's mapping mission as a medical mission. This case demonstrates the advantages of diffusion-based parallel generation in both response speed and task intent understanding when parsing compound instructions.
Ecosystem Partner
SigmaZ is our exclusive ecosystem partner for GenUI. They are bringing Clover into education, transforming static text and images into visual experiences that users can explore and interact with. SigmaZ has launched a closed beta demo for educational use. Apply for access at edu.tap8.ai. We are now beginning joint post-training, aiming to increase token throughput to 2,000 tokens per second and extend real-time visual coding to a broader range of interactive applications.
3D Science Exploration: Inside the Cell's Energy Factory
To explain how cells produce energy, the model builds an interactive 3D explainer with a cutaway view of an animal cell. Users can click organelles to move closer and read explanations, or toggle structural labels, the cutaway view, and rotation to explore the cell from the whole down to its individual components. By bringing knowledge explanations, a 3D scene, and interactive logic together on one page, the demo turns the conversion of glucose into cellular energy from a written description into a process users can understand through observation and interaction.
Interactive Astronomy: Visualizing How Eclipses Happen
To explain why solar and lunar eclipses do not occur every month, the model builds a time-driven 3D demonstration of the Sun, Earth, and Moon. It brings celestial positions, the tilt of the Moon's orbit, and changing lunar phases into a single scene, making the special alignments required for eclipses visible.
Users can drag a timeline to observe the bodies' relative motion, toggle orbits, labels, and shadow cones, or jump to a solar eclipse to see how the Moon's shadow falls on Earth. The page also notes where distances and sizes have been adjusted, balancing visual clarity with an accurate presentation of the underlying scientific concepts.
Decision Model
We are also introducing a decision model based on Clover-35B-A3B-Preview for fast, structured judgment tasks. It will be released at a later date.
The model supports two response modes:
Independent Mode: The model is called separately for each question, answering one question at a time. This mode is suited to standalone or routine judgments that require a fast response.
Joint Mode: Multiple questions about the same material are grouped together, and their answers are generated in parallel within a single denoising process, with up to eight answers per call.
What sets this model apart from TypeSafe Jev and other open-source decision models is Joint Mode's coordinated handling of related decisions. Rather than treating each decision in isolation, it substantially reduces the time spent on repeated interactions and allows judgments to inform one another during generation, helping avoid logical conflicts. Making multiple related decisions together is not only faster, but also more accurate.
In an internal validation set containing 500 groups and 2,420 related questions, Joint Mode answered 95.0% of individual questions correctly and got every answer right in 80.4% of question groups. In comparison, the same model in Independent Mode scored 89.8% and 62.4%, respectively. TypeSafe's decision model Jev scored 63.6% and 15.8% in historical evaluations on the same questions. In this validation set, processing related questions together improved both individual-question accuracy and the share of groups with all answers correct.
The Same Contract, Five Correlated Judgments
Rules within the same contract regarding "whether a notice was timely" and "whether a security deposit should be retained" share a logical dependency. When judged independently question by question, a model might simultaneously conclude that a notice was overdue while selecting the "retain deposit" option, which is only permissible if the notice is timely. Joint Mode enables these judgments to be generated concurrently, helping to minimize inconsistencies while reducing the overhead of repeatedly reading material and executing sequential calls.
In this validation case, Joint Mode answered all 5 judgments correctly, while Independent Mode answered 4 correctly, and Jev answered 2 correctly. The demo compares judgments on notice deadlines and deposit retention, demonstrating how Joint Mode handles mutually dependent rules.
Chinese Chess: Making Sequential Decisions from the Board State
The decision model reads the current board state, selects its next move, and outputs a structured move action. In this game against TypeSafe Jev, DiffuSpace's decision model played Black and won.
Next Steps
Moving forward, we will continue pushing the boundaries of diffusion language model scaling, expanding multimodal capabilities, enhancing stability in long-horizon tasks, optimizing inference efficiency, and advancing quantization and edge-device adaptation.
At the same time, we are developing an agent execution harness and a desktop coding client to bring natural-language instructions, tool calls, task execution, and result inspection into a more integrated workflow. These tools are currently under development and will be rolled out shortly.
None of this progress would be possible without the foundations laid by the open-source community. We thank Qwen for its open model architectures and training ecosystem, which provided a reliable starting point for the research behind Clover-35B-A3B-Preview. We also thank the SGLang community for its engineering support for block diffusion inference, and the llama.cpp community for its open integration interfaces and collaborative development process.
We invite you to visit diffuspace.ai to apply for access to Clover-35B-A3B-Preview.
Developers can also download model weights from Hugging Face and ModelScope, build applications, or contribute through GitHub Issues and integration efforts. Feedback from real-world tasks will help us continuously improve Clover. We look forward to building with the open-source community and believe this collaboration will make diffusion models better.
We also welcome ecosystem partners across a range of use cases to bring diffusion language models into real-world tasks, working together to validate and refine them.
"Clover" stands for good luck, while carrying our highest aspirations for diffusion models.
Today, Clover breaks new ground, ready to grow across the vast frontier of intelligence.