Reddit用户实测Strata引擎运行Qwen3.8-Flash-Next
Horsey challenge | Strata Fork | Qwen 3.8 FN NVFP4
提供了一手的高性能本地Agent工程实践，详细列出了Strata引擎与Qwen混合MoE模型的量化配置、内存管理及性能指标，对追求极致性价比的本地部署玩家极具参考价值。
So I wanted to try the horsey challenge myself on the latest engine i been testing which is a strata fork by sergqwer. https://github.com/sergqwer/strata-nvfp4. I seen a post earlier and seen i believe it was a iq3 i think and it was pretty dope so i thought i try out the NVFP4. I wasn't expecting to share but i actually that it did pretty darn good with its current time it took total.
所以我想亲自试试马术挑战，使用的是我最近测试的最新引擎，即由 sergqwer 开发的 Strata 分支。https://github.com/sergqwer/strata-nvfp4。我之前看到过一篇帖子，记得好像是 IQ3，我觉得挺酷的，所以我就想试试 NVFP4。我本来没打算分享，但它实际表现相当不错，总耗时很短。
I did change the prompt because this was the second try because the first one got botched because Qwen wanted to leave like 15+ headless windows open when testing the html. I scrolled to another window while it was working and realized my window was lagging lmao. That test i ended it like 45 min in. Which then i retried with new prompt and this time it was much faster.
我修改了提示词，因为这是第二次尝试，第一次搞砸了，因为在测试 HTML 时 Qwen 想要打开 15 多个无头窗口。我在它运行时滚动到另一个窗口，发现我的窗口卡顿了，哈哈。那次测试我在大约 45 分钟时终止了它。然后我用新的提示词重试了这次，速度要快得多。
I had this fork up and running a few days now. I had to tweak some settings for a higher compaction i believe at the 220k-230k mark. It didn't need compaction this time as it was under the limit but when I use it for long runs projects using unity headless, Blender headless and some other things like image gen for texture i will be compacting a lot. Its only been a couple days but so far its been fairly great to anything I have used prior. That goes for the engine and the current model i'm using.
这个分支我已经运行几天了。我认为在 220k-230k 左右我需要调整一些设置以提高压缩率。这次不需要压缩，因为它低于限制，但当我用它进行长时间运行的项目（如使用 Unity 无头模式、Blender 无头模式以及一些用于纹理生成的图像生成等）时，我会进行大量压缩。虽然才过了几天，但到目前为止，它比我之前使用的任何东西都要好得多。这适用于引擎和我当前使用的模型。
This is just to share and contribute to how far we come to accessing models on single gpu and being able to have a chance at some of these things. Maybe it'll motivate more people to show their work a bit more often. I know this sub has motivated me a lot to start sharing and also appreciate all that has helped to contribute for anything Local.
这只是为了分享并贡献于我们如何在单 GPU 上访问模型并取得进展。也许这会激励更多人更频繁地展示他们的工作。我知道这个子版块极大地激励我开始分享，并且我非常感谢所有为 Local 相关事物做出贡献的人。
Thank you to the original Strata Team, Sergqwer for the fork, windowsxp811203 for your model I converted over to work with Strata, and the Qwen Team for the amazing drops.
感谢原始的 Strata 团队，Sergqwer 的分支，windowsxp811203 为我转换的兼容 Strata 的模型，以及 Qwen 团队带来的精彩发布。
*The charts were organized by Ai only, it was easier to have it sort it nicer and clean it.
*图表仅由 AI 整理，让它排序和清理起来更容易且更整洁。
Also I did a prtsc video capture if you ask about quality of vid. Prompt here also that i had to edit myself lol.
另外，我还做了一个截图视频录制，如果你问起视频质量的话。这里也有我必须自己编辑的提示词，哈哈。
"Create a low-poly 3D horse running in a seamless loop using Three.js. Build it procedurally from primitive geometry, animate the legs, body, head, and tail, and return a single self-contained HTML file. Also when you use headless chrome make sure to kill your process that you don't need anymore. we dont need all these headless chromes lagging our system when we don't need it"
"使用 Three.js 创建一个无缝循环的低多边形 3D 奔跑马匹。从基本几何体程序化构建它，动画腿部、身体、头部和尾巴，并返回一个单一的自包含 HTML 文件。另外，当你使用无头 Chrome 时，请确保终止你不再需要的进程。我们不需要所有这些无头 Chrome 在我们不需要的时候拖慢系统"
The stack
技术栈
Hardware: RTX 5090 32GB · Ryzen 7 9800X3D · 192GB DDR5-5600 · Windows 11
硬件：RTX 5090 32GB · Ryzen 7 9800X3D · 192GB DDR5-5600 · Windows 11
Engine: strata-nvfp4 (fork of Niko1221/Strata by sergqwer) running on Windows, fronted by llama-swap as the model switcher
引擎：strata-nvfp4（由 sergqwer 基于 Niko1221/Strata 分叉），运行于 Windows，前端使用 llama-swap 作为模型切换器
Model: Qwen3.8-Flash-Next 125B-A6B hybrid MoE — abliterated checkpoint, self-converted to NVFP4
模型：Qwen3.8-Flash-Next 125B-A6B 混合 MoE —— abliterated 检查点，自行转换为 NVFP4
Weights: NVFP4 experts (4.5-bit, calibrated scales) + Q8_0 dense
权重：NVFP4 专家（4.5-bit，校准缩放）+ Q8_0 密集层
KV cache: int8
KV 缓存：int8
PLE table: bf16
PLE 表：bf16
Embeddings: bf16
嵌入：bf16
MTP: draft head, spec 5, spec-min-p 0.7
MTP：草稿头，spec 5，spec-min-p 0.7
Context: 262,144 max
上下文：最大 262,144
Compaction: custom trip line at 220,144
压缩：自定义触发线位于 220,144
Sampling: temp 0.8 / presence_penalty 0.3 / min_p 0.02
采样：温度 0.8 / 存在惩罚 0.3 / min_p 0.02
Vision: fp32 mmproj, encoder on CPU
视觉：fp32 mmproj，编码器在 CPU 上运行
RAM tier: 63 GiB pinned experts; hot experts cached in VRAM; misses computed CPU+PCIe overlapped with GPU
RAM 层级：63 GiB 固定专家；热专家缓存在 VRAM 中；未命中时通过 CPU+PCIe 与 GPU 重叠计算
Harness: dsh (web GUI), OpenAI API
框架：dsh（Web GUI），OpenAI API
| Metric | Value |
| 指标 | 值 |
|---|---|
| Wall: first message → last reply | 33m 45s |
| 墙钟时间：首条消息 → 最后回复 | 33m 45s |
| − my approval latency (25 permission prompts) | −10m 35s |
| − 我的审批延迟（25 个权限提示）| −10m 35s |
| **= pure agent working time** | 23m 10s |
| **= 纯代理工作时间** | 23m 10s |
| Steps | 87 |
| 步骤 | 87 |
| Output tokens (thinking + code + tool calls) | 124,443 |
| 输出 token（思考 + 代码 + 工具调用）| 124,443 |
| Fresh input tokens | 96,334 |
| 新输入 token | 96,334 |
| Cache-read tokens | 9,340,144 (**99.0% hit rate**) |
| 缓存读取 token | 9,340,144（**命中率 99.0%**）|
| Decode (per-step, approval-adjusted) | median 116 t/s · mean 103 · peak 171 |
| 解码（每步，经审批调整）| 中位数 116 t/s · 均值 103 · 峰值 171 |
| TTFT | median 0.7s (max 24.9s at deep context) |
| TTFT | 中位数 0.7s（深层上下文时最大 24.9s）|
| Prefill (fresh tokens) | mean 3,332 t/s · peak 6,623 t/s |
| Prefill（新 token）| 均值 3,332 t/s · 峰值 6,623 t/s |
| Longest single burst | 21,279 tokens |
| 最长单次突发 | 21,279 tokens |
| Compactions | 0 (never hit the 220k line in 33 min) |
| Compactions | 0（在 33 分钟内从未达到 220k 行） |
TOOL CALLS — 107 total, 14m 10s of tool runtime
TOOL CALLS — 共 107 次，工具运行时间 14 分 10 秒
更进一步：量化金融体系
看懂新闻只是起点——沿量化金融路径，把它变成能交付的工程能力