THX-01 — HAL-X AI 的本地 AI 模型
作者： Jakub Rusinowski · 最后更新：
HAL-X AI 的 322M 多语言决策模型：mmBERT-base 编码器加决策头，在 18 种语言上进行了后训练。除 choice、是/否和 score 外，还能从文档中返回数字、原文摘录和引用。
变体
THX-01 最小的变体在 Q4_K_M 下约需 1 GB 显存——量化权重加框架开销，不含 KV 缓存。
| 模型 | Q4 下显存 | 显存 | 上下文 | 运行 |
|---|---|---|---|---|
| THX-01 322M → 322M | ~1 GB | 1,024 tokens per question | thx01.server :8095 |
显存为 Q4_K_M 下的量化权重加开销，与 GPU 与显存检测器使用同一引擎计算。
如何在本地运行 THX-01
安装 Ollama，然后拉取标签。
Served by thx01.server (default port 8095). This model does not run in Ollama.
在上方选择一个尺寸，查看它自己的显存、速度估算和安装命令。
许可证
Commercial use permitted. No usage restrictions beyond attribution.
适用于: THX-01 322M推荐 GPU
目录中能在本地运行 THX-01（至少 1 GB 显存）的最便宜 GPU 是 Intel Arc B570 (10 GB).
THX-01 — 常见问题
How much VRAM does THX-01 need?
THX-01 needs about 1 GB VRAM at Q4_K_M quantization for its smallest variant. Variants: THX-01 322M (1 GB, Q4_K_M). On Apple Silicon, unified memory counts toward this requirement.
Can I run THX-01 on an RTX 4090 (24 GB)?
Yes — THX-01 runs on an RTX 4090 (24 GB) and other 24 GB cards such as the RTX 3090. Smaller variants also fit comfortably on 8–16 GB GPUs at Q4_K_M.
What quantization should I use for THX-01?
Q4_K_M is the best balance of quality and VRAM for THX-01 in most cases. Choose Q8_0 for near-lossless quality if you have spare VRAM, or smaller quants (Q3/Q2) only when memory is tight.
How do I run THX-01 locally?
THX-01 is a decision model: it is called over an HTTP endpoint (/v1/systemone), not chatted with. Where a variant is available in Ollama 0.35 or newer, pull it with `ollama pull` and send requests to the local server; the others ship their own server. Each variant page shows the exact commands for that model.