OpenClaw Beta – Early Access

[rfbjr]

Fair enough, it would be nice to see how this plays out. Thank you

[EnchongG]

anyone tried Open claw with Ollama local? is it working?

[rfbjr]

I might try this in a VMs, does TM have recommended settings for the VMs. I'm still not sure based upon the lack of any information about OpenClaw and TOS 7, aside from install instructions.

[TMroy]

Even though I work for TerraMaster, I actually use a TNAS at home to store my personal files and family photos. I'm running TOS 7 with OpenClaw 2026.6.5, and everything works great. Honestly, there's nothing to worry about — OpenClaw only does what you tell it to, and it strictly follows TerraMaster's skills and CLI. If something risky comes up, it will either refuse to run it or ask you to confirm first.

[rfbjr]

Even though I work for TerraMaster, I actually use a TNAS at home to store my personal files and family photos. I'm running TOS 7 with OpenClaw 2026.6.5, and everything works great. Honestly, there's nothing to worry about — OpenClaw only does what you tell it to, and it strictly follows TerraMaster's skills and CLI. If something risky comes up, it will either refuse to run it or ask you to confirm first.

Thank you for this. Now I just need to find time.

[sianderson]

i guess the best way is to play with it and see what it can or cant do, but am i hoping for too much, would this work in a similar way to chatGPT but on a personal local level i.e only i could access it but without any usage limits because its running on my own hardware rather than a cloud? i.e could i ask it to generate an image or ask it to generate some text, or is it really only for automation of TOS?

[Felix]

i guess the best way is to play with it and see what it can or cant do, but am i hoping for too much, would this work in a similar way to chatGPT but on a personal local level i.e only i could access it but without any usage limits because its running on my own hardware rather than a cloud? i.e could i ask it to generate an image or ask it to generate some text, or is it really only for automation of TOS?

Thank you for your question.

OpenClaw itself is mainly an AI agent and automation framework. It does not provide the same AI computing capability as ChatGPT by itself.

Running OpenClaw on a NAS can be useful for:

• AI assistant integration
• System management and automation tasks
• Connecting services and workflows

However, most NAS hardware is not designed for large AI model inference. Without a powerful GPU, running large language models locally will be very limited, and image generation models generally require much more GPU memory and computing power.

For example:

• Small language models may run on some NAS devices, depending on CPU, RAM, and model size.
• ChatGPT-level models require large-scale cloud infrastructure or high-end GPU hardware.
• Image generation normally requires a dedicated GPU with sufficient VRAM.

Therefore, a NAS running OpenClaw is better considered as a personal AI assistant platform for automation and integration, rather than a complete offline replacement for ChatGPT.

If you want ChatGPT-like performance, the common approach is to use OpenClaw with an external AI service/API

[Coundead]

i guess the best way is to play with it and see what it can or cant do, but am i hoping for too much, would this work in a similar way to chatGPT but on a personal local level i.e only i could access it but without any usage limits because its running on my own hardware rather than a cloud? i.e could i ask it to generate an image or ask it to generate some text, or is it really only for automation of TOS?

Thank you for your question.

OpenClaw itself is mainly an AI agent and automation framework. It does not provide the same AI computing capability as ChatGPT by itself.

Running OpenClaw on a NAS can be useful for:

• AI assistant integration
• System management and automation tasks
• Connecting services and workflows

However, most NAS hardware is not designed for large AI model inference. Without a powerful GPU, running large language models locally will be very limited, and image generation models generally require much more GPU memory and computing power.

For example:

• Small language models may run on some NAS devices, depending on CPU, RAM, and model size.
• ChatGPT-level models require large-scale cloud infrastructure or high-end GPU hardware.
• Image generation normally requires a dedicated GPU with sufficient VRAM.

Therefore, a NAS running OpenClaw is better considered as a personal AI assistant platform for automation and integration, rather than a complete offline replacement for ChatGPT.

If you want ChatGPT-like performance, the common approach is to use OpenClaw with an external AI service/API

I think that's the right way to look at it. The important distinction is between the agent/framework and the AI model actually doing the inference.

If OpenClaw is running locally, that doesn't automatically mean you have a local, unlimited ChatGPT replacement. The model still has to run somewhere. If you connect it to a cloud API, you're still subject to that provider's limits and costs. If you connect it to a local model, then you're limited mainly by your own hardware.