Raven — My Self‑Hosted Media Server AI

[rwnano]

OC is not banned by anthropic if you use it with API (super expensive though). But to answer your question directly, yes, it does work with OpenAI, and not only OpenAI. It works with at least 15 more AI providers out there, including KIMI 2.5, which is the best next thing after OpenAI.

Any local. LLM models that oc can be redirected to?

 

[arunvenkats]

Effectively banned for hobbyists like us.

They want to make sure their discounted (subscription) token capacity is available for the products which is part of their growth plan and business priorities. I always knew this would happen. Why would they subsidise something which is of no use to them. To be fair they did two things - [1] They provided free credits (equal to your subscription value) which you can use for a month before you decide how you are going to proceed. [2] Allow API access which means people who are using it for any critical use can continue using them though at a higher cost.

Regards,
Arun

 

[sam9s]

Any local. LLM models that oc can be redirected to?

Ollama or LM Studio but you need a beefy GPU for running smart models... If you run old dumb models, AI can screw up anything you want your AI to do for you.

 

[sam9s]

Effectively banned for hobbyists like us.

The API will still work. It's the subscription that has been banned.

 

[arunvenkats]

Any local. LLM models that oc can be redirected to?

It can. But the official documentation says this:

"Local is doable, but OpenClaw expects large context + strong defenses against prompt injection. Small cards truncate context and leak safety. Aim high: ≥2 maxed-out Mac Studios or equivalent GPU rig (~$30k+). A single 24 GB GPU works only for lighter prompts with higher latency. Use the largest / full-size model variant you can run; aggressively quantized or “small” checkpoints raise prompt-injection risk (see Security)."

If you have $30k to spend on this, you can pay for API access to Opus itself

I do not have experience with OpenClaw + Local models. But have lots of experience with LLMs. They are the core of my startup. I did try OpenClaw and I think it is amazing. But I am not using it till the security issues associated with it sorted out. Its probably okay for what the OP is doing, but for me the power of OpenClaw comes with its ability to do my chores like managing email, calendar, customer support, market research etc. I am not going to give it access to these high risk ops for now, let alone my passwords and credit card details.

Even the tiniest local LLM runs at an excruciatingly low speed on CPUs. You need GPUs for it to be effective for any process which is not interactive. OpenClaw is interactive and not a batch process mostly. GPUs are expensive. Also the capability of small open source LLMs (which can run on consumer grade GPUs at home) are ridiculously low compared to the frontier models provided by Anthropic/OpenAI/Google. Local models can and will work. Its just that the user should be willing to accept the degraded performance both functionally and speed wise. And most importantly small local LLMs do not have any kind of guard rails against prompt injection. This *is* a huge security risk if you have given OpenClaw substantial access to your personal accounts.

There are some serious efforts to make OpenClaw like systems run in a secure way. We need to wait a few months. The scenario will be very different. But till then I am going to wait and watch from the sidelines.

One of the efforts is from the Nvidia : NemoClaw : https://www.nvidia.com/en-us/ai/nemoclaw/

Regards,
Arun

 

[arunlouie]

How much exactly we are talking here for API based usage? Say If I use OC daily for 10 instructions per day.

 

[arunvenkats]

How much exactly we are talking here for API based usage? Say If I use OC daily for 10 instructions per day.

It completely depends on the instructions. For all practical reasons an instruction like "Will it rain today?" might take just $0.001 but an instruction like "Take a look at my current most played songs and create a playlist of new songs which I might like but never knew" *might* cost $0.2 or more based on the depth of research the AI will do. If you want it to a "deep research" on a say a marketing idea and create a document containing competitor landscape etc, it might cost much more than a dollar. Don't take on face value the values I gave for the examples. Consider it only in relative terms.

The API cost is based on tokens consumed. For example, as of today the middle level model from Anthropic (sonnet) costs $3 for million tokens for input (your instructions) and $15 for million tokens for output (text it produces). Million might look big from outside, but it adds up quickly. As a ball park an average english word takes about 2 tokens.

The real danger in using API based usage is runaway situations. Say for example the AI went into a rabbit hole doing your research and spends 2 hours researching, you might end up with a API invoice which you never expected. These types of situations can be mitigated with guard rails. I do not think we are there yet.

The alternative to API was subscriptions which were really affordable. A $20 subscription provided you with a month's usage but with a fixed amount of tokens / x hours. If you exhaust your tokens you can wait for the rate limit to expire and you get new tokens. There are $100 and $200 plans which give you more tokens for a given time period. Anthropic decided not extent these subscription mode to OpenClaw like use cases. They are limiting its use to their products like Claude Code.

Regards,
Arun

 

[sam9s]

How much exactly we are talking here for API based usage? Say If I use OC daily for 10 instructions per day.

Do not take risks on API based OC. Need to understand how open claw works, the hearbeat is a 24x7 pinging layer and even if you do nothing your credits will get consumed. Better to go for OpenAI sub based for OC. Open AI supports subscription-based access for OC (obviously, since OC creator Peter is now open AI employee)
or if OpenAI is also expensive for you go for KIMI 2.5 even cheaper than OpenAI and very capable model

 

[Nero9]

damn u back again after a very long time

 

[square_wave]

Ollama or LM Studio but you need a beefy GPU for running smart models... If you run old dumb models, AI can screw up anything you want your AI to do for you.

Is a mac mini good enough ?

 

[arunlouie]

Is a mac mini good enough ?

Here you go for Macmini setup with Gemma, latest udapte

 

[sam9s]

Is a mac mini good enough ?

Yes...it is... But mini is damm expensive

 

[sam9s]

damn u back again after a very long time

Yes bro .. Nice to see and feel that old members remember me, and yes, it's good to be back.

 

[sam9s]

Hi everybody,

First of all, I would like to appreciate the moderators for bringing the thread back. I understand there was a dilemma to keep the thread alive or let it perish, since there was a thin line between what I was trying to discuss and what I actually created this entire project for. Going forward, I will keep this thread active, purely from the technical standpoint, as well as logical.

I would also like to share that I am now working on a new personal assistant based on Harmees' agent, https://hermes-agent.nousresearch.com/ And I have named it Corves. (Italian for RAVEN ) which has gained even more popularity than Open Claw. It will be interesting. Thank you all!

 

[arunlouie]

Hey @sam9s How do you compare openclaw with Hermes ?

 

[sam9s]

Hey @sam9s How do you compare openclaw with Hermes ?

Hi brother, I might not be able to answer this as of now because I have just installed Hermes and started to configure it. My personal feedback will come when I spend some time and see how well it performs for the project I am developing it for. However, going by the reviews, posts, and discussions on Twitter, Hermes is better in every aspect than Open Claw, hence the reason for me to choose Hermes for Corves, not Open Claw. Let me complete the project, then I will share all the details. I will keep you all posted as I progress in my upcoming project in a new thread.

 

[sam9s]

Hi everybody,

First of all, I would like to appreciate the moderators for bringing the thread back. I understand there was a dilemma to keep the thread alive or let it perish, since there was a thin line between what I was trying to discuss and what I actually created this entire project for. Going forward, I will keep this thread active, purely from the technical standpoint, as well as logical.

I would also like to share that I am now working on a new personal assistant based on Harmees' agent, https://hermes-agent.nousresearch.com/ And I have named it Corves. (Italian for RAVEN ) which has gained even more popularity than Open Claw. It will be interesting. Thank you all!

Ok this was edited too heavily ... members will now be confused what this Agent is for ... This will be my personal business assistant for my business ... like Corvus will know in and out of my business and can assist me creating all documents like BRD, HLDs, Design Document ... take care of all digial marketing, creating images iusgin open AI image 2 model .... marketing videos using Hyperframes skill and posting on all SM. assist me in analyzing new potential clients .. etc etc..


@MODS: I have not mentioned what business I am running so request not to edit this. It is imperative that members at least know what this new AI agent I am building is for ... other wise what the point of telling what i am creating unless I tell why I am creating. I hope you all understand ..

 

[sam9s]

Hi everybody, just thought to update you that my Corvus thread is live. If anyone is interested, please keep following that thread, where I will be discussing how I build Corvus, what it does, and any other questions members might be interested in.

Thread 'Meet Corvus: My AI Business Assistant That Lives on Telegram 🐦‍⬛'

May 29, 2026

Hey everyone! Some of you might remember my earlier post about **Raven** — my self-hosted media server AI that runs on OpenClaw. That project taught me something powerful: once you have a persistent AI agent that can actually *execute* tasks (not just chat), the possibilities go way beyond movies and music.

So I built another one. This time, not for media — for **work**.

Meet **Corvus** (Latin for "RAVEN" — staying in the bird family ). Corvus is my personal AI business assistant. It lives 24/7 on a dedicated VPS, and I interact with it entirely through Telegram. No dashboards. No...

• sam9s
• Replies: 0
• Forum: General Lounge

•