What it is

BBAIS hosts only BLK-AI models. Any other model is yours to bring: your address, your key, your provider's bill. With BLK-AI: Bring Your Own Model you connect your own model and use it inside BLK-AI: Chat, BLK-AI: Artifacts, BLK-AI: BLKBUILDER, with BLK-AI's memory, tools, presets and builder working on it. All you need is a free MyBLKWorldHQ account.

Connecting and using your own model is free: it costs no BLKCOIN. Your model's provider bills you for its usage (the tokens your model reads and writes) at their own price. Every other BLK-AI feature keeps its BLKCOIN price.

Working today in: BLK-AI: Chat, BLK-AI: Artifacts, BLK-AI: BLKBUILDER. Coming soon: BLK-AI: Cinema Director, BLK-AI: Co-Pilot, BLK-AI: Accessibility chat, BLKBUILDER automations.

Connect your model now.

Connect in a few clicks

1. Open the model menu. In BLK-AI: Chat, Artifacts or BLKBUILDER, open the model menu and choose Connect your model.
2. Paste the address and key. Use your server's address (it ends in /v1 for most servers) or tap one of the common addresses. The model list fills itself.
3. Test & connect. BLK-AI sends a short live test, checks whether your model calls tools and reads images, then adds it to every model menu that can run it.

One model at the address? It is picked for you. Several? Choose one from the list, or type its name. An address that does not list its models still works: type the model name yourself.

For an open-weight model hosted on Hugging Face, type its hub id (for example org/model-name) in the hub id field and paste your own Hugging Face token as the key. Add :fastest or :cheapest to the id to choose how the hub picks a provider.

More options lets you name the model, set its format, its context window, and extra headers your server needs.

Addresses for common providers

Paste the address your provider gives you. These are the standard ones (checked live on 21 October 2026):

Most servers use the chat completions format, and that is the default. Anthropic uses its own Messages format, which BLK-AI speaks natively; it is chosen for you from the address.

Open-weight models you run yourself

BLK-AI calls your model from the internet, so your server needs a public https address. A model on your own computer (for example at http://localhost:11434) cannot be reached directly: give it a public address with a tunnel or a reverse proxy, then paste that address. Private-network and local addresses are refused for safety.

Any server that offers /v1/chat/completions works: vLLM, Text Generation Inference, SGLang, llama.cpp's server, LM Studio and Ollama all do. If your server lists its models at /v1/models, the model menu fills itself. To let BLK-AI's tools run natively on vLLM, start it with tool calling switched on for your model; without it, BLK-AI still drives tools by instruction.

BBAIS does not host, upload or store your weights. Your model stays on your server.

What is free and what is not

Free: connecting your model, testing it, and every reply it writes in BLK-AI: Chat, Artifacts and BLKBUILDER. No BLKCOIN is used for your model's replies, and there is no minimum charge.

Your provider's bill: your model's usage, the tokens it reads and writes, at your provider's price. Dashboard → Your Models shows the requests and tokens your provider reported for each model.

Still priced in BLKCOIN: the BLK-AI features that ride along with a reply. A chat turn that runs a live search (Smart Search, Deep Search, Web Scrape) pays the chat-message price, as every searched turn does; Internal Knowledge turns on your model are free. Envision images and videos, reading an image when your model cannot see images, BLKBUILDER hosting, domains and automations keep their own prices. Every charge shows in your BLKCOIN history before it lands.

Tools, vision and the live test

When you connect, BLK-AI sends your model a short live test and saves it only if it answers. It also checks two things: whether your model calls tools natively (badge TOOLS) and whether it reads images (badge VISION).

A model without native tools still uses BLK-AI's tools: BLK-AI describes them in the instructions and reads the call back from the reply. A model that cannot read images still works for text; images are read by BLK-AI's vision model at its usual price.

If your server rejects a setting (some models refuse a temperature, or want a different length field), BLK-AI retries without it and remembers that for next time.

Each turn carries BLK-AI's knowledge, your memory and your conversation, so a model with a context window of 32K tokens or more works best. Tell BLK-AI your model's window under More options and older turns are summarized to fit it; the summaries are written by your own model.

Privacy and your key

Your key is encrypted the moment it arrives. It is never shown again (you see only its last four characters), never sent to your browser, and never written to logs. Removing the model deletes the key with it, immediately.

Only you can use, test or see your models. Nobody else's model menu shows them, and no one else's request can run them.

What you send to your model goes to its provider under that provider's terms. BLK-AI's platform rules still apply to every request, whichever model answers.

Troubleshooting

• "Your model's provider rejected the key." Paste a fresh key in Dashboard → Your Models → Edit.
• "Could not find that model or address." Check the model name exactly as your provider spells it.
• "Out of credit." Top up with your provider; BLKCOIN cannot pay your provider.
• "Rate-limiting you right now." Your provider's limit; wait a moment or raise it with them.
• "Unreachable" or "did not answer in time." Your server is down, asleep or not public. Check it, then Re-test in Your Models.
• "Only works on your own computer or network." Give the server a public https address.

When your model fails, BLK-AI tells you why and never switches to a BLK-AI model on its own. Pick a BLK-AI model in the model menu if you want to keep going.