Offer local AI capacity

Get your computer ready to offer AI capacity.

Run the host check, verify your account, and create a one-time setup code in Account. Enrollment creates a pending node. It can receive work only after capacity and security review.

Check your host Open account setup

You control your host. Choose which approved local models to offer. After enrollment, your host connects out to TEKIZ.AI and you can remove it at any time. Personal OAuth subscriptions and unreviewed models cannot be offered as Community capacity.

01 / prepare

Prepare the host

On a Mac, the setup command installs Ollama and the tools it needs, starts Ollama, and downloads one starter model. Setup stays on your computer.

02 / verify

Measure capacity

Check that the starter model fits your available memory and can answer requests reliably. You can add another approved model later. By default, one model stays loaded at a time.

03 / submit

Enroll your host

Create a one-time setup code in your verified Account. Your host enters review after enrollment. It can receive work after its capacity and security checks are approved.

Draft provider terms

How accepted work may earn credits.

The closed beta may issue TEKIZ.AI Credits for accepted Community work. Credits apply to Pro or Max routing first, then Community inference. Rates and payouts are not active.

A listed or idle host earns no credit. Provider rates and payouts remain under manual review.

TEKIZ.AI considers recent accepted results, task fit, availability, and speed when assigning work. New approved hosts may receive limited evaluation requests to establish a record.

70% provider pool target: The draft beta allocates 70% of net Community inference revenue to workers whose results are accepted.
Verified work only: Failed, late, duplicate, cancelled, rejected, or unused work earns no credit.
How work is assigned: Recent accepted results, task fit, availability, and speed guide assignments. Hardware brands do not.
Using earned credits: Fusion and Community inference remain Pro or Max features. A Lite provider may save earned credits for later use on an eligible plan.
Provider inference benefit: An active approved node receives 15% off TEKIZ.AI's Community inference service fee while it remains in good standing.

Request review

Request Community provider access.

Send your details for review. An approved account will show Community setup in Account. Submitting this form does not enroll a host or allow it to receive customer work.

Host check

Run the local host check.

This verified command saves the public setup script, checks its SHA-256 digest, installs Ollama and jq if they are missing, pulls a local 7B starter model, and checks it. On macOS, Ollama uses Apple Silicon's Metal support. No GPU driver setup is needed.

If your computer has limited memory, choose a smaller model and keep one model loaded at a time. For Mixture-of-Experts models, the check counts all model weights that must fit in memory.

curl -fSLo enroll_community_node.sh https://tekiz.ai/downloads/enroll_community_node.txt && curl -fSLo enroll_community_node.sh.sha256 https://tekiz.ai/downloads/enroll_community_node.sha256.txt && (shasum -a 256 -c enroll_community_node.sh.sha256 || sha256sum -c enroll_community_node.sh.sha256) && chmod 700 enroll_community_node.sh && ./enroll_community_node.sh --setup

TEKIZAI_COMMUNITY_MODELS=qwen2.5:7b-instruct \
./enroll_community_node.sh --check

Optional task mapping

Map approved models to tasks.

After both models pass the capacity check, map tasks to enrolled local tags. Fusion can send implementation work to the smaller model and research work to the larger model.

This selects a model for each task. It does not keep both models loaded. A smaller host can continue to run one job and one model at a time.

TEKIZAI_COMMUNITY_MODELS=qwen2.5:7b-instruct,qwen2.5:14b-instruct \
TEKIZAI_COMMUNITY_TASK_MODEL_MAP=implementation=qwen2.5:7b-instruct,research=qwen2.5:14b-instruct \
./enroll_community_node.sh --check

Enroll a reviewed host from Account.

Run the host check first. Then create a one-time enrollment code in your verified Account. Enrollment starts the review; it does not make your host available for work or publish a paid listing.

# Run after Account issues a one-time host enrollment code.
TEKIZAI_COMMUNITY_NODE_LABEL=my-node \
TEKIZAI_COMMUNITY_MODELS=qwen2.5:7b-instruct,qwen2.5:14b-instruct \
TEKIZAI_COMMUNITY_TASK_MODEL_MAP=implementation=qwen2.5:7b-instruct,research=qwen2.5:14b-instruct \
TEKIZAI_COMMUNITY_ENROLLMENT_TOKEN=paste-token-from-account \
./enroll_community_node.sh

Account provides the one-time code and worker instructions. The host check runs locally. TEKIZ.AI reviews your host before it can receive work.

Check enrollment in Account

Keep it safe

Protect your computer from overload.

01Start with one selected local chat model and one accepted Community job at a time. More parallelism requires measured VRAM, thermals, and latency headroom.
02Do not advertise cloud catalog models, embeddings, OCR, or a personal provider subscription. Community work must use the approved local model list.
03Use --check again after changing hardware, model server configuration (Ollama or vLLM), or residency settings. A successful check is not an instruction to run a larger model.
04Community execution requires customer opt-in. Community Direct uses QVAC to send authorized work peer to peer, so TEKIZ.AI does not receive or review the result. In router-mediated Fusion, TEKIZ.AI receives the node's result and a router-controlled verifier owns the final customer response. Payment records never contain prompts.