From fees to compute.
A single loop: trading funds infrastructure, infrastructure runs open models, usage returns value to the protocol.
Protocol activity produces fees.
A governance-configured percentage flows into the Compute Treasury.
Treasury capital funds GPU capacity from integrated providers.
Open-source models run on the network.
Paid compute usage generates revenue back into the ecosystem.
Community compute, connected.
Every node is a real subsystem of the application. Hover a node to see what it does.
- LAZARUS COREProtocol coordination layer
- Protocol Fees
Fees generated by trading activity on the protocol.
- Compute Treasury
Protocol revenue allocated to GPU infrastructure and compute providers.
- GPU Pool
Active capacity available to run inference jobs.
- Model Registry
Open-weight models served through the network, health-checked against the provider.
- Holder Credits
Token holders receive compute credits according to protocol rules.
- Usage Revenue
Paid external usage that flows back into the Lazarus ecosystem.
Open models, health-checked.
Each card is a registry row; its status is what the configured provider answered on the last check — never a static claim.
Hold the network. Use the network.
Eligible holders receive compute credits that can be spent on models running through Lazarus Net.
Token balance determines your current compute tier.
Spend credits on supported models.
Protocol activity expands available capacity.
| Tier | Minimum balance | Credits / epoch | Notes |
|---|---|---|---|
| Node | 1,000 tokens | 200 | Entry allocation for verified holders. |
| Cluster | 10,000 tokens | 2,500 | Expanded allocation for larger positions. |
| Core | 100,000 tokens | 30,000 | Highest allocation, priority access when capacity is constrained. |
Note · Compute allocations are protocol-configurable and subject to capacity. Values shown are the current default configuration and can be changed by protocol admins.
Every fee, accounted for.
Treasury figures are read from the treasury contract or from admin-recorded events with transaction hashes. Nothing is estimated.
Open models. Open access.
Lazarus Net prioritizes open-weight models that can run across independent compute infrastructure.
Instruction-tuned chat and reasoning models for everyday inference.
Completion and repair models tuned on source code.
Long-form reasoning distillations for analysis and experimentation.
Dense vector encoders for retrieval and clustering.
Multimodal models that read images alongside text.
Models in the registry belong to their respective publishers and are used under their own licences. Lazarus Net does not claim ownership of any third-party model.
One endpoint. Metered compute.
Generate an API key from your wallet account and call the network directly. Usage is charged to your compute account and recorded as external revenue.
- AuthenticationBearer API key, hashed at rest, shown once.
- Rate limitsPer key, per wallet and per IP. HTTP 429 with Retry-After.
- AccountingCredits are reserved before the call and settled from real token usage.
- IdempotencySend idempotencyKey to make retries safe.
curl -X POST https://<your-host>/api/v1/inference \
-H "Authorization: Bearer lz_live_..." \
-H "Content-Type: application/json" \
-d '{
"model": "llama-3.1-8b-instruct",
"prompt": "Explain community-owned compute in one paragraph.",
"temperature": 0.7,
"maxTokens": 256
}'Questions, answered plainly.
From GPU providers integrated through the provider adapter layer. The Compute Treasury pays for capacity; the application routes inference jobs to it. If no provider is configured, the console says so instead of pretending.
No. Holders receive credit allocations per epoch according to their tier. Tiers, thresholds and allocations are protocol-configurable and subject to available capacity.
Open-weight models listed in the registry: general reasoning, code, research, embedding and vision. Availability is health-checked against the provider; only models the provider actually serves show as ONLINE.
Every completed job writes a usage record. External API usage is metered at the model list price and recorded as compute revenue; holder usage is recorded as subsidy cost. Nothing is projected or estimated.
The app shows "Token contract not configured" and "Awaiting treasury contract integration." Wallet sign-in, jobs, the registry and the developer API work independently of those contracts.
Only a SHA-256 hash of the prompt is stored with the job. Output text is kept for wallet-session jobs so the Jobs page can display it, and is capped in length.