LogicBasis Market Docs
Secure Gateway API
A gateway layer for external inference systems. LogicBasis provisions the selected Core during request transit while you retain control over models, providers, inference parameters, context construction, and execution flow.
The Pass-Through Principle
LogicBasis is not a hosted language model provider. Your system sends inference requests through the LogicBasis gateway while continuing to use your own external provider accounts.
The gateway validates your LogicBasis Market key, deducts exactly 1 request unit, provisions the selected Core, forwards the request to the chosen provider, and returns a normalized response structure.
LogicBasis does not rewrite or reinterpret provider outputs. It provisions the selected Core within the provider request and normalizes the resulting response.
Gateway Endpoint
Authentication
Every request requires both a LogicBasis Market key and a provider API key.
Request Format
Each Core is exposed as a dedicated POST endpoint. The gateway accepts a minimal unified request structure independent of the selected provider.
| Field | Type | Required | Description |
|---|---|---|---|
| model | string | Yes | Target provider model identifier |
| input | string | Yes | Raw user input sequence |
| temperature | number | No | Sampling temperature. Default: 0.0 |
| top_p | number | No | Nucleus sampling parameter. Default: 1.0 |
| provider | string | No | Explicit provider selector: openai, anthropic, deepseek |
| max_tokens | number | No | Supported only for Anthropic requests |
Response Normalization
Different providers return different response formats. LogicBasis normalizes provider responses into a unified structure independent of provider-specific schemas.
The returned output field contains the extracted provider-generated response content.
Core Namespace Routing
| Core | Endpoint |
|---|---|
| Analysis Core | /api/v1/lb-1.1 |
| Clarify Core | /api/v1/lb-1.2 |
| Reduction Core | /api/v1/lb-1.3 |
| Strategy Core | /api/v1/lb-1.4 |
| Synthesis Core | /api/v1/lb-1.5 |
Provider Routing
LogicBasis currently supports OpenAI, Anthropic, and DeepSeek provider routing.
If no explicit provider is supplied, the gateway infers the provider from the model identifier.
| Model Prefix | Provider |
|---|---|
| claude- | Anthropic |
| deepseek- | DeepSeek |
| all other models | OpenAI |
Units & Billing
Each successfully processed request deducts exactly 1 unit from your LogicBasis Market balance.
Provider-side inference costs are billed directly through your own provider account.
A provider-side error does not reverse the Market unit. The unit covers successful Market-side request processing and Core provisioning.
When your balance is insufficient, the gateway returns a 402 response.
Security & Data Handling
All requests are transmitted over TLS.
The gateway is designed not to persist provider API keys, payload content, provider outputs, or conversation history beyond transient request execution.
LogicBasis does not provide persistent memory, conversation storage, inference trace retention, or behavioral analytics.
Stability & Execution Responsibility
LogicBasis is designed to provide stable Core identity conditioning during provider transit.
Final output behavior remains dependent on: model family, provider infrastructure, inference parameters, runtime conditions, and context construction.
You remain responsible for output supervision, factual verification, safety evaluation, and downstream system integration.
Error Handling
| Code | Meaning |
|---|---|
| 400 | Invalid input or model |
| 401 | Missing or invalid LogicBasis key |
| 402 | Insufficient balance |
| 5xx | Internal gateway or upstream processing failure |
| Provider Status | Provider-side error codes may be forwarded transparently |
For calm and minimal communication regarding integration or account matters: contact@logicbasis.ch