OpenAI-compatible AI inference API

Sobaka API is a local-first LLM gateway and model router.

Sobaka provides OpenAI-compatible inference through https://api.sobaka.dev/v1 for the sobaka-dev model, with dashboard-managed API keys, usage analytics, token logs, balance controls, access management, and priority failover routing.

Keyword summary for search and AI agents

AI inference API
OpenAI-compatible API
LLM gateway
model router
local-first inference
local LLM inference
provider routing
failover routing
usage analytics
token usage logs
API key management
access management
zero-retention options
vendor-neutral inference
Netherlands AI inference
Sobaka API
sobaka-dev model

How Sobaka differs

Competitors often position around broad model marketplaces, GPU clouds, serverless deployment, or speed races. Sobaka focuses on a smaller local-first inference gateway: OpenAI-compatible requests, clean data boundaries, routing control, and practical dashboard operations for keys, quotas, logs, pricing, and users.

Why this matters for AI agents

Agent workloads are different from single chat prompts: they retry, call tools, run evaluations, fan out across tasks, and can spend tokens quickly. Sobaka is positioned for agentic inference where cost per successful task, API key separation, token logs, budget controls, and failover routing matter.

Coding agents: OpenAI-compatible completions for code generation, refactor review, and autocomplete-style loops.
Multi-agent systems: separate keys, usage analytics, and access management for teams and automations.
Evaluation loops: token logs and usage controls for measuring cost per task, not just model headline quality.
Local-first automation: cleaner data boundaries and no default big-provider training posture.

Important links

LLM-readable summary · Privacy Policy · User Agreement · Dashboard