AISBF
AI Service Broker Framework — AI Should Be Free — Intelligent routing for AI providers
AISBFAI Service Broker Framework — AI Should Be Free — Intelligent routing for AI providers
AISBF routes requests across local and cloud AI providers, analyzes prompts and context to choose better models, and gives you operational control over privacy, reliability, and cost through one unified API.
Get Started Free Get Pro Test PlanTesting-period offer: unlimited Pro access for just €2/month while AISBF is in active development.
View Source Code on GitLab — Self-host it, audit it, and run it your way.
Developer guide landing page — See the multi-provider API angle in plain English.
Privacy & self-host landing page — Explore the local-first, TOR-friendly, anti-lock-in angle.
Browser companion extension — Download the BYOK/BYOT Chrome extension for AISBF routes.
🌐 Also available on TOR:
aisbfity4ud6nsht53tsh2iauaur2e4dah2gplcprnikyjpkg72vfjad.onion
AISBF is already live, useful, and evolving quickly. During the testing period, early adopters can unlock the full hosted platform for €2/month with unlimited usage. That price is temporary and may change once testing ends.
Get unlimited requests during the current test phase instead of starter-tier caps.
Use unlimited providers, rotations, and autoselections without self-hosting first.
The €2/month price is specifically for the testing period and may not stay this low later.
AISBF is open source, so you can self-host it. But if you want the fastest path to using the full platform, the hosted service removes setup friction completely — and during testing, Pro is unlimited for just €2/month.
Best when you want results now instead of another deployment project.
Best when you want maximum control over infrastructure, privacy, and routing behavior.
Different buyers care about different things. AISBF now has focused pages for the four strongest entry motivations: cheap early access, developer simplicity, privacy/self-host control, and operational governance for teams.
Best if the main thing you care about is getting unlimited Pro during testing for just €2/month.
Go to the early-adopter page →Best if you want one gateway for multi-provider AI, cleaner routing, and less glue code.
Go to the developer page →Best if you care about local-first routing, self-hosting, TOR-friendly exposure, and less vendor lock-in.
Go to the privacy/self-host page →Best if you want a quick browser popup that talks to AISBF using your own hosted or self-hosted routes.
Get the browser extension →Best if you need quotas, analytics, multi-user operations, and governance over who can use which AI providers.
Go to the teams/ops page →Analyze prompts and context before routing, so requests can land on the provider or model that best fits the actual task.
Unify Google, OpenAI, Anthropic, Ollama, and other backends behind one interface. Switch providers without rewriting your application.
Self-host, route to local models, restrict exposure, and keep operational control over where model traffic goes.
Condense long prompts and conversations with hierarchical, conversational, semantic, and algorithmic strategies to stay within model limits.
Use weighted balancing, fallback routing, and provider failover so your AI stack keeps working when a backend slows down or disappears.
Track usage, enforce TPM/TPH/TPD controls, and understand spend by user, provider, and model.
Run AISBF for teams, customers, or projects with per-user keys, quotas, and activity visibility.
Support interactive workloads with streaming responses, WebSocket-friendly behavior, and low-latency delivery.
Expose AISBF through TOR, keep sensitive workloads on local backends, and reduce reliance on any single cloud vendor.
Reduce wasted spend by routing work intelligently, caching responses where appropriate, and using prompt-aware model selection instead of static rules.
Keep control of your model traffic with self-hosting, local-model routing, traffic governance, and deployment patterns that reduce unnecessary exposure.
Use prompt and context analysis to make better routing decisions for quality, cost, and latency instead of treating every request the same.
Adopt a unified API layer so you can move between providers, mix local and remote backends, and avoid hard-coding your stack around one vendor.
Keep sensitive or cheap workloads local, burst into cloud models when needed, and fail over automatically when a backend becomes unavailable.
Scale from prototype to production with quotas, rate limits, token tracking, and operational visibility across users, providers, and models.
Condense conversations and context intelligently to stay inside model limits while preserving the parts that actually matter.
Support streaming workloads and interactive applications through APIs that work well for real-time UX, not just batch request forwarding.
Migrate quickly using familiar API patterns while keeping the option to self-host, expose over TOR, and operate with stricter privacy boundaries.
AISBF is not just a thin proxy: it is an operational layer for routing, governance, resilience, and observability across AI backends.
Use AISBF in real workflows: unify provider access, add failover, and keep sensitive prompts on local or self-hosted models when needed.
Follow the full service-user, MySQL, Redis, systemd, nginx, TLS, and verification walkthrough.
Read the install tutorial →Route your app through AISBF instead of wiring every provider directly.
Read the gateway tutorial →Use rotations, fallback routes, quotas, and route separation for production AI reliability.
Read the failover tutorial →Keep sensitive prompts on local or self-hosted backends while retaining hosted providers for general work.
Read the privacy tutorial →Use the dashboard to add providers, build failover rotations, and create autoselect routes with real screenshots.
Read the dashboard tutorial →Proxy media and embedding workloads, connect CoderAI broker workers, and configure MySQL/Redis for production.
Read the proxy tutorial →OpenAI-compatible APIs, user-scoped routes, rotations, autoselect, dashboard surfaces, and MCP automation.
Read the API docs →Use AISBF as a stable control plane above local broker workers and burst GPU endpoints.
Read the integration tutorial →Run daily provider checks, safe route changes, cache validation, and external smoke tests from one repeatable operator flow.
Read the runbook →Apply privacy-first, cost-aware, and burst GPU routing recipes, then debug multi-node drift across MySQL, Redis, and load balancers.
Read routing recipes →Use node-by-node checks and route smoke-test matrices to find inconsistent provider state, token mismatch, and cache prefix bugs.
Read the playbook →R0/month
Best for personal testing, evaluation, and small self-hosted workflows.
€2/month
Unlimited hosted usage during the testing period for early adopters who want the full platform before normal pricing lands.
OSS
Run AISBF your own way with maximum control over privacy, providers, and infrastructure.
The €2/month Pro test plan is temporary and available only during the current testing period. Need private deployment, implementation help, or production support? Contact Nexlab.
AISBF is free and open source. Both AISBF and LiteLLM are capable OSS tools, but AISBF is more opinionated about prompt-aware routing, context handling, local-cloud mixing, and privacy-conscious deployment.