Open-source client · Cross-platform · Upstream keys stay local
Share idle AI model capacity and put every call to work
Fasten Share connects model-capacity producers with callers. Publish local models, private deployments, or compatible APIs you are allowed to share while the platform handles discovery, streaming routing, metering, and credit settlement.
Only connect model capacity you are authorized to use and share, and follow the upstream provider terms.
Model network online HTTPS + WebSocket
01 Consumer Compatible API request
→Compatible API request
02 Fasten Share Discover · route · meter
→Stream response
03 Producer client Inject key locally
Model backendOllama · API · Private
KEY Open-source producer clientMajor model protocolsWindows · macOS · LinuxStreaming request and response routing
One client, two ways to participate
Provide model capacity or call shared nodes
Producers and consumers use the same desktop client with clearly separated credentials, credits, and workflows.
PRODUCER
As a producer
Connect a model backend you are allowed to share and earn producer credits after successful consumer calls.
- Local models, private deployments, and compatible online APIs
- Upstream API keys are stored and injected locally, never distributed to consumers
- Control models, concurrency, credit multiplier, and sharing status
Read the producer guide → CONSUMER
As a consumer
Find online model nodes, select a producer, and connect AI tools or API clients through one generated endpoint.
- Discover nodes by model, protocol, rating, and availability
- Generate a consumer API key and compatible endpoint
- Streaming responses, follows, and monthly ratings
Read the consumer guide → From publishing to inference
How the platform works
Consumers never receive a producer’s upstream key. Inference requests and responses pass through the Fasten Share service for routing, metering, and settlement.
- 01
Connect a backend
Configure the protocol, base URL, models, and locally stored credential in the client.
- 02
Publish a healthy node
The producer opens an outbound connection and runs health checks so unhealthy backends leave discovery.
- 03
Choose and call
A consumer finds a node and sends standard HTTP or streaming requests to the generated Fasten Share endpoint.
- 04
Meter and settle
The service meters actual forwarded usage and tracks consumer and producer credits separately.
Works with your existing setup
Connect local models, online APIs, and popular AI tools
Exact compatibility depends on the protocol and tool capabilities published by each producer. The current client UI is authoritative.
Producer backends
OllamaLM StudiovLLMPrivate deployments
Model protocols
OpenAIResponses APIAnthropicAzure OpenAI
Consumer tools
CodexClaude CodeOpenCodeOpenClawHermesPi
Core capabilities
The complete path from sharing to inference
01 One compatible entry point
Consumers call the endpoint generated by Fasten Share without receiving a producer’s original upstream credential.
02 Local and online backends
Connect Ollama, LM Studio, vLLM, and OpenAI-, Anthropic-, or Azure OpenAI-style APIs.
03 Continuous health checks
Periodic checks reduce the exposure of nodes with authentication failures, exhausted quota, or unavailable endpoints.
04 Discovery and trust signals
Search by model and protocol, then use availability, follows, and recent three-month ratings to choose a producer.
05 Streaming transport
The platform forwards requests and responses between consumer HTTP and the producer’s outbound connection, including streaming content.
06 Authoritative metering
Actual forwarded data is metered centrally while consumer and producer credits remain separate.
Security boundary
Keys stay local, and the traffic path stays clear
A producer’s upstream API key is stored by the local client and injected only when it calls the model backend. It is not sent to consumers or the Fasten Share service. Consumer prompts and model responses do pass through the service to enable routing, metering, and settlement, so always use the official HTTPS domain.
Inspect the client on GitHub →
1Inspect the open-source client
The producer client is MIT-licensed, including the local credential and request-forwarding implementation.
2Stay in control of sharing
Adjust concurrency and the credit multiplier, stop an individual backend, or take the client offline at any time.
3Producers confirm sharing rights
Do not connect capacity you may not forward or that violates upstream account, subscription, or commercial-use terms.
Start with a concrete problem
Practical producer and consumer guides
Step-by-step help for local models, compatible APIs, CLIProxyAPI, and AI tool configuration.
View all guides → Producer From authorization and backend preparation to health checks and producer credits, learn how to share idle model capacity without ignoring credential, policy, or income risks.
Read guide →
Producer & consumer Learn how Fasten Share connects model-capacity producers and consumers through discovery, streaming routing, credential separation, usage metering, and credit settlement.
Read guide →
Producer Configure Ollama connectivity, model IDs, an OpenAI-compatible endpoint, Fasten Share health checks, and resource limits to share a local model responsibly.
Read guide →
Common questions before you start
Is a producer’s upstream API key uploaded?+
No. The upstream credential stays on the producer device and is injected by the local client when it calls the backend. Consumers and the Fasten Share service do not receive it.
Do prompts and model responses pass through Fasten Share?+
Yes. The service performs discovery, routing, authoritative metering, and credit settlement, so prompts and responses are forwarded through the official HTTPS service.
What model capacity can I share?+
You can connect local models, private deployments, or compatible online APIs you are authorized to share. Exact protocols and models depend on current client support and your configuration.
Do producer credits guarantee cash income?+
No. Producer credits come from successful usage. Withdrawal eligibility, exchange rates, thresholds, review, and payout rules are those shown in the client when you apply; no fixed income is promised.
How does a consumer call a shared model?+
Sign in, create a consumer API key, find and select a model node, then copy its generated endpoint or use the tool configuration assistant.
Which operating systems are supported?+
Desktop packages are published for Windows, macOS, and Linux. Check the download page for currently available builds and versions.
Ready to share or call model capacity?
Use one Fasten Share client for producer setup, node discovery, and AI tool configuration.