Free Qwen3.8-27B gateway with isolated Minis provider, SSE, retry/failover, and MCP.
Maintain this server? Claim it
Call this server's tools from your browser. Nothing to install.
Requests go through mcptry because browsers can't call most MCP servers directly. Header values are sent to the server for that request only and are never stored or logged.
Tool list from our last check. Connect live to refresh it.
Check both zero-cost Qwen3.8 providers.
This tool takes no arguments.
This tool isn't marked read-only, so it may change something on the server or in an account you connected.
Check both zero-cost Qwen3.8 providers.
No arguments.
{
"type": "object",
"properties": {}
}List Qwen models through the active provider.
No arguments.
{
"type": "object",
"properties": {}
}Show evaluated HAPI and Hugging Face provider candidates.
No arguments.
{
"type": "object",
"properties": {}
}List configured zero-cost providers.
No arguments.
{
"type": "object",
"properties": {}
}Run a real routed gateway inference and report the exact-model ZeroGPU fallback state.
No arguments.
{
"type": "object",
"properties": {}
}Rank the healthy free providers by priority and latency.
No arguments.
{
"type": "object",
"properties": {}
}Ask Qwen3.8-27B through H200 with verified exact-model ZeroGPU failover.
| Argument | Type | Description |
|---|---|---|
| prompt* | string | |
| system | string | |
| max_tokens | integer | |
| temperature | number | |
| reasoning_effort | string |
{
"type": "object",
"required": [
"prompt"
],
"properties": {
"prompt": {
"type": "string"
},
"system": {
"type": "string"
},
"max_tokens": {
"type": "integer"
},
"temperature": {
"type": "number"
},
"reasoning_effort": {
"type": "string"
}
}
}