80 lines
2.2 KiB
JSON
80 lines
2.2 KiB
JSON
{
|
|
"providers": {
|
|
"sparkl": {
|
|
"baseUrl": "http://192.168.10.111:8000/v1",
|
|
"api": "openai-completions",
|
|
"apiKey": "vllm",
|
|
"compat": {
|
|
"supportsDeveloperRole": false,
|
|
"supportsReasoningEffort": false
|
|
},
|
|
"models": [
|
|
{
|
|
"id": "Qwen/Qwen3.8-27B",
|
|
"name": "Qwen3.8 27B (local vLLM)",
|
|
"reasoning": true,
|
|
"input": ["text"],
|
|
"contextWindow": 16384,
|
|
"maxTokens": 16384,
|
|
"samplingParams": {
|
|
"temperature": 0.6,
|
|
"top_p": 0.95,
|
|
"top_k": 20
|
|
},
|
|
"cost": { "input": 0, "output": 0, "cacheRead": 0, "cacheWrite": 0 },
|
|
"compat": {
|
|
"thinkingFormat": "qwen-chat-template",
|
|
"maxTokensField": "max_tokens"
|
|
}
|
|
},
|
|
{
|
|
"id": "Qwen/Qwen3.8-27B-FP8",
|
|
"name": "Qwen3.8 27B FP8 (local vLLM)",
|
|
"reasoning": true,
|
|
"input": ["text"],
|
|
"contextWindow": 65536,
|
|
"maxTokens": 65536,
|
|
"samplingParams": {
|
|
"temperature": 0.6,
|
|
"top_p": 0.95,
|
|
"top_k": 20
|
|
},
|
|
"cost": { "input": 0, "output": 0, "cacheRead": 0, "cacheWrite": 0 },
|
|
"compat": {
|
|
"thinkingFormat": "qwen-chat-template",
|
|
"maxTokensField": "max_tokens"
|
|
}
|
|
}
|
|
]
|
|
},
|
|
"sparkr": {
|
|
"baseUrl": "http://192.168.10.110:8000/v1",
|
|
"api": "openai-completions",
|
|
"apiKey": "vllm",
|
|
"compat": {
|
|
"supportsDeveloperRole": false,
|
|
"supportsReasoningEffort": false
|
|
},
|
|
"models": [
|
|
{
|
|
"id": "Qwen/Qwen3-Coder-Next-FP8",
|
|
"name": "Qwen3 Coder Next FP8 (local vLLM)",
|
|
"reasoning": false,
|
|
"input": ["text"],
|
|
"contextWindow": 262144,
|
|
"maxTokens": 8192,
|
|
"samplingParams": {
|
|
"temperature": 0.6,
|
|
"top_p": 0.95,
|
|
"top_k": 20
|
|
},
|
|
"cost": { "input": 0, "output": 0, "cacheRead": 0, "cacheWrite": 0 },
|
|
"compat": {
|
|
"maxTokensField": "max_tokens"
|
|
}
|
|
}
|
|
]
|
|
}
|
|
}
|
|
}
|