42 lines
1.2 KiB
JSON
42 lines
1.2 KiB
JSON
{
|
|
"providers": {
|
|
"llama-cpp": {
|
|
"baseUrl": "http://localhost:8080/v1",
|
|
"api": "openai-completions",
|
|
"apiKey": "sk-no-key",
|
|
"models": [
|
|
{
|
|
"id": "unsloth/Qwen3.5-9B-GGUF:Q5_K_M",
|
|
"name": "Qwen 3.5 9B Q5_K_M (Local M1 Max - Unsloth)",
|
|
"reasoning": true,
|
|
"input": ["text"],
|
|
"contextWindow": 262144,
|
|
"maxTokens": 32768,
|
|
"cost": { "input": 0, "output": 0, "cacheRead": 0, "cacheWrite": 0 },
|
|
"compat": {
|
|
"supportsDeveloperRole": false,
|
|
"supportsReasoningEffort": false,
|
|
"maxTokensField": "max_tokens",
|
|
"thinkingFormat": "qwen"
|
|
}
|
|
},
|
|
{
|
|
"id": "unsloth/Qwen3.5-4B-GGUF:Q5_K_M",
|
|
"name": "Qwen 3.5 4B Q5_K_M (Local M1 Max - Unsloth)",
|
|
"reasoning": true,
|
|
"input": ["text"],
|
|
"contextWindow": 262144,
|
|
"maxTokens": 32768,
|
|
"cost": { "input": 0, "output": 0, "cacheRead": 0, "cacheWrite": 0 },
|
|
"compat": {
|
|
"supportsDeveloperRole": false,
|
|
"supportsReasoningEffort": false,
|
|
"maxTokensField": "max_tokens",
|
|
"thinkingFormat": "qwen"
|
|
}
|
|
}
|
|
]
|
|
}
|
|
}
|
|
}
|