/**
* LM Studio Model Specifications
* Models are discovered dynamically from the LM Studio server
* This file provides helper functions and type definitions
*/
import { ModelSpec } from '../modelTypes';
/**
* LM Studio uses dynamic model discovery via /v1/models endpoint
* Models are loaded by the user and can vary
* No static model list needed - adapter queries server for available models
*/
/**
* Create a dynamic model spec from LM Studio model data
* Used when models are discovered from the server
*/
export function createLMStudioModelSpec(
modelId: string,
contextWindow?: number,
maxTokens?: number
): ModelSpec {
return {
provider: 'lmstudio',
name: modelId,
apiName: modelId,
contextWindow: contextWindow || 4096,
maxTokens: maxTokens || 2048,
inputCostPerMillion: 0, // Local models are free
outputCostPerMillion: 0,
capabilities: {
supportsJSON: true,
supportsImages: detectVisionSupport(modelId),
supportsFunctions: detectToolSupport(modelId),
supportsStreaming: true,
supportsThinking: false
}
};
}
/**
* Detect if a model supports vision based on common name patterns
*/
function detectVisionSupport(modelId: string): boolean {
const visionKeywords = ['vision', 'llava', 'bakllava', 'cogvlm', 'yi-vl', 'moondream'];
const lowerModelId = modelId.toLowerCase();
return visionKeywords.some(keyword => lowerModelId.includes(keyword));
}
/**
* Detect if a model supports tool/function calling based on common name patterns
* Many newer models support function calling
*
* Note: Models with "nexus" or "tools-sft" use [TOOL_CALLS] content format
*/
function detectToolSupport(modelId: string): boolean {
const toolSupportedKeywords = [
'gpt', 'mistral', 'mixtral', 'hermes', 'nous', 'qwen',
'deepseek', 'dolphin', 'functionary', 'gorilla',
// Fine-tuned models with [TOOL_CALLS] format
'nexus', 'tools-sft', 'tool-calling'
];
const lowerModelId = modelId.toLowerCase();
return toolSupportedKeywords.some(keyword => lowerModelId.includes(keyword));
}
/**
* Common LM Studio model families and their typical characteristics
* Used for display purposes and capability hints
*/
export const LM_STUDIO_MODEL_FAMILIES = {
llama: {
name: 'Llama',
description: 'Meta\'s Llama models - general purpose, high quality',
typical: {
contextWindow: 4096,
supportsJSON: true,
supportsFunctions: true
}
},
mistral: {
name: 'Mistral',
description: 'Mistral AI models - efficient and capable',
typical: {
contextWindow: 8192,
supportsJSON: true,
supportsFunctions: true
}
},
phi: {
name: 'Phi',
description: 'Microsoft Phi models - compact and efficient',
typical: {
contextWindow: 2048,
supportsJSON: true,
supportsFunctions: false
}
},
qwen: {
name: 'Qwen',
description: 'Alibaba Qwen models - multilingual and capable',
typical: {
contextWindow: 8192,
supportsJSON: true,
supportsFunctions: true
}
},
deepseek: {
name: 'DeepSeek',
description: 'DeepSeek models - coding and reasoning focused',
typical: {
contextWindow: 16384,
supportsJSON: true,
supportsFunctions: true
}
}
};
/**
* Get model family information if available
*/
export function getModelFamily(modelId: string) {
const lowerModelId = modelId.toLowerCase();
for (const [familyKey, familyInfo] of Object.entries(LM_STUDIO_MODEL_FAMILIES)) {
if (lowerModelId.includes(familyKey)) {
return familyInfo;
}
}
return null;
}
