diff --git a/build.gradle.kts b/build.gradle.kts
index 1bd96e1..eb01cc2 100644
--- a/build.gradle.kts
+++ b/build.gradle.kts
@@ -46,6 +46,12 @@ dependencies {
     // Coroutines
     implementation("org.jetbrains.kotlinx:kotlinx-coroutines-core:1.8.1")
     implementation("org.jetbrains.kotlinx:kotlinx-coroutines-swing:1.8.1")
+
+    val langchain4jVersion = "0.31.0"
+    implementation("dev.langchain4j:langchain4j:$langchain4jVersion")
+    // llama.cpp 서버가 OpenAI API와 호환되므로 이 라이브러리를 사용합니다.
+    implementation("dev.langchain4j:langchain4j-open-ai:$langchain4jVersion")
+
 }
 
 compose.desktop {
diff --git a/src/main/kotlin/Main.kt b/src/main/kotlin/Main.kt
index 3f3a185..f5e9144 100644
--- a/src/main/kotlin/Main.kt
+++ b/src/main/kotlin/Main.kt
@@ -18,6 +18,7 @@ import ui.SettingsScreen
 enum class AppScreen { Settings, Dashboard }
 
 fun main() = application {
+
     // 앱 실행 시 필요한 바이너리 경로 (실행 파일 위치)
     val binPath = "./src/main/resources/bin/llama-server"
 
@@ -40,7 +41,8 @@ fun main() = application {
                         vtsAccountNo = it[ConfigTable.vtsAccountNo],
                         isSimulation = it[ConfigTable.isSimulation],
                         htsId = it[ConfigTable.htsId],
-                        modelPath = it[ConfigTable.modelPath]
+                        modelPath = it[ConfigTable.modelPath],
+                        embedModelPath = it[ConfigTable.embedModelPath]
                     )
                 }
             }
@@ -62,7 +64,10 @@ fun main() = application {
 
                             // LLM 서버 시작 (설정된 모델 경로 사용)
                             if (config.modelPath.isNotEmpty()) {
-                                LlamaServerManager.startServer(binPath, config.modelPath)
+                                LlamaServerManager.startServer(binPath, config.modelPath,port = 8080)
+                            }
+                            if (config.embedModelPath.isNotEmpty()) {
+                                LlamaServerManager.startServer(binPath, config.embedModelPath, port = 8081)
                             }
 
                             // 대시보드로 화면 전환
diff --git a/src/main/kotlin/database/DatabaseFactory.kt b/src/main/kotlin/database/DatabaseFactory.kt
index 05ec296..3860edc 100644
--- a/src/main/kotlin/database/DatabaseFactory.kt
+++ b/src/main/kotlin/database/DatabaseFactory.kt
@@ -27,6 +27,7 @@ object ConfigTable : Table("app_config") {
     val vtsAccountNo = varchar("vts_account_no", 20).default("")
     val isSimulation = bool("is_simulation").default(true)
     val modelPath = varchar("model_path", 512).default("")
+    val embedModelPath = varchar("embed_model_path", 512).default("")
     val htsId = varchar("hts_id", 50).default("") // HTS ID 컬럼 추가
     override val primaryKey = PrimaryKey(id)
 }
@@ -61,6 +62,27 @@ object TradeLogTable : Table("trade_logs") {
     override val primaryKey = PrimaryKey(id)
 }
 
+class VectorColumnType(private val dimension: Int) : ColumnType<String>() {
+    // [수정] H2에서 ARRAY는 내부 타입을 명시해야 합니다.
+    // VECTOR 대신 FLOAT8 ARRAY를 사용하면 벡터 연산 함수와 100% 호환됩니다.
+    override fun sqlType(): String = "FLOAT8 ARRAY"
+
+    override fun valueFromDB(value: Any): String = value.toString()
+
+    override fun notNullValueToDB(value: String): Any = value
+}
+
+object VectorStoreTable : Table("VECTOR_STORE") {
+    val id = integer("id").autoIncrement()
+    val content = text("content")
+    val metadata = text("metadata")
+
+    // 이제 FLOAT8 ARRAY 타입으로 컬럼이 생성됩니다.
+    val embedding = registerColumn<String>("embedding", VectorColumnType(1024))
+
+    override val primaryKey = PrimaryKey(id)
+}
+
 object DatabaseFactory {
     fun init() {
         val dbPath = File("db/autotrade_db").absolutePath
@@ -71,10 +93,12 @@ object DatabaseFactory {
 
         transaction {
             // 테이블 생성 (AutoTradeTable 포함)
-            SchemaUtils.create(ConfigTable, TradeLogTable, AutoTradeTable)
+            SchemaUtils.createMissingTablesAndColumns(ConfigTable, TradeLogTable, AutoTradeTable,VectorStoreTable)
         }
     }
 
+
+
     // --- 자동매매(감시) 관련 함수 ---
 
 
@@ -174,7 +198,8 @@ object DatabaseFactory {
                 vtsAccountNo = it[ConfigTable.vtsAccountNo],
                 isSimulation = it[ConfigTable.isSimulation],
                 htsId = it[ConfigTable.htsId], // htsId 로드
-                modelPath = it[ConfigTable.modelPath]
+                modelPath = it[ConfigTable.modelPath],
+                embedModelPath = it[ConfigTable.embedModelPath],
             )
         }
     }
@@ -192,6 +217,7 @@ object DatabaseFactory {
                 it[isSimulation] = config.isSimulation
                 it[htsId] = config.htsId
                 it[modelPath] = config.modelPath
+                it[embedModelPath] = config.embedModelPath
             }
         }
     }
diff --git a/src/main/kotlin/model/AppConfig.kt b/src/main/kotlin/model/AppConfig.kt
index 7d60502..3cf1e3e 100644
--- a/src/main/kotlin/model/AppConfig.kt
+++ b/src/main/kotlin/model/AppConfig.kt
@@ -25,7 +25,8 @@ data class AppConfig(
 
     var websocketToken: String = "",
     val isSimulation: Boolean = true,
-    val modelPath: String = "") {
+    val modelPath: String = "",
+    val embedModelPath: String = "") {
 
     val accountNo : String
         get() {
diff --git a/src/main/kotlin/network/AiService.kt b/src/main/kotlin/network/AiService.kt
index 6249660..19572ef 100644
--- a/src/main/kotlin/network/AiService.kt
+++ b/src/main/kotlin/network/AiService.kt
@@ -3,6 +3,7 @@ package network
 import io.ktor.client.*
 import io.ktor.client.call.*
 import io.ktor.client.engine.cio.*
+import io.ktor.client.plugins.HttpTimeout
 import io.ktor.client.plugins.contentnegotiation.*
 import io.ktor.client.request.*
 import io.ktor.http.*
@@ -19,10 +20,15 @@ object AiService {
                 coerceInputValues = true
             })
         }
+        install(HttpTimeout) {
+            requestTimeoutMillis = 60_000  // 전체 요청 대기 시간을 60초로 설정
+            connectTimeoutMillis = 10_000  // 서버 연결 대기 시간 10초
+            socketTimeoutMillis = 60_000   // 데이터 수신 대기 시간 60초
+        }
     }
 
-    private const val LLM_URL = "http://localhost:8080/completion"
-
+//    private const val LLM_URL = "http://localhost:8080/completion"
+    private const val LLM_URL = "http://127.0.0.1:8080/completion"
     /**
      * 종목명, 현재가, 실시간 체결내역을 바탕으로 AI 분석 결과를 가져옵니다.
      */
@@ -38,26 +44,15 @@ object AiService {
 
         // Gemma에게 전달할 프롬프트 구성
         val prompt = """
-            <start_of_turn>user
-            당신은 20년 경력의 전문 주식 트레이더이자 데이터 분석가입니다. 
-            다음 데이터를 바탕으로 해당 종목의 현재 '수급 상황'과 '단기 전망'을 분석하여 3줄 이내로 핵심만 말해주세요.
+        <|begin_of_text|><|start_header_id|>system<|end_header_id|>
+        당신은 20년 경력의 주식 트레이더입니다. 데이터를 분석하여 짧고 단호하게 조언합니다.<|eot_id|><|start_header_id|>user<|end_header_id|>
+        다음 데이터를 분석하여 '수급 상황'과 '단기 전망'을 3줄 이내로 요약하세요.
 
-            [종목 정보]
-            - 종목명: $stockName
-            - 현재가: $currentPrice
-            
-            [최근 실시간 체결 내역]
-            $tradeSummary
-
-            분석 기준:
-            1. 매수 체결 비중이 높은지, 매도 체결 비중이 높은지 판단하세요.
-            2. 대량 체결(고래)의 움직임이 있는지 확인하세요.
-            3. 단기적으로 진입하기에 적절한 시점인지 조언하세요.
-            
-            답변은 한국어로, 친절하지만 단호한 전문가 말투를 사용하세요.<end_of_turn>
-            <start_of_turn>model
-            
-        """.trimIndent()
+        [종목] $stockName ($currentPrice)
+        [최근 체결]
+        $tradeSummary
+        <|eot_id|><|start_header_id|>assistant<|end_header_id|>
+    """.trimIndent()
 
         return try {
             val response = client.post(LLM_URL) {
@@ -72,20 +67,53 @@ object AiService {
                 "AI 서버 응답 오류: ${response.status}"
             }
         } catch (e: Exception) {
-            "분석 실패: 로컬 AI 서버(llama.cpp)가 실행 중인지 확인하세요. (${e.message})"
+            var msg = "분석 실패: 로컬 AI 서버(llama.cpp)가 실행 중인지 확인하세요. (${e.message})"
+            println(msg)
+            msg
         }
     }
+
+
+    suspend fun getEmbedding(text: String): List<Double>? {
+        return try {
+            val response = client.post("http://127.0.0.1:8080/embedding") {
+                contentType(ContentType.Application.Json)
+                setBody(EmbeddingRequest(content = text))
+            }
+            if (response.status == HttpStatusCode.OK) {
+                val res: EmbeddingResponse = response.body()
+                res.embedding
+            } else null
+        } catch (e: Exception) {
+            null
+        }
+    }
+
+
 }
 
+
+@Serializable
+data class EmbeddingRequest(val content: String)
+
+@Serializable
+data class EmbeddingResponse(val embedding: List<Double>)
+
 /**
  * llama.cpp 서버 요청 데이터 구조
  */
 @Serializable
 data class LlamaRequest(
     val prompt: String,
-    val n_predict: Int = 256,
-    val temperature: Double = 0.7,
-    val stop: List<String> = listOf("<|end_of_turn|>", "<end_of_turn>")
+    val n_predict: Int = 256,       // 답변 길이를 엄격히 제한
+    val temperature: Double = 0.4, // M3 Pro에서 더 일관된 답변을 위해 낮춤
+    val stop: List<String> = listOf(
+        "<|eot_id|>",
+        "<|end_of_text|>",
+        "<|start_header_id|>",
+        "user",
+        "model"
+    ) // [중요] AI가 멈춰야 할 지점들을 명확히 지정
 )
 
 /**
diff --git a/src/main/kotlin/network/LlamaServerManager.kt b/src/main/kotlin/network/LlamaServerManager.kt
index b6f1da3..de478d7 100644
--- a/src/main/kotlin/network/LlamaServerManager.kt
+++ b/src/main/kotlin/network/LlamaServerManager.kt
@@ -4,50 +4,64 @@ import java.io.File
 import java.io.BufferedReader
 import java.io.InputStreamReader
 import kotlinx.coroutines.*
+import java.util.concurrent.ConcurrentHashMap
 
 object LlamaServerManager {
-    private var process: Process? = null
-    private val scope = CoroutineScope(Dispatchers.IO + Job())
+    // 포트별로 프로세스를 관리합니다.
+    private val processes = ConcurrentHashMap<Int, Process>()
+    private val scope = CoroutineScope(Dispatchers.IO + SupervisorJob())
+    init {
+        Runtime.getRuntime().addShutdownHook(Thread {
+            stopAll()
+        })
+    }
 
-    fun startServer(binPath: String, modelPath: String) {
-        if (process != null || modelPath.isNullOrBlank()) return // 이미 실행 중이면 무시
+    fun startServer(binPath: String, modelPath: String, port: Int, nGpuLayers: Int = 99) {
+        // 이미 해당 포트에서 실행 중이거나 모델 경로가 비었으면 무시합니다.
+        if (processes.containsKey(port) || modelPath.isBlank()) return
 
         val command = listOf(
             binPath,
             "-m", modelPath,
-            "--port", "8080",
-            "-c", "2048",        // 컨텍스트 길이
-            "-t", "4",           // 인텔 맥 코어 수에 맞춰 스레드 제한 (부하 방지)
-            "--embedding"        // 나중에 유사도 분석 등을 위해 활성화
+            "--port", port.toString(),
+            "-c", if (port == 8081) "512" else "4096", // 임베딩용은 컨텍스트가 짧아도 충분합니다.
+            "-ngl", nGpuLayers.toString(),
+            "-t", "6", // M3 Pro의 성능 코어를 고려하여 6~8개 권장
+            "--embedding" // 임베딩 기능을 활성화합니다.
         )
 
         scope.launch {
             try {
                 val pb = ProcessBuilder(command)
-                // 실행 파일 권한 확인 (자동 부여)
+
+                pb.redirectErrorStream(true)
                 File(binPath).setExecutable(true)
 
-                process = pb.start()
-                println("✅ AI 서버 시작됨: http://localhost:8080")
+                val process = pb.start()
+                processes[port] = process
+                println("✅ AI 서버 시작 시도 (Port: $port, Model: ${File(modelPath).name})")
 
-                // 서버 로그 모니터링 (에러 디버깅용)
-                val reader = BufferedReader(InputStreamReader(process?.inputStream))
+                val reader = BufferedReader(InputStreamReader(process.inputStream))
                 var line: String?
                 while (reader.readLine().also { line = it } != null) {
-                    // 서버 준비 완료 로그 확인용
-                    if (line?.contains("HTTP server listening") == true) {
-                        println("🚀 AI 모델 로딩 완료 및 대기 중")
+                    // 로그 출력 (디버깅용)
+                     println("[Server $port] $line")
+                    if (line?.contains("server is listening") == true) {
+                        println("🚀 AI 서버 준비 완료 (Port: $port)")
                     }
                 }
             } catch (e: Exception) {
-                println("❌ AI 서버 실행 실패: ${e.message}")
+                println("❌ AI 서버 실행 실패 (Port: $port): ${e.message}")
+                processes.remove(port)
             }
         }
     }
 
-    fun stopServer() {
-        process?.destroy()
-        process = null
-        println("🛑 AI 서버 종료")
+    fun stopAll() {
+        processes.forEach { (port, process) ->
+            process.destroy()
+            println("🛑 AI 서버 종료 (Port: $port)")
+        }
+        processes.clear()
     }
 }
\ No newline at end of file
diff --git a/src/main/kotlin/network/RagService.kt b/src/main/kotlin/network/RagService.kt
new file mode 100644
index 0000000..dae93f4
--- /dev/null
+++ b/src/main/kotlin/network/RagService.kt
@@ -0,0 +1,75 @@
+// src/main/kotlin/network/RagService.kt
+
+import dev.langchain4j.data.segment.TextSegment
+import dev.langchain4j.model.openai.OpenAiChatModel
+import dev.langchain4j.model.openai.OpenAiEmbeddingModel
+import org.jetbrains.exposed.sql.*
+import org.jetbrains.exposed.sql.SqlExpressionBuilder.plus
+import org.jetbrains.exposed.sql.transactions.transaction
+import java.time.Duration
+
+object RagService {
+    // 임베딩 모델 (8081) 및 채팅 모델 (8080) 설정
+    private val embeddingModel = OpenAiEmbeddingModel.builder()
+        .baseUrl("http://127.0.0.1:8081/v1")
+        .apiKey("unused")
+        .build()
+
+    private val chatModel = OpenAiChatModel.builder()
+        .baseUrl("http://127.0.0.1:8080/v1")
+        .apiKey("unused")
+        .timeout(Duration.ofSeconds(60))
+        .build()
+
+    /**
+     * 텍스트를 임베딩하여 H2 DB에 저장합니다.
+     */
+    fun ingest(text: String, meta: String = "") {
+        val embedding = embeddingModel.embed(text).content().vector()
+
+        transaction {
+            VectorStoreTable.insert {
+                it[content] = text
+                it[metadata] = meta
+                // 벡터 데이터를 문자열 형태로 저장 (H2 포맷)
+                it[VectorStoreTable.embedding] = embedding.joinToString(",", "[", "]")
+            }
+        }
+        println("💾 H2 벡터 저장 완료: ${text.take(15)}...")
+    }
+
+    /**
+     * 질문과 가장 유사한 정보를 H2에서 검색하여 AI 답변을 생성합니다.
+     */
+    fun ask(question: String): String {
+        val queryVector = embeddingModel.embed(question).content().vector()
+        val vectorStr = queryVector.joinToString(",", "[", "]")
+
+        // H2의 VECTOR_COSINE_SIMILARITY 함수를 사용하여 검색
+        val context = transaction {
+            val query = "SELECT content FROM VECTOR_STORE " +
+                    "ORDER BY VECTOR_COSINE_SIMILARITY(embedding, '$vectorStr') DESC " +
+                    "LIMIT 3"
+
+            val results = mutableListOf<String>()
+            exec(query) { rs ->
+                while (rs.next()) {
+                    results.add(rs.getString("content"))
+                }
+            }
+            results.joinToString("\n\n")
+        }
+
+        val prompt = """
+            [참고 정보]
+            $context
+            
+            [질문]
+            $question
+            
+            위 정보를 참고하여 분석 결과를 말해주세요.
+        """.trimIndent()
+
+        return chatModel.generate(prompt)
+    }
+}
\ No newline at end of file
diff --git a/src/main/kotlin/ui/SettingsScreen.kt b/src/main/kotlin/ui/SettingsScreen.kt
index 9908930..0d5b44c 100644
--- a/src/main/kotlin/ui/SettingsScreen.kt
+++ b/src/main/kotlin/ui/SettingsScreen.kt
@@ -98,6 +98,19 @@ fun SettingsScreen(onAuthSuccess: () -> Unit) {
             ) {
                 Text(if(config.modelPath.isEmpty()) "GGUF 모델 파일을 여기로 드래그하세요" else config.modelPath, fontSize = 12.sp)
             }
+            Box(
+                modifier = Modifier.fillMaxWidth().height(100.dp).border(1.dp, Color.Gray, RoundedCornerShape(8.dp))
+                    .onExternalDrag(onDrop = { state ->
+                        val data = state.dragData
+                        if (data is DragData.FilesList) {
+                            val embedModelPath = data.readFiles().firstOrNull()?.removePrefix("file:")
+                            if (embedModelPath?.endsWith(".gguf") == true) config = config.copy(embedModelPath = embedModelPath)
+                        }
+                    }),
+                contentAlignment = Alignment.Center
+            ) {
+                Text(if(config.embedModelPath.isEmpty()) "임베드용 GGUF 모델 파일을 여기로 드래그하세요" else config.embedModelPath, fontSize = 12.sp)
+            }
 
             Spacer(Modifier.height(24.dp))
 
diff --git a/src/main/resources/bin/libggml-base.0.9.5.dylib b/src/main/resources/bin/libggml-base.0.9.5.dylib
new file mode 100755
index 0000000..9ef58e8
Binary files /dev/null and b/src/main/resources/bin/libggml-base.0.9.5.dylib differ
diff --git a/src/main/resources/bin/libggml-base.0.dylib b/src/main/resources/bin/libggml-base.0.dylib
new file mode 120000
index 0000000..b95fcb8
--- /dev/null
+++ b/src/main/resources/bin/libggml-base.0.dylib
@@ -0,0 +1 @@
+libggml-base.0.9.5.dylib
\ No newline at end of file
diff --git a/src/main/resources/bin/libggml-base.dylib b/src/main/resources/bin/libggml-base.dylib
new file mode 120000
index 0000000..f08d1fc
--- /dev/null
+++ b/src/main/resources/bin/libggml-base.dylib
@@ -0,0 +1 @@
+libggml-base.0.dylib
\ No newline at end of file
diff --git a/src/main/resources/bin/libggml-blas.0.9.5.dylib b/src/main/resources/bin/libggml-blas.0.9.5.dylib
new file mode 100755
index 0000000..ddef480
Binary files /dev/null and b/src/main/resources/bin/libggml-blas.0.9.5.dylib differ
diff --git a/src/main/resources/bin/libggml-blas.0.dylib b/src/main/resources/bin/libggml-blas.0.dylib
new file mode 120000
index 0000000..5efe2a6
--- /dev/null
+++ b/src/main/resources/bin/libggml-blas.0.dylib
@@ -0,0 +1 @@
+libggml-blas.0.9.5.dylib
\ No newline at end of file
diff --git a/src/main/resources/bin/libggml-blas.dylib b/src/main/resources/bin/libggml-blas.dylib
new file mode 120000
index 0000000..28748a5
--- /dev/null
+++ b/src/main/resources/bin/libggml-blas.dylib
@@ -0,0 +1 @@
+libggml-blas.0.dylib
\ No newline at end of file
diff --git a/src/main/resources/bin/libggml-cpu.0.9.5.dylib b/src/main/resources/bin/libggml-cpu.0.9.5.dylib
new file mode 100755
index 0000000..423f283
Binary files /dev/null and b/src/main/resources/bin/libggml-cpu.0.9.5.dylib differ
diff --git a/src/main/resources/bin/libggml-cpu.0.dylib b/src/main/resources/bin/libggml-cpu.0.dylib
new file mode 120000
index 0000000..d58caa5
--- /dev/null
+++ b/src/main/resources/bin/libggml-cpu.0.dylib
@@ -0,0 +1 @@
+libggml-cpu.0.9.5.dylib
\ No newline at end of file
diff --git a/src/main/resources/bin/libggml-cpu.dylib b/src/main/resources/bin/libggml-cpu.dylib
new file mode 120000
index 0000000..ac40938
--- /dev/null
+++ b/src/main/resources/bin/libggml-cpu.dylib
@@ -0,0 +1 @@
+libggml-cpu.0.dylib
\ No newline at end of file
diff --git a/src/main/resources/bin/libggml-metal.0.9.5.dylib b/src/main/resources/bin/libggml-metal.0.9.5.dylib
new file mode 100755
index 0000000..e414d5c
Binary files /dev/null and b/src/main/resources/bin/libggml-metal.0.9.5.dylib differ
diff --git a/src/main/resources/bin/libggml-metal.0.dylib b/src/main/resources/bin/libggml-metal.0.dylib
new file mode 120000
index 0000000..adbf41f
--- /dev/null
+++ b/src/main/resources/bin/libggml-metal.0.dylib
@@ -0,0 +1 @@
+libggml-metal.0.9.5.dylib
\ No newline at end of file
diff --git a/src/main/resources/bin/libggml-metal.dylib b/src/main/resources/bin/libggml-metal.dylib
new file mode 120000
index 0000000..8210fd7
--- /dev/null
+++ b/src/main/resources/bin/libggml-metal.dylib
@@ -0,0 +1 @@
+libggml-metal.0.dylib
\ No newline at end of file
diff --git a/src/main/resources/bin/libggml-rpc.0.9.5.dylib b/src/main/resources/bin/libggml-rpc.0.9.5.dylib
new file mode 100755
index 0000000..73a1782
Binary files /dev/null and b/src/main/resources/bin/libggml-rpc.0.9.5.dylib differ
diff --git a/src/main/resources/bin/libggml-rpc.0.dylib b/src/main/resources/bin/libggml-rpc.0.dylib
new file mode 120000
index 0000000..0ad03d3
--- /dev/null
+++ b/src/main/resources/bin/libggml-rpc.0.dylib
@@ -0,0 +1 @@
+libggml-rpc.0.9.5.dylib
\ No newline at end of file
diff --git a/src/main/resources/bin/libggml-rpc.dylib b/src/main/resources/bin/libggml-rpc.dylib
new file mode 120000
index 0000000..a871ded
--- /dev/null
+++ b/src/main/resources/bin/libggml-rpc.dylib
@@ -0,0 +1 @@
+libggml-rpc.0.dylib
\ No newline at end of file
diff --git a/src/main/resources/bin/libggml.0.9.5.dylib b/src/main/resources/bin/libggml.0.9.5.dylib
new file mode 100755
index 0000000..deccb07
Binary files /dev/null and b/src/main/resources/bin/libggml.0.9.5.dylib differ
diff --git a/src/main/resources/bin/libggml.0.dylib b/src/main/resources/bin/libggml.0.dylib
new file mode 120000
index 0000000..7770e33
--- /dev/null
+++ b/src/main/resources/bin/libggml.0.dylib
@@ -0,0 +1 @@
+libggml.0.9.5.dylib
\ No newline at end of file
diff --git a/src/main/resources/bin/libggml.dylib b/src/main/resources/bin/libggml.dylib
new file mode 120000
index 0000000..0ef7111
--- /dev/null
+++ b/src/main/resources/bin/libggml.dylib
@@ -0,0 +1 @@
+libggml.0.dylib
\ No newline at end of file
diff --git a/src/main/resources/bin/libllama.0.0.7787.dylib b/src/main/resources/bin/libllama.0.0.7787.dylib
new file mode 100755
index 0000000..051fb6b
Binary files /dev/null and b/src/main/resources/bin/libllama.0.0.7787.dylib differ
diff --git a/src/main/resources/bin/libllama.0.dylib b/src/main/resources/bin/libllama.0.dylib
new file mode 120000
index 0000000..88f39b3
--- /dev/null
+++ b/src/main/resources/bin/libllama.0.dylib
@@ -0,0 +1 @@
+libllama.0.0.7787.dylib
\ No newline at end of file
diff --git a/src/main/resources/bin/libllama.dylib b/src/main/resources/bin/libllama.dylib
new file mode 120000
index 0000000..d76d521
--- /dev/null
+++ b/src/main/resources/bin/libllama.dylib
@@ -0,0 +1 @@
+libllama.0.dylib
\ No newline at end of file
diff --git a/src/main/resources/bin/libmtmd.0.0.7787.dylib b/src/main/resources/bin/libmtmd.0.0.7787.dylib
new file mode 100755
index 0000000..80f07f7
Binary files /dev/null and b/src/main/resources/bin/libmtmd.0.0.7787.dylib differ
diff --git a/src/main/resources/bin/libmtmd.0.dylib b/src/main/resources/bin/libmtmd.0.dylib
new file mode 120000
index 0000000..fbbfcfd
--- /dev/null
+++ b/src/main/resources/bin/libmtmd.0.dylib
@@ -0,0 +1 @@
+libmtmd.0.0.7787.dylib
\ No newline at end of file
diff --git a/src/main/resources/bin/libmtmd.dylib b/src/main/resources/bin/libmtmd.dylib
new file mode 120000
index 0000000..21fd179
--- /dev/null
+++ b/src/main/resources/bin/libmtmd.dylib
@@ -0,0 +1 @@
+libmtmd.0.dylib
\ No newline at end of file
diff --git a/src/main/resources/bin/llama-server b/src/main/resources/bin/llama-server
index b05226c..38f864c 100755
Binary files a/src/main/resources/bin/llama-server and b/src/main/resources/bin/llama-server differ