diff --git a/cortex-js/package.json b/cortex-js/package.json
index 1e4b08149..62231be75 100644
--- a/cortex-js/package.json
+++ b/cortex-js/package.json
@@ -48,11 +48,13 @@
     "decompress": "^4.2.1",
     "js-yaml": "^4.1.0",
     "nest-commander": "^3.13.0",
+    "openai": "^4.50.0",
     "readline": "^1.3.0",
     "reflect-metadata": "^0.2.0",
     "rxjs": "^7.8.1",
     "sqlite": "^5.1.1",
     "sqlite3": "^5.1.7",
+    "systeminformation": "^5.22.10",
     "typeorm": "^0.3.20",
     "ulid": "^2.3.0",
     "update-notifier": "^5.0.0",
diff --git a/cortex-js/src/command.module.ts b/cortex-js/src/command.module.ts
index f25d65cea..76f9a00a1 100644
--- a/cortex-js/src/command.module.ts
+++ b/cortex-js/src/command.module.ts
@@ -28,6 +28,7 @@ import { PSCommand } from './infrastructure/commanders/ps.command';
 import { KillCommand } from './infrastructure/commanders/kill.command';
 import { PresetCommand } from './infrastructure/commanders/presets.command';
 import { EmbeddingCommand } from './infrastructure/commanders/embeddings.command';
+import { BenchmarkCommand } from './infrastructure/commanders/benchmark.command';
 
 @Module({
   imports: [
@@ -56,6 +57,7 @@ import { EmbeddingCommand } from './infrastructure/commanders/embeddings.command
     KillCommand,
     PresetCommand,
     EmbeddingCommand,
+    BenchmarkCommand,
 
     // Questions
     InitRunModeQuestions,
diff --git a/cortex-js/src/file-manager/file-manager.service.ts b/cortex-js/src/file-manager/file-manager.service.ts
index 78b29270c..498509bb6 100644
--- a/cortex-js/src/file-manager/file-manager.service.ts
+++ b/cortex-js/src/file-manager/file-manager.service.ts
@@ -12,6 +12,7 @@ export class FileManagerService {
   private modelFolderName = 'models';
   private presetFolderName = 'presets';
   private extensionFoldername = 'extensions';
+  private benchmarkFoldername = 'benchmark';
   private cortexCppFolderName = 'cortex-cpp';
 
   /**
@@ -116,4 +117,14 @@ export class FileManagerService {
     const dataFolderPath = await this.getDataFolderPath();
     return join(dataFolderPath, this.extensionFoldername);
   }
+
+  /**
+   * Get the benchmark folder path
+   * Usually it is located at the home directory > cortex > extensions
+   * @returns the path to the extensions folder
+   */
+  async getBenchmarkPath(): Promise<string> {
+    const dataFolderPath = await this.getDataFolderPath();
+    return join(dataFolderPath, this.benchmarkFoldername);
+  }
 }
diff --git a/cortex-js/src/infrastructure/commanders/benchmark.command.ts b/cortex-js/src/infrastructure/commanders/benchmark.command.ts
new file mode 100644
index 000000000..e3e6a69fd
--- /dev/null
+++ b/cortex-js/src/infrastructure/commanders/benchmark.command.ts
@@ -0,0 +1,18 @@
+import { CommandRunner, SubCommand } from 'nest-commander';
+import { BenchmarkCliUsecases } from './usecases/benchmark.cli.usecases';
+
+@SubCommand({
+  name: 'benchmark',
+  subCommands: [],
+  description:
+    'Benchmark and analyze the performance of a specific AI model using a variety of system resources',
+})
+export class BenchmarkCommand extends CommandRunner {
+  constructor(private readonly benchmarkUsecases: BenchmarkCliUsecases) {
+    super();
+  }
+
+  async run(): Promise<void> {
+    return this.benchmarkUsecases.benchmark();
+  }
+}
diff --git a/cortex-js/src/infrastructure/commanders/chat.command.ts b/cortex-js/src/infrastructure/commanders/chat.command.ts
index d08d12037..718541d0d 100644
--- a/cortex-js/src/infrastructure/commanders/chat.command.ts
+++ b/cortex-js/src/infrastructure/commanders/chat.command.ts
@@ -6,8 +6,9 @@ import {
 } from 'nest-commander';
 import { ChatCliUsecases } from './usecases/chat.cli.usecases';
 import { exit } from 'node:process';
-import { ModelStat, PSCliUsecases } from './usecases/ps.cli.usecases';
+import { PSCliUsecases } from './usecases/ps.cli.usecases';
 import { ModelsUsecases } from '@/usecases/models/models.usecases';
+import { ModelStat } from './types/model-stat.interface';
 
 type ChatOptions = {
   threadId?: string;
diff --git a/cortex-js/src/infrastructure/commanders/cortex-command.commander.ts b/cortex-js/src/infrastructure/commanders/cortex-command.commander.ts
index 88fbe7a70..9bb9c33fd 100644
--- a/cortex-js/src/infrastructure/commanders/cortex-command.commander.ts
+++ b/cortex-js/src/infrastructure/commanders/cortex-command.commander.ts
@@ -10,6 +10,7 @@ import { KillCommand } from './kill.command';
 import pkg from '@/../package.json';
 import { PresetCommand } from './presets.command';
 import { EmbeddingCommand } from './embeddings.command';
+import { BenchmarkCommand } from './benchmark.command';
 
 interface CortexCommandOptions {
   version: boolean;
@@ -26,6 +27,7 @@ interface CortexCommandOptions {
     KillCommand,
     PresetCommand,
     EmbeddingCommand,
+    BenchmarkCommand,
   ],
   description: 'Cortex CLI',
 })
diff --git a/cortex-js/src/infrastructure/commanders/embeddings.command.ts b/cortex-js/src/infrastructure/commanders/embeddings.command.ts
index 45b23cc93..872715762 100644
--- a/cortex-js/src/infrastructure/commanders/embeddings.command.ts
+++ b/cortex-js/src/infrastructure/commanders/embeddings.command.ts
@@ -5,9 +5,10 @@ import {
   SubCommand,
 } from 'nest-commander';
 import { ModelsUsecases } from '@/usecases/models/models.usecases';
-import { ModelStat, PSCliUsecases } from './usecases/ps.cli.usecases';
+import { PSCliUsecases } from './usecases/ps.cli.usecases';
 import { ChatCliUsecases } from './usecases/chat.cli.usecases';
 import { inspect } from 'util';
+import { ModelStat } from './types/model-stat.interface';
 
 interface EmbeddingCommandOptions {
   encoding_format?: string;
diff --git a/cortex-js/src/infrastructure/commanders/types/benchmark-config.interface.ts b/cortex-js/src/infrastructure/commanders/types/benchmark-config.interface.ts
new file mode 100644
index 000000000..aa49c313e
--- /dev/null
+++ b/cortex-js/src/infrastructure/commanders/types/benchmark-config.interface.ts
@@ -0,0 +1,28 @@
+import { ChatCompletionMessageParam } from 'openai/resources';
+
+export interface BenchmarkConfig {
+  api: {
+    base_url: string;
+    api_key: string;
+    parameters: {
+      messages: ChatCompletionMessageParam[];
+      model: string;
+      stream?: boolean;
+      max_tokens?: number;
+      stop?: string[];
+      frequency_penalty?: number;
+      presence_penalty?: number;
+      temperature?: number;
+      top_p?: number;
+    };
+  };
+  prompts?: {
+    min: number;
+    max: number;
+    samples: number;
+  };
+  output: string;
+  concurrency: number;
+  num_rounds: number;
+  hardware: string[];
+}
diff --git a/cortex-js/src/infrastructure/commanders/types/model-stat.interface.ts b/cortex-js/src/infrastructure/commanders/types/model-stat.interface.ts
new file mode 100644
index 000000000..336917b4f
--- /dev/null
+++ b/cortex-js/src/infrastructure/commanders/types/model-stat.interface.ts
@@ -0,0 +1,8 @@
+export interface ModelStat {
+  modelId: string;
+  engine?: string;
+  duration?: string;
+  status: string;
+  vram?: string;
+  ram?: string;
+}
diff --git a/cortex-js/src/infrastructure/commanders/usecases/benchmark.cli.usecases.ts b/cortex-js/src/infrastructure/commanders/usecases/benchmark.cli.usecases.ts
new file mode 100644
index 000000000..0ac842877
--- /dev/null
+++ b/cortex-js/src/infrastructure/commanders/usecases/benchmark.cli.usecases.ts
@@ -0,0 +1,223 @@
+import { Injectable } from '@nestjs/common';
+import si from 'systeminformation';
+import fs, { existsSync, mkdirSync, readFileSync, writeFileSync } from 'fs';
+import OpenAI from 'openai';
+import { Presets, SingleBar } from 'cli-progress';
+import yaml from 'js-yaml';
+import { FileManagerService } from '@/file-manager/file-manager.service';
+import { join } from 'path';
+import { ModelsCliUsecases } from './models.cli.usecases';
+import { spawn } from 'child_process';
+import { BenchmarkConfig } from '../types/benchmark-config.interface';
+import { CortexUsecases } from '@/usecases/cortex/cortex.usecases';
+import { inspect } from 'util';
+import { defaultBenchmarkConfiguration } from '@/infrastructure/constants/benchmark';
+
+@Injectable()
+export class BenchmarkCliUsecases {
+  constructor(
+    private readonly modelsCliUsecases: ModelsCliUsecases,
+    private readonly cortexUsecases: CortexUsecases,
+    private readonly fileService: FileManagerService,
+  ) {}
+
+  config: BenchmarkConfig;
+  openai?: OpenAI;
+  /**
+   * Benchmark and analyze the performance of a specific AI model using a variety of system resources
+   */
+  async benchmark() {
+    return this.getBenchmarkConfig().then((config) => {
+      this.config = config;
+
+      // TODO: Using OpenAI client or Cortex client to benchmark?
+      this.openai = new OpenAI({
+        apiKey: this.config.api.api_key,
+        baseURL: this.config.api.base_url,
+        timeout: 20 * 1000,
+      });
+
+      spawn('cortex', ['serve'], {
+        detached: false,
+      });
+
+      return this.cortexUsecases
+        .startCortex()
+        .then(() =>
+          this.modelsCliUsecases.startModel(this.config.api.parameters.model),
+        )
+        .then(() => this.runBenchmarks())
+        .then(() => process.exit(0));
+    });
+  }
+
+  async getBenchmarkConfig() {
+    const benchmarkFolder = await this.fileService.getBenchmarkPath();
+    const configurationPath = join(benchmarkFolder, 'config.yaml');
+    console.log(configurationPath, existsSync(configurationPath));
+    if (existsSync(configurationPath)) {
+      return yaml.load(
+        readFileSync(configurationPath, 'utf8'),
+      ) as BenchmarkConfig;
+    } else {
+      const config = yaml.dump(defaultBenchmarkConfiguration);
+      if (!existsSync(benchmarkFolder)) {
+        mkdirSync(benchmarkFolder, {
+          recursive: true,
+        });
+      }
+      await writeFileSync(configurationPath, config, 'utf8');
+      return defaultBenchmarkConfiguration;
+    }
+  }
+
+  async getSystemResources() {
+    return {
+      cpu: await si.currentLoad(),
+      mem: await si.mem(),
+      gpu: (await si.graphics()).controllers,
+    };
+  }
+
+  async getResourceChange(startData: any, endData: any) {
+    return {
+      cpu:
+        startData.cpu && endData.cpu
+          ? ((endData.cpu.currentload - startData.cpu.currentload) /
+              startData.cpu.currentload) *
+            100
+          : null,
+      mem:
+        startData.mem && endData.mem
+          ? ((endData.mem.used - startData.mem.used) / startData.mem.total) *
+            100
+          : null,
+    };
+  }
+
+  async benchmarkUser() {
+    const startResources = await this.getSystemResources();
+    const start = Date.now();
+    let tokenCount = 0;
+    let firstTokenTime = null;
+
+    try {
+      const stream = await this.openai!.chat.completions.create({
+        model: this.config.api.parameters.model,
+        messages: this.config.api.parameters.messages,
+        max_tokens: this.config.api.parameters.max_tokens,
+        stream: true,
+      });
+
+      for await (const chunk of stream) {
+        if (!firstTokenTime && chunk.choices[0]?.delta?.content) {
+          firstTokenTime = Date.now();
+        }
+        tokenCount += (chunk.choices[0]?.delta?.content || '').split(
+          /\s+/,
+        ).length;
+      }
+    } catch (error) {
+      console.error('Error during API call:', error);
+      return null;
+    }
+
+    const latency = Date.now() - start;
+    const ttft = firstTokenTime ? firstTokenTime - start : null;
+    const endResources = await this.getSystemResources();
+    const resourceChange = await this.getResourceChange(
+      startResources,
+      endResources,
+    );
+
+    return {
+      tokens: this.config.api.parameters.max_tokens,
+      token_length: tokenCount, // Dynamically calculated token count
+      latency,
+      resourceChange,
+      tpot: tokenCount ? latency / tokenCount : 0,
+      throughput: tokenCount / (latency / 1000),
+      ttft,
+    };
+  }
+
+  calculatePercentiles(data: number[], percentile: number) {
+    if (data.length === 0) return null;
+    const sorted = data
+      .filter((x: number) => x !== null)
+      .sort((a: number, b: number) => a - b);
+    const pos = (percentile / 100) * sorted.length;
+    if (pos < 1) return sorted[0];
+    if (pos >= sorted.length) return sorted[sorted.length - 1];
+    const lower = sorted[Math.floor(pos) - 1];
+    const upper = sorted[Math.ceil(pos) - 1];
+    return lower + (upper - lower) * (pos - Math.floor(pos));
+  }
+
+  async runBenchmarks() {
+    const allResults: any[] = [];
+    const rounds = this.config.num_rounds || 1;
+
+    const bar = new SingleBar({}, Presets.shades_classic);
+    bar.start(rounds, 0);
+
+    for (let i = 0; i < rounds; i++) {
+      const roundResults = [];
+      const hardwareBefore = await this.getSystemResources();
+
+      for (let j = 0; j < this.config.concurrency; j++) {
+        const result = await this.benchmarkUser();
+        if (result) {
+          roundResults.push(result);
+        }
+      }
+
+      const hardwareAfter = await this.getSystemResources();
+      const hardwareChanges = await this.getResourceChange(
+        hardwareBefore,
+        hardwareAfter,
+      );
+
+      allResults.push({
+        round: i + 1,
+        results: roundResults,
+        hardwareChanges,
+      });
+
+      bar.update(i + 1);
+    }
+
+    const metrics: any = {
+      p50: {},
+      p75: {},
+      p95: {},
+    };
+    const keys = ['latency', 'tpot', 'throughput', 'ttft'];
+    keys.forEach((key) => {
+      const data = allResults.flatMap((r) =>
+        r.results.map((res: object) => res[key as keyof typeof res]),
+      );
+      metrics.p50[key] = this.calculatePercentiles(data, 50);
+      metrics.p75[key] = this.calculatePercentiles(data, 75);
+      metrics.p95[key] = this.calculatePercentiles(data, 95);
+    });
+
+    const output = {
+      hardware: await this.getSystemResources(),
+      results: allResults,
+      metrics,
+    };
+    bar.stop();
+
+    const outputFilePath = join(
+      await this.fileService.getBenchmarkPath(),
+      'output.json',
+    );
+    fs.writeFileSync(outputFilePath, JSON.stringify(output, null, 2));
+    console.log(`Benchmark results and metrics saved to ${outputFilePath}`);
+
+    console.log(
+      inspect(output, { showHidden: false, depth: null, colors: true }),
+    );
+  }
+}
diff --git a/cortex-js/src/infrastructure/commanders/usecases/cli.usecases.module.ts b/cortex-js/src/infrastructure/commanders/usecases/cli.usecases.module.ts
index feead245a..de14a704d 100644
--- a/cortex-js/src/infrastructure/commanders/usecases/cli.usecases.module.ts
+++ b/cortex-js/src/infrastructure/commanders/usecases/cli.usecases.module.ts
@@ -11,6 +11,7 @@ import { AssistantsModule } from '@/usecases/assistants/assistants.module';
 import { MessagesModule } from '@/usecases/messages/messages.module';
 import { FileManagerModule } from '@/file-manager/file-manager.module';
 import { PSCliUsecases } from './ps.cli.usecases';
+import { BenchmarkCliUsecases } from './benchmark.cli.usecases';
 
 @Module({
   imports: [
@@ -28,7 +29,14 @@ import { PSCliUsecases } from './ps.cli.usecases';
     ModelsCliUsecases,
     ChatCliUsecases,
     PSCliUsecases,
+    BenchmarkCliUsecases,
+  ],
+  exports: [
+    InitCliUsecases,
+    ModelsCliUsecases,
+    ChatCliUsecases,
+    PSCliUsecases,
+    BenchmarkCliUsecases,
   ],
-  exports: [InitCliUsecases, ModelsCliUsecases, ChatCliUsecases, PSCliUsecases],
 })
 export class CliUsecasesModule {}
diff --git a/cortex-js/src/infrastructure/commanders/usecases/ps.cli.usecases.ts b/cortex-js/src/infrastructure/commanders/usecases/ps.cli.usecases.ts
index 5e2e8db38..885a10c8b 100644
--- a/cortex-js/src/infrastructure/commanders/usecases/ps.cli.usecases.ts
+++ b/cortex-js/src/infrastructure/commanders/usecases/ps.cli.usecases.ts
@@ -6,15 +6,8 @@ import {
 } from '@/infrastructure/constants/cortex';
 import { HttpService } from '@nestjs/axios';
 import { firstValueFrom } from 'rxjs';
+import { ModelStat } from '../types/model-stat.interface';
 
-export interface ModelStat {
-  modelId: string;
-  engine?: string;
-  duration?: string;
-  status: string;
-  vram?: string;
-  ram?: string;
-}
 interface ModelStatResponse {
   object: string;
   data: any;
diff --git a/cortex-js/src/infrastructure/constants/benchmark.ts b/cortex-js/src/infrastructure/constants/benchmark.ts
new file mode 100644
index 000000000..01dfc59bd
--- /dev/null
+++ b/cortex-js/src/infrastructure/constants/benchmark.ts
@@ -0,0 +1,37 @@
+import { BenchmarkConfig } from '../commanders/types/benchmark-config.interface';
+
+export const defaultBenchmarkConfiguration: BenchmarkConfig = {
+  api: {
+    base_url: 'http://127.0.0.1:1337/',
+    api_key: '<api_key>',
+    parameters: {
+      messages: [
+        {
+          content: 'You are a helpful assistant.',
+          role: 'system',
+        },
+        {
+          content: 'Hello!',
+          role: 'user',
+        },
+      ],
+      model: 'tinyllama',
+      stream: true,
+      max_tokens: 2048,
+      stop: [],
+      frequency_penalty: 0,
+      presence_penalty: 0,
+      temperature: 0.7,
+      top_p: 0.95,
+    },
+  },
+  prompts: {
+    min: 102,
+    max: 2048,
+    samples: 10,
+  },
+  output: 'json',
+  hardware: ['cpu', 'gpu', 'psu', 'chassis', 'ram'],
+  concurrency: 1,
+  num_rounds: 10,
+};