diff --git a/src/benchmarks/README.md b/src/benchmarks/README.md
index 8d43cf9..46d634e 100644
--- a/src/benchmarks/README.md
+++ b/src/benchmarks/README.md
@@ -36,6 +36,7 @@ interface Benchmark {
 | `locomo` | GitHub snap-research/locomo | Long context memory benchmark |
 | `longmemeval` | HuggingFace xiaowu0162/longmemeval-cleaned | Long-term memory evaluation |
 | `convomem` | HuggingFace Salesforce/ConvoMem | Conversational memory benchmark |
+| `infinitebench` | HuggingFace xinrongzhang2022/InfiniteBench | Long-context understanding benchmark for testing retrieval, reasoning, and comprehension over extremely long inputs |
 
 ## Question Types
 
diff --git a/src/benchmarks/index.ts b/src/benchmarks/index.ts
index b790e87..f05e06c 100644
--- a/src/benchmarks/index.ts
+++ b/src/benchmarks/index.ts
@@ -2,11 +2,13 @@ import type { Benchmark, BenchmarkName } from "../types/benchmark"
 import { LoCoMoBenchmark } from "./locomo"
 import { LongMemEvalBenchmark } from "./longmemeval"
 import { ConvoMemBenchmark } from "./convomem"
+import { InfiniteBenchBenchmark } from "./infinitebench"
 
 const benchmarks: Record<BenchmarkName, new () => Benchmark> = {
   locomo: LoCoMoBenchmark,
   longmemeval: LongMemEvalBenchmark,
   convomem: ConvoMemBenchmark,
+  infinitebench: InfiniteBenchBenchmark,
 }
 
 export function createBenchmark(name: BenchmarkName): Benchmark {
@@ -21,4 +23,4 @@ export function getAvailableBenchmarks(): BenchmarkName[] {
   return Object.keys(benchmarks) as BenchmarkName[]
 }
 
-export { LoCoMoBenchmark, LongMemEvalBenchmark, ConvoMemBenchmark }
+export { LoCoMoBenchmark, LongMemEvalBenchmark, ConvoMemBenchmark, InfiniteBenchBenchmark }
diff --git a/src/benchmarks/infinitebench/index.ts b/src/benchmarks/infinitebench/index.ts
new file mode 100644
index 0000000..56a5680
--- /dev/null
+++ b/src/benchmarks/infinitebench/index.ts
@@ -0,0 +1,162 @@
+import { existsSync, readFileSync } from "fs"
+import { join } from "path"
+
+import type { Benchmark, BenchmarkConfig, QuestionFilter } from "../../types/benchmark"
+import type {
+  UnifiedMessage,
+  UnifiedQuestion,
+  UnifiedSession,
+  QuestionTypeRegistry,
+} from "../../types/unified"
+
+import type { InfiniteBenchItem } from "./types"
+import { logger } from "../../utils/logger"
+
+const DEFAULT_DATA_PATH = "./data/benchmarks/infinitebench"
+
+const TASKS = [
+  "passkey",
+  "kv_retrieval",
+  "number_string",
+  "code_run",
+  "code_debug",
+  "math_find",
+  "math_calc",
+  "longdialogue_qa_eng",
+  "longbook_qa_eng",
+  "longbook_sum_eng",
+  "longbook_choice_eng",
+  "longbook_qa_chn",
+] as const
+
+type InfiniteBenchTask = (typeof TASKS)[number]
+
+export const INFINITEBENCH_QUESTION_TYPES: QuestionTypeRegistry = Object.fromEntries(
+  TASKS.map((task) => [
+    task,
+    {
+      id: task,
+      alias: task,
+      description: `InfiniteBench ${task} task`,
+    },
+  ])
+)
+
+export class InfiniteBenchBenchmark implements Benchmark {
+  name = "infinitebench"
+
+  private questions: UnifiedQuestion[] = []
+  private sessions: Map<string, UnifiedSession[]> = new Map()
+
+  private dataPath = DEFAULT_DATA_PATH
+
+  async load(config?: BenchmarkConfig): Promise<void> {
+    this.dataPath = config?.dataPath ?? DEFAULT_DATA_PATH
+
+    const datasetPath = join(process.cwd(), this.dataPath)
+
+    if (!existsSync(datasetPath)) {
+      throw new Error(`InfiniteBench dataset missing: ${datasetPath}`)
+    }
+
+    this.loadDataset(datasetPath)
+  }
+
+  private loadDataset(datasetPath: string): void {
+    for (const task of TASKS) {
+      const file = join(datasetPath, `${task}.jsonl`)
+
+      if (!existsSync(file)) {
+        logger.warn(`Skipping missing task file: ${file}`)
+        continue
+      }
+
+      const rows = readFileSync(file, "utf-8").split("\n").filter(Boolean)
+
+      for (const row of rows) {
+        try {
+          const item: InfiniteBenchItem = JSON.parse(row)
+
+          this.addQuestion(item, task)
+        } catch (error) {
+          logger.error(`Failed parsing ${file}: ${error}`)
+        }
+      }
+    }
+
+    logger.info(`Loaded ${this.questions.length} InfiniteBench questions`)
+  }
+
+  private addQuestion(item: InfiniteBenchItem, task: InfiniteBenchTask): void {
+    const questionId = `infinitebench-${task}-${item.id}`
+
+    const session = this.createSession(item.context, questionId)
+
+    const question: UnifiedQuestion = {
+      questionId,
+
+      question: item.input,
+
+      questionType: task,
+
+      groundTruth: item.answer.at(0) ?? "",
+
+      haystackSessionIds: [session.sessionId],
+
+      metadata: {
+        task,
+        options: item.options,
+      },
+    }
+
+    this.questions.push(question)
+
+    this.sessions.set(questionId, [session])
+  }
+
+  private createSession(context: string, questionId: string): UnifiedSession {
+    const message: UnifiedMessage = {
+      role: "user",
+
+      content: context,
+    }
+
+    return {
+      sessionId: `${questionId}-session`,
+
+      messages: [message],
+    }
+  }
+
+  getQuestions(filter?: QuestionFilter): UnifiedQuestion[] {
+    let result = this.questions
+
+    if (filter?.questionTypes?.length) {
+      result = result.filter((q) => filter.questionTypes!.includes(q.questionType))
+    }
+
+    if (filter?.offset) {
+      result = result.slice(filter.offset)
+    }
+
+    if (filter?.limit) {
+      result = result.slice(0, filter.limit)
+    }
+
+    return result
+  }
+
+  getHaystackSessions(questionId: string): UnifiedSession[] {
+    return this.sessions.get(questionId) ?? []
+  }
+
+  getGroundTruth(questionId: string): string {
+    return this.questions.find((q) => q.questionId === questionId)?.groundTruth ?? ""
+  }
+
+  getQuestionTypes(): QuestionTypeRegistry {
+    return INFINITEBENCH_QUESTION_TYPES
+  }
+}
+
+export default InfiniteBenchBenchmark
\ No newline at end of file
diff --git a/src/benchmarks/infinitebench/types.ts b/src/benchmarks/infinitebench/types.ts
new file mode 100644
index 0000000..12139f8
--- /dev/null
+++ b/src/benchmarks/infinitebench/types.ts
@@ -0,0 +1,7 @@
+export interface InfiniteBenchItem {
+  id: number
+  context: string
+  input: string
+  answer: string[]
+  options: string[]
+}
\ No newline at end of file
diff --git a/src/cli/index.ts b/src/cli/index.ts
index b3c29d6..4fb20eb 100644
--- a/src/cli/index.ts
+++ b/src/cli/index.ts
@@ -150,11 +150,16 @@ Available benchmark datasets for evaluation:
   convomem       ConvoMem - Conversational memory benchmark
                  Tests: user facts, assistant facts, preferences, implicit connections
                  Source: HuggingFace Salesforce/ConvoMem (downloaded on first use)
+  
+  infinitebench  InfiniteBench - Ultra-long context understanding benchmark
+                 Tests: document QA, long-context retrieval, code understanding, multi-hop reasoning
+                 Source: GitHub OpenBMB/InfiniteBench (downloaded on first use)
 
 Usage:
   -b locomo        Run LoCoMo benchmark
   -b longmemeval   Run LongMemEval benchmark
   -b convomem      Run ConvoMem benchmark
+  -b infinitebench Run InfiniteBench benchmark
 `)
 }
 
diff --git a/src/types/benchmark.ts b/src/types/benchmark.ts
index 07961e4..c73b384 100644
--- a/src/types/benchmark.ts
+++ b/src/types/benchmark.ts
@@ -20,4 +20,4 @@ export interface Benchmark {
   getQuestionTypes(): QuestionTypeRegistry
 }
 
-export type BenchmarkName = "locomo" | "longmemeval" | "convomem"
+export type BenchmarkName = "locomo" | "longmemeval" | "convomem" | "infinitebench"