diff --git a/mteb/benchmarks/benchmarks.py b/mteb/benchmarks/benchmarks.py
index 2c2a0ed6b2..f2aef081ca 100644
--- a/mteb/benchmarks/benchmarks.py
+++ b/mteb/benchmarks/benchmarks.py
@@ -1999,3 +1999,33 @@
 }
 """,
 )
+
+R2MED = Benchmark(
+    name="R2MED",
+    display_name="Reasoning-driven medical retrieval",
+    tasks=get_tasks(
+        tasks=[
+            "R2MEDBiologyRetrieval",
+            "R2MEDBioinformaticsRetrieval",
+            "R2MEDMedicalSciencesRetrieval",
+            "R2MEDMedXpertQAExamRetrieval",
+            "R2MEDMedQADiagRetrieval",
+            "R2MEDPMCTreatmentRetrieval",
+            "R2MEDPMCClinicalRetrieval",
+            "R2MEDIIYiClinicalRetrieval",
+        ]
+    ),
+    description="""R2MED: First Reasoning-Driven Medical Retrieval Benchmark.
+    R2MED is a high-quality, high-resolution information retrieval (IR) dataset designed for medical scenarios. 
+    It contains 876 queries with three retrieval tasks, five medical scenarios, and twelve body systems.
+    """,
+    reference="https://r2med.github.io/",
+    citation=r"""
+@article{li2025r2med,
+  author = {Li, Lei and Zhou, Xiao and Liu, Zheng},
+  journal = {arXiv preprint arXiv:2505.14558},
+  title = {R2MED: A Benchmark for Reasoning-Driven Medical Retrieval},
+  year = {2025},
+}
+""",
+)
diff --git a/mteb/tasks/Retrieval/__init__.py b/mteb/tasks/Retrieval/__init__.py
index b2abdd19e5..39932c47ec 100644
--- a/mteb/tasks/Retrieval/__init__.py
+++ b/mteb/tasks/Retrieval/__init__.py
@@ -87,6 +87,7 @@
 from .eng.PiqaRetrieval import *
 from .eng.QuailRetrieval import *
 from .eng.QuoraRetrieval import *
+from .eng.R2MEDRetrieval import *
 from .eng.RARbCodeRetrieval import *
 from .eng.RARbMathRetrieval import *
 from .eng.SCIDOCSRetrieval import *
diff --git a/mteb/tasks/Retrieval/eng/R2MEDRetrieval.py b/mteb/tasks/Retrieval/eng/R2MEDRetrieval.py
new file mode 100644
index 0000000000..9cac3cb0c0
--- /dev/null
+++ b/mteb/tasks/Retrieval/eng/R2MEDRetrieval.py
@@ -0,0 +1,393 @@
+from __future__ import annotations
+
+from collections import defaultdict
+
+import datasets
+
+from mteb.abstasks.AbsTaskRetrieval import AbsTaskRetrieval
+from mteb.abstasks.TaskMetadata import TaskMetadata
+
+
+def load_r2med_data(
+    path: str,
+    eval_splits: list,
+    cache_dir: str,
+    revision: str,
+):
+    eval_split = eval_splits[0]
+    corpus = {eval_split: None}
+    queries = {eval_split: None}
+    relevant_docs = {eval_split: None}
+    domain_corpus = datasets.load_dataset(
+        path, name="corpus", split="corpus", cache_dir=cache_dir, revision=revision
+    )
+    domain_queries = datasets.load_dataset(
+        path, name="query", split="query", cache_dir=cache_dir, revision=revision
+    )
+    domain_qrels = datasets.load_dataset(
+        path, name="qrels", split="qrels", cache_dir=cache_dir, revision=revision
+    )
+    corpus[eval_split] = {e["id"]: {"text": e["text"]} for e in domain_corpus}
+    queries[eval_split] = {e["id"]: e["text"] for e in domain_queries}
+    relevant_docs[eval_split] = defaultdict(dict)
+    for e in domain_qrels:
+        qid = e["q_id"]
+        pid = e["p_id"]
+        relevant_docs[eval_split][qid][pid] = int(e["score"])
+
+    corpus = datasets.DatasetDict(corpus)
+    queries = datasets.DatasetDict(queries)
+    relevant_docs = datasets.DatasetDict(relevant_docs)
+    return corpus, queries, relevant_docs
+
+
+class R2MEDBiologyRetrieval(AbsTaskRetrieval):
+    metadata = TaskMetadata(
+        name="R2MEDBiologyRetrieval",
+        dataset={
+            "path": "R2MED/Biology",
+            "revision": "8b9fec2db9eda4b5742d03732213fbaee8169556",
+        },
+        reference="https://huggingface.co/R2MED/Biology",
+        description="Biology retrieval dataset.",
+        type="Retrieval",
+        category="s2p",
+        eval_splits=["test"],
+        eval_langs=["eng-Latn"],
+        main_score="ndcg_at_10",
+        domains=["Medical"],
+        task_subtypes=["Article retrieval"],
+        license="cc-by-4.0",
+        annotations_creators="LM-generated and reviewed",
+        dialect=[],
+        sample_creation="found",
+        modalities=["text"],
+        bibtex_citation=r"""
+@article{li2025r2med,
+  author = {Li, Lei and Zhou, Xiao and Liu, Zheng},
+  journal = {arXiv preprint arXiv:2505.14558},
+  title = {R2MED: A Benchmark for Reasoning-Driven Medical Retrieval},
+  year = {2025},
+}
+""",
+    )
+
+    def load_data(self, **kwargs):
+        if self.data_loaded:
+            return
+
+        self.corpus, self.queries, self.relevant_docs = load_r2med_data(
+            path=self.metadata.dataset["path"],
+            eval_splits=self.metadata.eval_splits,
+            cache_dir=kwargs.get("cache_dir", None),
+            revision=self.metadata.dataset["revision"],
+        )
+        self.data_loaded = True
+
+
+class R2MEDBioinformaticsRetrieval(AbsTaskRetrieval):
+    metadata = TaskMetadata(
+        name="R2MEDBioinformaticsRetrieval",
+        dataset={
+            "path": "R2MED/Bioinformatics",
+            "revision": "6021fce366892cbfd7837fa85a4128ea93315e18",
+        },
+        reference="https://huggingface.co/R2MED/Bioinformatics",
+        description="Bioinformatics retrieval dataset.",
+        type="Retrieval",
+        category="s2p",
+        eval_splits=["test"],
+        eval_langs=["eng-Latn"],
+        main_score="ndcg_at_10",
+        domains=["Medical"],
+        task_subtypes=["Article retrieval"],
+        license="cc-by-4.0",
+        annotations_creators="LM-generated and reviewed",
+        dialect=[],
+        sample_creation="found",
+        modalities=["text"],
+        bibtex_citation=r"""
+@article{li2025r2med,
+  author = {Li, Lei and Zhou, Xiao and Liu, Zheng},
+  journal = {arXiv preprint arXiv:2505.14558},
+  title = {R2MED: A Benchmark for Reasoning-Driven Medical Retrieval},
+  year = {2025},
+}
+""",
+    )
+
+    def load_data(self, **kwargs):
+        if self.data_loaded:
+            return
+
+        self.corpus, self.queries, self.relevant_docs = load_r2med_data(
+            path=self.metadata.dataset["path"],
+            eval_splits=self.metadata.eval_splits,
+            cache_dir=kwargs.get("cache_dir", None),
+            revision=self.metadata.dataset["revision"],
+        )
+        self.data_loaded = True
+
+
+class R2MEDMedicalSciencesRetrieval(AbsTaskRetrieval):
+    metadata = TaskMetadata(
+        name="R2MEDMedicalSciencesRetrieval",
+        dataset={
+            "path": "R2MED/Medical-Sciences",
+            "revision": "1b48911514c80bf9182222d99752ad75e23b4b47",
+        },
+        reference="https://huggingface.co/R2MED/Medical-Sciences",
+        description="Medical-Sciences retrieval dataset.",
+        type="Retrieval",
+        category="s2p",
+        eval_splits=["test"],
+        eval_langs=["eng-Latn"],
+        main_score="ndcg_at_10",
+        domains=["Medical"],
+        task_subtypes=["Article retrieval"],
+        license="cc-by-4.0",
+        annotations_creators="LM-generated and reviewed",
+        dialect=[],
+        sample_creation="found",
+        modalities=["text"],
+        bibtex_citation=r"""
+@article{li2025r2med,
+  author = {Li, Lei and Zhou, Xiao and Liu, Zheng},
+  journal = {arXiv preprint arXiv:2505.14558},
+  title = {R2MED: A Benchmark for Reasoning-Driven Medical Retrieval},
+  year = {2025},
+}
+""",
+    )
+
+    def load_data(self, **kwargs):
+        if self.data_loaded:
+            return
+
+        self.corpus, self.queries, self.relevant_docs = load_r2med_data(
+            path=self.metadata.dataset["path"],
+            eval_splits=self.metadata.eval_splits,
+            cache_dir=kwargs.get("cache_dir", None),
+            revision=self.metadata.dataset["revision"],
+        )
+        self.data_loaded = True
+
+
+class R2MEDMedXpertQAExamRetrieval(AbsTaskRetrieval):
+    metadata = TaskMetadata(
+        name="R2MEDMedXpertQAExamRetrieval",
+        dataset={
+            "path": "R2MED/MedXpertQA-Exam",
+            "revision": "b457ea43db9ae5db74c3a3e5be0a213d0f85ac3a",
+        },
+        reference="https://huggingface.co/R2MED/MedXpertQA-Exam",
+        description="MedXpertQA-Exam retrieval dataset.",
+        type="Retrieval",
+        category="s2p",
+        eval_splits=["test"],
+        eval_langs=["eng-Latn"],
+        main_score="ndcg_at_10",
+        domains=["Medical"],
+        task_subtypes=["Article retrieval"],
+        license="cc-by-4.0",
+        annotations_creators="LM-generated and reviewed",
+        dialect=[],
+        sample_creation="found",
+        modalities=["text"],
+        bibtex_citation=r"""
+@article{li2025r2med,
+  author = {Li, Lei and Zhou, Xiao and Liu, Zheng},
+  journal = {arXiv preprint arXiv:2505.14558},
+  title = {R2MED: A Benchmark for Reasoning-Driven Medical Retrieval},
+  year = {2025},
+}
+""",
+    )
+
+    def load_data(self, **kwargs):
+        if self.data_loaded:
+            return
+
+        self.corpus, self.queries, self.relevant_docs = load_r2med_data(
+            path=self.metadata.dataset["path"],
+            eval_splits=self.metadata.eval_splits,
+            cache_dir=kwargs.get("cache_dir", None),
+            revision=self.metadata.dataset["revision"],
+        )
+        self.data_loaded = True
+
+
+class R2MEDMedQADiagRetrieval(AbsTaskRetrieval):
+    metadata = TaskMetadata(
+        name="R2MEDMedQADiagRetrieval",
+        dataset={
+            "path": "R2MED/MedQA-Diag",
+            "revision": "78b585990279cc01a493f876c1b0cf09557fba57",
+        },
+        reference="https://huggingface.co/R2MED/MedQA-Diag",
+        description="MedQA-Diag retrieval dataset.",
+        type="Retrieval",
+        category="s2p",
+        eval_splits=["test"],
+        eval_langs=["eng-Latn"],
+        main_score="ndcg_at_10",
+        domains=["Medical"],
+        task_subtypes=["Article retrieval"],
+        license="cc-by-4.0",
+        annotations_creators="LM-generated and reviewed",
+        dialect=[],
+        sample_creation="found",
+        modalities=["text"],
+        bibtex_citation=r"""
+@article{li2025r2med,
+  author = {Li, Lei and Zhou, Xiao and Liu, Zheng},
+  journal = {arXiv preprint arXiv:2505.14558},
+  title = {R2MED: A Benchmark for Reasoning-Driven Medical Retrieval},
+  year = {2025},
+}
+""",
+    )
+
+    def load_data(self, **kwargs):
+        if self.data_loaded:
+            return
+
+        self.corpus, self.queries, self.relevant_docs = load_r2med_data(
+            path=self.metadata.dataset["path"],
+            eval_splits=self.metadata.eval_splits,
+            cache_dir=kwargs.get("cache_dir", None),
+            revision=self.metadata.dataset["revision"],
+        )
+        self.data_loaded = True
+
+
+class R2MEDPMCTreatmentRetrieval(AbsTaskRetrieval):
+    metadata = TaskMetadata(
+        name="R2MEDPMCTreatmentRetrieval",
+        dataset={
+            "path": "R2MED/PMC-Treatment",
+            "revision": "53c489a44a3664ba352c07550b72b4525a5968d5",
+        },
+        reference="https://huggingface.co/R2MED/PMC-Treatment",
+        description="PMC-Treatment retrieval dataset.",
+        type="Retrieval",
+        category="s2p",
+        eval_splits=["test"],
+        eval_langs=["eng-Latn"],
+        main_score="ndcg_at_10",
+        domains=["Medical"],
+        task_subtypes=["Article retrieval"],
+        license="cc-by-4.0",
+        annotations_creators="LM-generated and reviewed",
+        dialect=[],
+        sample_creation="found",
+        modalities=["text"],
+        bibtex_citation=r"""
+@article{li2025r2med,
+  author = {Li, Lei and Zhou, Xiao and Liu, Zheng},
+  journal = {arXiv preprint arXiv:2505.14558},
+  title = {R2MED: A Benchmark for Reasoning-Driven Medical Retrieval},
+  year = {2025},
+}
+""",
+    )
+
+    def load_data(self, **kwargs):
+        if self.data_loaded:
+            return
+
+        self.corpus, self.queries, self.relevant_docs = load_r2med_data(
+            path=self.metadata.dataset["path"],
+            eval_splits=self.metadata.eval_splits,
+            cache_dir=kwargs.get("cache_dir", None),
+            revision=self.metadata.dataset["revision"],
+        )
+        self.data_loaded = True
+
+
+class R2MEDPMCClinicalRetrieval(AbsTaskRetrieval):
+    metadata = TaskMetadata(
+        name="R2MEDPMCClinicalRetrieval",
+        dataset={
+            "path": "R2MED/PMC-Clinical",
+            "revision": "812829522f7eaa407ef82b96717be85788a50f7e",
+        },
+        reference="https://huggingface.co/R2MED/PMC-Clinical",
+        description="PMC-Clinical retrieval dataset.",
+        type="Retrieval",
+        category="s2p",
+        eval_splits=["test"],
+        eval_langs=["eng-Latn"],
+        main_score="ndcg_at_10",
+        domains=["Medical"],
+        task_subtypes=["Article retrieval"],
+        license="cc-by-4.0",
+        annotations_creators="LM-generated and reviewed",
+        dialect=[],
+        sample_creation="found",
+        modalities=["text"],
+        bibtex_citation=r"""
+@article{li2025r2med,
+  author = {Li, Lei and Zhou, Xiao and Liu, Zheng},
+  journal = {arXiv preprint arXiv:2505.14558},
+  title = {R2MED: A Benchmark for Reasoning-Driven Medical Retrieval},
+  year = {2025},
+}
+""",
+    )
+
+    def load_data(self, **kwargs):
+        if self.data_loaded:
+            return
+
+        self.corpus, self.queries, self.relevant_docs = load_r2med_data(
+            path=self.metadata.dataset["path"],
+            eval_splits=self.metadata.eval_splits,
+            cache_dir=kwargs.get("cache_dir", None),
+            revision=self.metadata.dataset["revision"],
+        )
+        self.data_loaded = True
+
+
+class R2MEDIIYiClinicalRetrieval(AbsTaskRetrieval):
+    metadata = TaskMetadata(
+        name="R2MEDIIYiClinicalRetrieval",
+        dataset={
+            "path": "R2MED/IIYi-Clinical",
+            "revision": "974abbc9bc281c3169180a6aa5d7586cfd2f5877",
+        },
+        reference="https://huggingface.co/R2MED/IIYi-Clinical",
+        description="IIYi-Clinical retrieval dataset.",
+        type="Retrieval",
+        category="s2p",
+        eval_splits=["test"],
+        eval_langs=["eng-Latn"],
+        main_score="ndcg_at_10",
+        domains=["Medical"],
+        task_subtypes=["Article retrieval"],
+        license="cc-by-4.0",
+        annotations_creators="LM-generated and reviewed",
+        dialect=[],
+        sample_creation="found",
+        modalities=["text"],
+        bibtex_citation=r"""
+@article{li2025r2med,
+  author = {Li, Lei and Zhou, Xiao and Liu, Zheng},
+  journal = {arXiv preprint arXiv:2505.14558},
+  title = {R2MED: A Benchmark for Reasoning-Driven Medical Retrieval},
+  year = {2025},
+}
+""",
+    )
+
+    def load_data(self, **kwargs):
+        if self.data_loaded:
+            return
+
+        self.corpus, self.queries, self.relevant_docs = load_r2med_data(
+            path=self.metadata.dataset["path"],
+            eval_splits=self.metadata.eval_splits,
+            cache_dir=kwargs.get("cache_dir", None),
+            revision=self.metadata.dataset["revision"],
+        )
+        self.data_loaded = True