Back to Benchmarks
RULER
RULER: What's the Real Context Size of Your Long-Context Language Models
Benchmark Metadata
PublisherNVIDIA
VenueCOLM 2024
Evaluation Typeautomatic
Dimensions13
Test Prompts4,000
ScoringHigher is better
Update Frequencyannual
PaperView Paper
LeaderboardView Leaderboard
What It Measures
- Single and multi-key needle retrieval
- Variable tracking
- Common and frequent word extraction
- Question answering with long contexts
- Effective context length
What It Does Not Measure
- Multi-session consistency
- Personalization
- Generation quality
All Systems Evaluated(71 systems)
| Rank | System | Score |
|---|---|---|
| #1 | Titanslucidrains (community) / paper by Google Research | 97.9 |
| #2 | PineconePinecone Systems | 82.5 |
| #3 | WeaviateWeaviate | 81.5 |
| #4 | Recurrent Memory TransformerMIPT / DeepPavlov (Bulatov, Kuratov, Burtsev) | 79.5 |
| #5 | MilvusZilliz | 78.2 |
| #6 | QdrantQdrant | 77.9 |
| #7 | MambaCMU / Princeton (Gu, Dao) | 77.8 |
| #8 | Jina AI EmbeddingsJina AI GmbH | 76.1 |
| #9 | RWKVRWKV Foundation / BlinkDL community | 75.8 |
| #10 | txtaiNeuML | 75.7 |
| #11 | Landmark AttentionEPFL (Mohtashami, Jaggi) | 75.6 |
| #12 | Compressive TransformerDeepMind (Rae et al.) | 75.4 |
| #13 | Cohere EmbedCohere Inc. | 75 |
| #14 | MiniRAGHKUDS | 74.8 |
| #15 | LM-InfiniteIllinois / Meta (Han et al.) | 74.5 |
| #16 | H2OUT Austin / Rice / CMU / Stanford / Meta (Zhang et al.) | 74.4 |
| #17 | ChromaChroma | 74.2 |
| #18 | MarkerDatalab (datalab-to) | 74.1 |
| #19 | AllegroGraphFranz Inc. | 74 |
| #20 | PathRAGBUPT-GAMMA | 73.6 |
| #21 | TRIMEPrinceton NLP (Zhong, Lei, Chen) | 73.6 |
| #22 | LanceDBLanceDB Inc. (YC S22) | 73.3 |
| #23 | R2RSciPhi-AI | 73.3 |
| #24 | ScissorhandsRice / Stanford / Meta (Liu et al.) | 73.2 |
| #25 | ∞ FormerInstituto de Telecomunicações / DeepMind / IST (Martins, Marinho, Martins) | 72.7 |
| #26 | KAGOpenSPG / Ant Group | 72.5 |
| #27 | GraphRAGMicrosoft | 72.4 |
| #28 | ICAEMicrosoft Research (Ge et al.) | 72.3 |
| #29 | Neo4j LLM Graph BuilderNeo4j Labs | 72.3 |
| #30 | Mixedbread AIMixedbread AI | 71.9 |
| #31 | Voyage AIVoyage AI (acquired by MongoDB, Feb 2025) | 71.9 |
| #32 | Ontotext GraphDBOntotext / Graphwise (merged with Semantic Web Company, 2025) | 71.6 |
| #33 | Supabase VectorSupabase Inc. | 71.4 |
| #34 | Nano GraphRAGgusye1234 | 71.3 |
| #35 | DiffbotDiffbot Inc. | 71.1 |
| #36 | SingleStore VectorSingleStore Inc. | 71 |
| #37 | Activeloop Deep LakeActiveloop Inc. | 70.7 |
| #38 | LightRAGHKUDS (HKU Data Intelligence Lab) | 70.7 |
| #39 | RAPTORStanford (Sarthi, Abdullah et al.) | 70.1 |
| #40 | AtlasMeta AI FAIR (Izacard et al.) | 69.9 |
| #41 | StardogStardog Union Inc. | 69.9 |
| #42 | Unstructured IOUnstructured Technologies Inc. | 69.9 |
| #43 | MyScaleMyScale Inc. | 69.5 |
| #44 | Elasticsearch VectorElastic N.V. | 69.4 |
| #45 | ValdYahoo Japan | 69.4 |
| #46 | pgvector Supabase Neonpgvector OSS / Supabase Inc. / Neon Inc. | 69.3 |
| #47 | PrivateGPTZylon AI | 69.3 |
| #48 | OpenSearch VectorOpenSearch Project (AWS-led) | 69.2 |
| #49 | PaperQA2FutureHouse | 69.1 |
| #50 | RETRODeepMind (Borgeaud et al.) | 68.9 |
| #51 | Neon VectorNeon Inc. | 68.8 |
| #52 | TrustRAGGoMate Community | 68.7 |
| #53 | REALMGoogle Research (Guu et al.) | 68.6 |
| #54 | SelfmemTsinghua / Microsoft (Cheng et al.) | 68.4 |
| #55 | Carbon AICarbon (acquired by Perplexity, Dec 2024) | 68.3 |
| #56 | GraphRAG-SDKFalkorDB | 68.3 |
| #57 | LlamaCloudLlamaIndex Inc. | 67.7 |
| #58 | MarqoMarqo Pty Ltd | 67.6 |
| #59 | Astra DBDataStax | 67.1 |
| #60 | Activation BeaconBAAI / Renmin University (Zhang et al.) | 66.7 |
| #61 | Vespa AIYahoo / Vespa.ai (independent OSS project) | 66.6 |
| #62 | vectorizeVectorize Inc. | 66.5 |
| #63 | Manticore SearchManticore Software Ltd. | 66.4 |
| #64 | VectaraVectara Inc. | 66.4 |
| #65 | ColPaliilluin-tech | 66.2 |
| #66 | ParadeDBParadeDB Inc. (YC S23) | 66.1 |
| #67 | CognitaTrueFoundry | 66 |
| #68 | VerbaWeaviate | 66 |
| #69 | TurboPufferTurboPuffer Inc. | 65.6 |
| #70 | MendableMendable (YC-backed) | 64.5 |
| #71 | StreamingLLMMIT Han Lab / Meta AI (Xiao et al.) | 57.2 |