Loading page…
Preferred languages
More languages supported by Google Translate
Open the selector to load the current language catalog.
Translations are provided by Google Translate.
Loading page…
Preferred languages
More languages supported by Google Translate
Open the selector to load the current language catalog.
Translations are provided by Google Translate.
external benchmark
Decision Index 0.4 Retrieval. Every model the catalog carries a published Decision Index Retrieval value for, ranked by that value.
Decision Index 0.4 publishes chance-corrected retrieval scores for 209 evaluated systems.
120 models with a published Decision Index Retrieval value, ordered by that value, highest first. Models the source has not scored on this benchmark are not listed — they are unmeasured, not last.
| Rank | Model | Provider | Chance-corrected area score |
|---|---|---|---|
| 1 | Perplexity | 69.4 | |
| 2 | BAmammoth-1 | Baseten | 68.3 |
| 3 | BAelephant-1 | Baseten | 65.1 |
What the benchmark measures and how its value is published, from the source’s own record.
Decision Index Retrieval (Decision Index 0.4 Retrieval) is an external benchmark published by Decision Index leaderboard. Decision Index 0.4 publishes chance-corrected retrieval scores for 209 evaluated systems. Its evaluation set is described by the source as retrieval area of the Decision Index evaluation. The source rates its difficulty as retrieval and classification decision tasks.
Each model's result is published by Decision Index leaderboard as chance-corrected area score. ModelScale prints that value to 1 decimal place, exactly as published, and never converts or rescales it. This page ranks the 120 models that carry a published value, highest value first; a model the source never scored is left out rather than ranked at zero. The source does not state whether a higher value is the better result, so neither does this page.
| 4 | NLdeck31b (Gemma 4 31B) | Not listed | 64.8 |
| 4 | TAJev 1.14 | TypeSafe AI | 64.8 |
| 6 | Perplexity | 64.7 |
| 7 | FAsimple-jev · Qwen3.8-27B | Featherless AI | 63.5 |
| 8 | MADecider chat · Gemma-4-31B | Mapika | 63.4 |
| 9 | AUGEV-26B-Decide | autotrust | 63.3 |
| 10 | CHQuyet-1.0-Large | chinhnc | 63.2 |
| 10 | NLTorchcast Decision 27B | Not listed | 63.2 |
| 12 | PIBlink v0.3 26B-A4B (NVFP4) | PixilabAI | 63.1 |
| 13 | SUSurogate Rune 26B-A4B v3 | Surogate | 62.3 |
| 14 | Perplexity | 62.2 |
| 15 | NLJev | Not listed | 61.4 |
| 16 | NLvLLM-SR Decision 2.0 Vega 27B | Not listed | 61.1 |
| 16 | JUXor 26B-A4B | juspay | 61.1 |
| 18 | NLFastino GLiDE no-thinking (28B) | Not listed | 60.7 |
| 19 | CLClef | Cloudflare | 59.8 |
| 20 | CAEikos-27B | caiovicentino1 | 59.3 |
| 21 | NADrex v1.5 | Nace.AI | 59.2 |
| 22 | NADrex v2.0 | Nace.AI | 58.5 |
| 22 | AUJEV-27B | autotrust | 58.5 |
| 24 | KSreflex-27b | kshetrajna12 | 58.1 |
| 25 | NLXor 1.2 | Not listed | 57.9 |
| 26 | CLClef-Omni | Cloudflare | 56.5 |
| 26 | NLdecisio 0.4.0 (Gemma 4 12B) | Not listed | 56.5 |
| 28 | NLJADE | Not listed | 56.0 |
| 29 | FIJebadiah 27B | frontier-infra | 55.8 |
| 30 | JAKev 27B | jaredpalmer | 55.6 |
| 31 | ANAJev lora5 (Gemma 4 12B) | andyzhang232 | 55.3 |
| 32 | KIJPT-35B-A3B | kirp | 55.1 |
| 33 | ERWinnow-12B Q8 | Eldan Ring | 54.6 |
| 34 | NLvLLM-SR Decision 2.0 Lux 9B | Not listed | 54.3 |
| 35 | NLJoshuaSP diffusiongemma (open-jev) | Not listed | 54.2 |
| 36 | NLdecisio 0.4.0 (Qwen3.6-35B-A3B) | Not listed | 54.0 |
| 37 | NLDecider chat · Qwen3.6-27B | Not listed | 53.8 |
| 38 | NLJevfire | Not listed | 53.7 |
| 39 | BEBespoke Nimble 9B v3 | bespokelabs | 52.9 |
| 39 | NLHopper 12B (trained) | Not listed | 52.9 |
| 41 | NLvLLM-SR Decision 2.0 Nox 4B | Not listed | 52.3 |
| 42 | MAdecider-35b-a3b | Mapika | 52.1 |
| 43 | EVezjev 4B s2 | everettjf | 52.0 |
| 43 | ALJevK5 9B v0.3.3 | alibiserikbay | 52.0 |
| 45 | ELjiwo 4B | eljiwo | 51.3 |
| 46 | NLJev-Style 2B v3 | Not listed | 49.5 |
| 47 | AUJEV-9B | autotrust | 48.4 |
| 48 | SHRSI-Jev v5.0 VL 3B | shgao | 48.3 |
| 49 | MIJet v6.2 | michaljach | 47.8 |
| 50 | NLKev 9B v2 | Not listed | 47.4 |
| 51 | HOHopper (G) 1.2 | HopitAI | 46.6 |
| 52 | NLRyotide Qwen v0.3.1 | Not listed | 45.8 |
| 52 | SASieve 9B | sthanika-ai | 45.8 |
| 54 | NLJevK5 v0.3 | Not listed | 45.6 |
| 55 | MAdjev | Maisa | 44.8 |
| 56 | TONeoHorse-Jev-4B | TokenRhythm | 44.7 |
| 57 | NLLev1 3.7B | Not listed | 44.0 |
| 58 | ELjiwo 0.8B | eljiwo | 43.0 |
| 59 | KIJPT-9B | kirp | 42.3 |
| 59 | IAlev | interfaze-ai | 42.3 |
| 61 | MADecider 4B | Mapika | 42.2 |
| 62 | JAKev 4B r10 | jaredpalmer | 41.7 |
| 63 | CLClef-Flash | Cloudflare | 40.8 |
| 63 | KPKas 4B | kpiya | 40.8 |
| 63 | ELWinnow-E4B | EldanRing | 40.8 |
| 66 | KIJPT-4B | kirp | 40.4 |
| 66 | NLvLLM-SR Decision 2.0 Sol 2B | Not listed | 40.4 |
| 68 | internlm | 39.5 |
| 69 | AKJev-Omni | akhilaaa3 | 39.3 |
| 70 | DAGevva E2B | davidburhans | 38.5 |
| 71 | HLBosun v3.1 1.7B | Hanno-Labs | 37.9 |
| 71 | NLthis-that 1.2 | Not listed | 37.9 |
| 73 | HOBekko System One v0 400M | hotchpotch | 37.7 |
| 74 | NLrazorback16 openjev diffusiongemma (NVFP4, vLLM) | Not listed | 37.6 |
| 75 | NLJobe | Not listed | 37.5 |
| 76 | MAdecider-2b | Mapika | 36.7 |
| 77 | NLIntelif Qwen3-4B | Not listed | 36.6 |
| 77 | AMJevstral 8B | AmirBraham | 36.6 |
| 79 | BEBespoke Nimble 9B v2 | bespokelabs | 36.0 |
| 80 | LUDinah-0 | Lukitaduarte | 33.6 |
| 81 | NLvLLM-SR Decision 2.0 Eos 0.8B | Not listed | 33.1 |
| 82 | DAGevva E4B | davidburhans | 32.9 |
| 83 | NLEXAONE-4.0-1.2B-JEV v0.3 | Not listed | 31.2 |
| 84 | OMOneJev 0.8B | OmniJev | 30.9 |
| 85 | NLopenvons | Not listed | 30.0 |
| 86 | CAKodiak v0.2 1B (accuracy) | cortex-agent-llc | 29.9 |
| 87 | HOBekko System One v0 68M | hotchpotch | 28.5 |
| 88 | CLBosun v3.1 0.6B | Clause Logic | 28.1 |
| 89 | SASieve 2B | sthanika-ai | 28.0 |
| 90 | NLSolomon v1.1 | Not listed | 26.9 |
| 91 | NLvLLM-SR Decision 2.0 Kai 0.6B | Not listed | 26.8 |
| 92 | NLSifr 0.8B v3.1 | Not listed | 26.0 |
| 93 | NLopen-jev (pngwn) | Not listed | 24.0 |
| 94 | NLmmastrac diffusiongemma (vLLM PR 57250) | Not listed | 23.0 |
| 95 | NLJev-Style 0.8B v3 | Not listed | 21.7 |
| 96 | NLVon 1.3 | Not listed | 20.9 |
| 97 | ALJevK5 Lite | alibiserikbay | 20.3 |
| 98 | TOTev1-4B-experimental | togethercomputer | 18.5 |
| 99 | FAGLiNER2.5-Decide | fastino | 18.3 |
| 100 | NLSemIf | Not listed | 17.8 |
| 101 | internlm | 17.7 |
| 102 | LMjeff | Logan Markewich | 17.2 |
| 103 | MOLavoir | moganai | 17.1 |
| 104 | NLKev 0.8B r15 | Not listed | 16.7 |
| 105 | KIJPT-0.8B | kirp | 14.6 |
| 106 | HOBekko System One v0 17M | hotchpotch | 14.5 |
| 107 | W(metask-jev-4b | Wayfind (metask-ai) | 14.1 |
| 108 | NLmini-jev | Not listed | 10.0 |
| 109 | CILaya | Convai Innovations | 9.2 |
| 110 | NLsystem-one-gemma | Not listed | 8.8 |
| 111 | internlm | 8.5 |
| 112 | MOLFM2.5-2.6B-RLCD | monotykamary | 4.9 |
| 113 | CLCLM-v0.1-8B | Contrastive-LM | 4.6 |
| 114 | NLDM-JEPA 1.1 | Not listed | 4.3 |
| 115 | SUJulia 1 | SupersonicLabs | 3.5 |
| 116 | TOTev1-0.8B-experimental | togethercomputer | 3.4 |
| 117 | HAQwen-2.5-1B-RLCD | harshatheg | 1.0 |
| 118 | NOLFM2.5-350M-RLCD | notnotsamuel | 0.0 |
| 118 | NLVerdict | Not listed | 0.0 |
| 118 | NLZev | Not listed | 0.0 |
Evidence key: ObservedLast good
Rows are ordered by the value Decision Index leaderboard published, highest first. The source does not state whether a higher value is the better result, so this page does not either: for a benchmark that measures a rate of failure, read the table from the bottom.