Benchmark Almanac

OpenLegalRetrieval-DE

SauerkrautLM Legal-API 0.55 tops its leaderboard Qwen3-Embedding-8B 0.44 leads the models we track
Not in model cards

Models retrieve German statutes and court decisions across seven tasks. Each task scores where relevant documents appear in the first ten results. Each task carries equal weight.

VAGO reports 23,620 queries. Two tasks use GerLayQA and GerDaLIRSmall. The full leaderboard also ranks models across other benchmarks.

Made by
VAGO Solutions
Released
November 2026
Kind
Law
Level
University
Scored
Checked automatically
Answers
Short answer
Tasks
All public
Measures
Capability
Subject
Law
Language
German
In and out
Text → Text
Interaction
Single turn
Refresh
One fixed set
Run by
Anyone
In views
Law and finance
Last model added
October 2026
Scores refreshed
November 20th, 2026
33
Entries on VAGO Solutions’s leaderboard, 4 models we track

SauerkrautLM Legal-API tops the leaderboard with 0.55 Qwen3-Embedding-8B leads with 0.44 among models released by October 2026

Best score per entry from VAGO Solutions’s leaderboard, read on November 20th, 2026. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. The best model we track is Qwen3-Embedding-8B, at 0.44. This view shows only models we track, each at its best.

The newest model with a score here came out in October 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 SauerkrautLM Legal-API 0.55
  2. 2 Nemotron-3-Embed-8B st / bf16 / 8192 tokens 0.52
  3. 3 Nemotron-3.5-Embed-8B st / bf16 / 8192 tokens 0.52
  4. 4 dinghy-law-8b-v1 ⚠ st / bf16 / 8192 tokens 0.52
  5. 5 SauerkrautLM-MoEdernColBERT-115M-Legal-DE 0.48
  6. 6 most-embed-de st / bf16 / 8192 tokens 0.48
  7. 7 Nemotron-3-Embed-1B st / bf16 / 8192 tokens 0.47
  8. 8 Nemotron-3.5-Embed-1B st / bf16 / 8192 tokens 0.47
  9. 9 SauerkrautLM-MoEdernColBERT-27M-Legal-DE 0.46
  10. 10 Euler-Legal-Embedding-V1 qwen3inst / bf16 / 8192 tokens 0.45
  11. 11 octen-law-8b-v1 ⚠ qwen3inst / bf16 / 8192 tokens 0.45
  12. 12 dinghy-law-4b-v1 ⚠ qwen3inst / bf16 / 8192 tokens 0.44
  13. 13 Qwen3-Embedding-8B qwen3inst / bf16 / 8192 tokens 0.44
  14. 14 MoD-Embedding qwen3inst / bf16 / 8192 tokens 0.43
  15. 15 Qwen3-Embedding-4B qwen3inst / bf16 / 8192 tokens 0.42
  16. 16 greenleaf-law-embed-tiny st / bf16 / 8192 tokens 0.41
  17. 17 jina-embeddings-v5-text-small st / bf16 / 8192 tokens 0.39
  18. 18 SauerkrautLM-MoEdernColBERT-1.5M-Legal-DE 0.39
  19. 19 BM25 0.38
  20. 20 jina-embeddings-v5-text-nano st / bf16 / 8192 tokens 0.38
  21. 21 bge-multilingual-gemma2 gemma2inst / bf16 / 8192 tokens 0.38
  22. 22 litil-embed-0.6b ⚠ qwen3inst / bf16 / 8192 tokens 0.37
  23. 23 embeddinggemma-300m st / bf16 / 8192 tokens 0.35
  24. 24 dinghy-law-0.6b-v1 ⚠ qwen3inst / bf16 / 8192 tokens 0.34
  25. 25 EmbeddingGemma 2 st / bf16 / 8192 tokens 0.32
  26. 26 Qwen3-Embedding-0.6B qwen3inst / bf16 / 8192 tokens 0.32
  27. 27 bge-m3 st / bf16 / 8192 tokens 0.32
  28. 28 multilingual-e5-large-instruct e5inst / bf16 / 512 tokens 0.31
  29. 29 multilingual-e5-base e5 / bf16 / 512 tokens 0.27
  30. 30 multilingual-e5-small e5 / bf16 / 512 tokens 0.26
  31. 31 granite-embedding-278m-multilingual st / bf16 / 512 tokens 0.25
  32. 32 granite-embedding-107m-multilingual st / bf16 / 512 tokens 0.23
  33. 33 paraphrase-multilingual-MiniLM-L12-v2 st / bf16 / 128 tokens 0.13

Source: VAGO Solutions’s leaderboard

#EntryScoreSettingDateLinkSource
1 SauerkrautLM Legal-API 1.1B 0.55 huggingface.co Official leaderboard
2 Nemotron-3-Embed-8B open weights, 8B 0.52 st / bf16 / 8192 tokens huggingface.co Official leaderboard
3 Nemotron-3.5-Embed-8B open weights, 8B 0.52 st / bf16 / 8192 tokens huggingface.co Official leaderboard
4 dinghy-law-8b-v1 ⚠ open weights, 8B 0.52 st / bf16 / 8192 tokens huggingface.co Official leaderboard
5 SauerkrautLM-MoEdernColBERT-115M-Legal-DE open weights, 0.1B 0.48 huggingface.co Official leaderboard
6 most-embed-de open weights, 1.1B 0.48 st / bf16 / 8192 tokens huggingface.co Official leaderboard
7 Nemotron-3-Embed-1B open weights, 1.1B 0.47 st / bf16 / 8192 tokens huggingface.co Official leaderboard
8 Nemotron-3.5-Embed-1B open weights, 1.1B 0.47 st / bf16 / 8192 tokens huggingface.co Official leaderboard
9 SauerkrautLM-MoEdernColBERT-27M-Legal-DE open weights, 0B 0.46 huggingface.co Official leaderboard
10 Euler-Legal-Embedding-V1 open weights, 8B 0.45 qwen3inst / bf16 / 8192 tokens huggingface.co Official leaderboard
11 octen-law-8b-v1 ⚠ open weights, 8B 0.45 qwen3inst / bf16 / 8192 tokens huggingface.co Official leaderboard
12 dinghy-law-4b-v1 ⚠ open weights, 4B 0.44 qwen3inst / bf16 / 8192 tokens huggingface.co Official leaderboard
13 Qwen3-Embedding-8B 0.44 qwen3inst / bf16 / 8192 tokens Jun 2025 huggingface.co Official leaderboard
14 MoD-Embedding open weights, 4B 0.43 qwen3inst / bf16 / 8192 tokens huggingface.co Official leaderboard
15 Qwen3-Embedding-4B 0.42 qwen3inst / bf16 / 8192 tokens Jun 2025 huggingface.co Official leaderboard
16 greenleaf-law-embed-tiny open weights, 0.6B 0.41 st / bf16 / 8192 tokens huggingface.co Official leaderboard
17 jina-embeddings-v5-text-small open weights, 0.7B 0.39 st / bf16 / 8192 tokens huggingface.co Official leaderboard
18 SauerkrautLM-MoEdernColBERT-1.5M-Legal-DE open weights, 0B 0.39 huggingface.co Official leaderboard
19 BM25 open weights, 0B 0.38 huggingface.co Official leaderboard
20 jina-embeddings-v5-text-nano open weights, 0.2B 0.38 st / bf16 / 8192 tokens huggingface.co Official leaderboard
21 bge-multilingual-gemma2 open weights, 9B 0.38 gemma2inst / bf16 / 8192 tokens huggingface.co Official leaderboard
22 litil-embed-0.6b ⚠ open weights, 0.6B 0.37 qwen3inst / bf16 / 8192 tokens huggingface.co Official leaderboard
23 embeddinggemma-300m open weights, 0.3B 0.35 st / bf16 / 8192 tokens huggingface.co Official leaderboard
24 dinghy-law-0.6b-v1 ⚠ open weights, 0.6B 0.34 qwen3inst / bf16 / 8192 tokens huggingface.co Official leaderboard
25 EmbeddingGemma 2 0.32 st / bf16 / 8192 tokens Oct 2026 huggingface.co Official leaderboard
26 Qwen3-Embedding-0.6B 0.32 qwen3inst / bf16 / 8192 tokens Jun 2025 huggingface.co Official leaderboard
27 bge-m3 open weights, 0.6B 0.32 st / bf16 / 8192 tokens huggingface.co Official leaderboard
28 multilingual-e5-large-instruct open weights, 0.6B 0.31 e5inst / bf16 / 512 tokens huggingface.co Official leaderboard
29 multilingual-e5-base open weights, 0.3B 0.27 e5 / bf16 / 512 tokens huggingface.co Official leaderboard
30 multilingual-e5-small open weights, 0.1B 0.26 e5 / bf16 / 512 tokens huggingface.co Official leaderboard
31 granite-embedding-278m-multilingual open weights, 0.3B 0.25 st / bf16 / 512 tokens huggingface.co Official leaderboard
32 granite-embedding-107m-multilingual open weights, 0.1B 0.23 st / bf16 / 512 tokens huggingface.co Official leaderboard
33 paraphrase-multilingual-MiniLM-L12-v2 open weights, 0.1B 0.13 st / bf16 / 128 tokens huggingface.co Official leaderboard
#ModelScoreSettingDateLinkSource
1 Qwen3-Embedding-8B 0.44 qwen3inst / bf16 / 8192 tokens Jun 2025 huggingface.co Official leaderboard
2 Qwen3-Embedding-4B 0.42 qwen3inst / bf16 / 8192 tokens Jun 2025 huggingface.co Official leaderboard
3 EmbeddingGemma 2 0.32 st / bf16 / 8192 tokens Oct 2026 huggingface.co Official leaderboard
4 Qwen3-Embedding-0.6B 0.32 qwen3inst / bf16 / 8192 tokens Jun 2025 huggingface.co Official leaderboard

Number one over time

Lab claims count from the day the card came out, independent runs from the model’s release.

Qwen
VAGO Solutions leaderboard 1 leader

Source: VAGO Solutions’s leaderboard

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

Plot with other benchmarks
On VAGO Solutions’s leaderboard
0.35 0.40 0.45 2025 2026 0.35 0.40 0.45 2025 2026

Source: VAGO Solutions’s leaderboard

No lab model card reports it yet

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.

More in Law, finance and medicine

All 637 in Law, finance and medicine
  • MedXpertQA-MM In use Models answer expert-level medical exam questions that combine images with clinical records and test results. Scoring is accuracy. Qwen 3.8 Max 80.4%
  • HealthBench Professional In use 525 physician-authored tasks from real clinician chats (care consult, documentation, research) with physician-adjudicated rubrics. Claude Sonnet 5.5 69.2%
  • HealthBench In use 5,000 realistic health conversations graded against physician-written rubrics. A length-adjusted variant penalizes verbosity. GPT-5 67.2%
  • FinSearchComp In use Agents search the web to fetch live financial data, look up history, and run multi-step investigations, across 635 expert-written questions. Scored by answer accuracy. Ling-3.1-flash 81.7%
  • KorMedMCQA In use Models answer 7,469 multiple-choice questions from Korean licensing exams for doctors, nurses, pharmacists and dentists, held from 2012 to 2024. o1-preview 92.7%
  • Harvey LAB-AA v1.1 In use Artificial Analysis's run of Harvey's Legal Agent Benchmark on Harvey's newer private set of 120 tasks across 24 practice areas. A deliverable with a material hallucination scores zero, and the headline is the share of judges who find every rubric criterion passed. GPT-6.2 Astra 15%

Data audit

All 9 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 3 pass 2 notes
  • Pass Independent runs Scores from the official leaderboard (33 entries).
  • Pass Models scored 4 models we track, plus 29 other entries on its leaderboard.
  • Pass Newest score The newest is from October 2026, for EmbeddingGemma 2.
  • Note Frontier models None of its models are on the capabilities index.
  • Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on VAGO Solutions: 4 rows for models we track and 29 other entries at the last read.
  • Pass Last read November 20th, 2026.
  • Pass Read worked The last read worked.
Consistency 1 pass 1 note
  • Note Lab claims No model has both a lab number and an independent run.
  • Pass Known issues No published audit found problems.

How we check