MedLFQA
Models write long answers to biomedical questions. The benchmark checks whether answers include required facts and avoid claims identified as incorrect.
- Made by
- Minbyul Jeong and colleagues
- Released
- May 2024
- Kind
- Medicine
- Level
- Expert
- Scored
- Checked automatically
- Answers
- Open-ended
- Tasks
- All public
- Measures
- Capability
- Subject
- Clinical medicine
- Language
- English
- Interaction
- Single turn
- Refresh
- One fixed set
- Run by
- Anyone
- In views
- Biology and medicine
- Strongest model tested
- Claude 3 Sonnet #187 on the capability index
- Last model added
- February 2024
- Scores refreshed
- November 20th, 2026
GPT-4o (snapshot unspecified) tops the leaderboard with 78.3 Claude 3 Sonnet leads with 77.1 among models released by February 2024
Best score per entry from the benchmark's paper, read on November 20th, 2026. A paper’s table doesn’t add newer models. The headline averages factuality, the paper’s primary focus, equally across its five datasets. 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 Claude 3 Sonnet, at 77.1. This view shows only models we track, each at its best.
The newest model with a score here came out in February 2024, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- GPT-6.2 Astra
- Claude Fable 5.2
-
Gemini 4 Carbon
- GPT-6 Astra
- and 16 more
- 1 GPT-4o (snapshot unspecified) Zero-shot 78.3
- 2 Claude 3 Sonnet Zero-shot 77.1
- 3 GPT-3.5-Turbo (snapshot unspecified) Zero-shot 53.1
- 4
Mistral 7B v0.1 Zero-shot, Olaph step 1 48.3
- 5 BioMistral Zero-shot, Olaph step 1 40.8
- 6 Self-BioRAG Zero-shot, Olaph step 1 37.4
- 7
Llama 2-7B Zero-shot, Olaph step 1 25.2
- 8 Meditron Zero-shot, Olaph step 1 24.1
Source: The MedLFQA paper
Source: The MedLFQA paper
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | GPT-4o (snapshot unspecified) | 78.3 | Zero-shot | Official leaderboard | |
| 2 | Claude 3 Sonnet | 77.1 | Zero-shot | Feb 2024 | Official leaderboard |
| 3 | GPT-3.5-Turbo (snapshot unspecified) | 53.1 | Zero-shot | Official leaderboard | |
| 4 | | 48.3 | Zero-shot, Olaph step 1 | Sep 2023 | Official leaderboard |
| 5 | BioMistral | 40.8 | Zero-shot, Olaph step 1 | Official leaderboard | |
| 6 | Self-BioRAG | 37.4 | Zero-shot, Olaph step 1 | Official leaderboard | |
| 7 | | 25.2 | Zero-shot, Olaph step 1 | Jul 2023 | Official leaderboard |
| 8 | Meditron | 24.1 | Zero-shot, Olaph step 1 | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Claude 3 Sonnet | 77.1 | Zero-shot | Feb 2024 | Official leaderboard |
| 2 | | 48.3 | Zero-shot, Olaph step 1 | Sep 2023 | Official leaderboard |
| 3 | | 25.2 | Zero-shot, Olaph step 1 | Jul 2023 | Official leaderboard |
Rankings by part
We read 15 parts from the benchmark’s paper in 5 groups: liveqa, medicationqa, healthsearchqa, k-qa golden and k-qa silver. Each ranks models on one part of the test. Claude 3 Sonnet leads every one of them. Arrows show how far a model moved from its overall rank among the models we track.
Words Composition
3 models we track have a score here.
- 1 Claude 3 Sonnet 44.3
- 2
Mistral 7B v0.1 8.50
- 3
Llama 2-7B 7.73
Semantic Similarity
3 models we track have a score here.
- 1 Claude 3 Sonnet 116.5
- 2
Llama 2-7B ↑1 67.0
- 3
Mistral 7B v0.1 ↓1 64.3
Factuality
3 models we track have a score here.
- 1 Claude 3 Sonnet 71.2
- 2
Llama 2-7B ↑1 42.3
- 3
Mistral 7B v0.1 ↓1 33.6
Words Composition
3 models we track have a score here.
- 1 Claude 3 Sonnet 48.9
- 2
Mistral 7B v0.1 6.30
- 3
Llama 2-7B 5.30
Semantic Similarity
3 models we track have a score here.
- 1 Claude 3 Sonnet 122.3
- 2
Llama 2-7B ↑1 67.0
- 3
Mistral 7B v0.1 ↓1 65.2
Factuality
3 models we track have a score here.
- 1 Claude 3 Sonnet 79.9
- 2
Mistral 7B v0.1 35.0
- 3
Llama 2-7B 14.1
Words Composition
3 models we track have a score here.
- 1 Claude 3 Sonnet 41.2
- 2
Mistral 7B v0.1 15.8
- 3
Llama 2-7B 12.0
Semantic Similarity
3 models we track have a score here.
- 1 Claude 3 Sonnet 115.1
- 2
Mistral 7B v0.1 65.2
- 3
Llama 2-7B 63.6
Factuality
3 models we track have a score here.
- 1 Claude 3 Sonnet 71.3
- 2
Mistral 7B v0.1 67.6
- 3
Llama 2-7B 36.4
Words Composition
3 models we track have a score here.
- 1 Claude 3 Sonnet 48.5
- 2
Mistral 7B v0.1 10.9
- 3
Llama 2-7B 8.40
Semantic Similarity
3 models we track have a score here.
- 1 Claude 3 Sonnet 119.3
- 2
Mistral 7B v0.1 66.3
- 3
Llama 2-7B 65.4
Factuality
3 models we track have a score here.
- 1 Claude 3 Sonnet 82.8
- 2
Mistral 7B v0.1 50.4
- 3
Llama 2-7B 38.2
Words Composition
3 models we track have a score here.
- 1 Claude 3 Sonnet 51.3
- 2
Mistral 7B v0.1 18.2
- 3
Llama 2-7B 7.50
Semantic Similarity
3 models we track have a score here.
- 1 Claude 3 Sonnet 117.7
- 2
Mistral 7B v0.1 66.2
- 3
Llama 2-7B 63.9
Factuality
3 models we track have a score here.
- 1 Claude 3 Sonnet 80.1
- 2
Mistral 7B v0.1 54.9
- 3
Llama 2-7B -5.20
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: The MedLFQA paper
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official MedLFQA 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
- 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
8 of 9 checks pass, 1 warning. Look first at the Newest score check. The newest came 2 years ago, in February 2024, for Claude 3 Sonnet. Every benchmark
Catalog record 2 pass
- Pass Task access All tasks are public.
- Pass Confidence Checked against a primary source.
Scores 1 warning 2 pass 2 notes
- Pass Independent runs Scores from the benchmark’s paper (8 entries).
- Pass Models scored 3 models we track, plus 5 other entries on its leaderboard.
- Warning Newest score The newest came 2 years ago, in February 2024, for Claude 3 Sonnet.
- Note Frontier models None of the top 10 has a score. The strongest tested is Claude 3 Sonnet (#187).
- Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
- Pass Importer We read the board on arxiv.org: 5 rows for models we track and 5 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.