BasedAGIBasedAGI
Menu
Rankings live

biomed_science

Cross-paper contradiction analysis

Identify contradictions and uncertainty across papers with citations.

#1 Recommendation

gemini-3-pro-preview

Strong on FACTS Benchmark Suite facts_grounding_score_pct (88%) and Vals GPQA overall_accuracy_pct (94%)

external/google/gemini-3-pro-preview

35.6%

Score

46.1%

Confidence

Limited benchmark evidence for this use case.

52 ranked models with average evidence of 14.8 points. Rankings may shift as more benchmark data is ingested.

Ranked Models

30

Evidence Quality

85%

Scoring

Benchmark-backed

Top Signal

FACTS Benchmark Suite: facts_grounding_score_pct

All Ranked Models

Max params:
Min confidence:
30 of 30
RankModelScore
#1gemini-3-pro-preview

Strong on FACTS Benchmark Suite facts_grounding_score_pct (88%) and Vals GPQA overall_accuracy_pct (94%)

35.6%
#2gemini-2.5-pro

Strong on FACTS Benchmark Suite facts_grounding_score_pct (100%) and Vectara HHEM Leaderboard overall_hallucination_error_pct (76%)

30.0%
#3google/gemini-3.1-pro-preview

Strong on Vals GPQA overall_accuracy_pct (100%) and Vals MedCode overall_accuracy_pct (100%)

28.6%
#4gpt-4.1-20250414
27.0%
#5gpt-5-mini-2025-08-07
26.3%
#6anthropic/claude-sonnet-4.6
25.5%
#7claude-opus-4-5-20251101
25.2%
#8gpt-5-2025-08-07
25.2%
#9openai/gpt-5.4-2026-03-05
25.1%
#10Grok-4-0709
23.6%
#11gemini-3-flash-preview
23.2%
#12claude-sonnet-4-20250514
22.0%
#13google/gemini-3.1-flash-lite-preview
21.9%
#14gpt-5.1-2025-11-13
21.4%
#15gemini-2.5-flash
20.5%
#16xai-org/grok-4-fast-reasoning
19.6%
#17gpt-5.2-2025-12-11
18.7%
#18anthropic/claude-opus-4-6-thinking
18.6%
#19xai-org/grok-4-1-fast-reasoning
18.2%
#20kimi/kimi-k2.5-thinking
17.8%
#21anthropic/claude-opus-4-5-20251101-thinking
17.3%
#22x-ai/grok-3
16.8%
#23anthropic/claude-opus-4-1-20250805
16.3%
#24anthropic/claude-sonnet-4-5-20250929-thinking
16.0%
#25o3-20250416
15.3%
#26mistralai/mistral-large-2512
15.3%
#28zai/glm-5-thinking
14.4%
#29xai-org/grok-4-1-fast-non-reasoning
14.3%
#30alibaba/qwen3.5-flash
13.6%
#31anthropic/claude-haiku-4-5-20251001-thinking
13.3%

Compare Models

Model A leads by +5.7%

Shareable Link →

Model A

gemini-3-pro-preview

external/google/gemini-3-pro-preview

35.6%

Rank #1

Confidence 46.1%25 evidence pts

FACTS Benchmark Suite: facts_grounding_score_pct

Value 88.3% · Conf 100.0% · Weight 3.6%

facts_benchmark_suite.facts_grounding_score_pct (Mar 12, 2026)

Vals GPQA: overall_accuracy_pct

Value 94.1% · Conf 100.0% · Weight 2.5%

vals_gpqa.overall_accuracy_pct (Mar 12, 2026)

FACTS Benchmark Suite: facts_search_score_pct

Value 100.0% · Conf 100.0% · Weight 2.3%

facts_benchmark_suite.facts_search_score_pct (Mar 12, 2026)

FACTS Benchmark Suite: average_score_pct

Value 100.0% · Conf 100.0% · Weight 2.1%

facts_benchmark_suite.average_score_pct (Mar 12, 2026)

Model B

gemini-2.5-pro

external/google/gemini-2-5-pro

30.0%

Rank #2

Confidence 44.2%26 evidence pts

FACTS Benchmark Suite: facts_grounding_score_pct

Value 100.0% · Conf 100.0% · Weight 4.1%

facts_benchmark_suite.facts_grounding_score_pct (Mar 12, 2026)

Vectara HHEM Leaderboard: overall_hallucination_error_pct

Value 76.0% · Conf 100.0% · Weight 2.4%

vectara_hhem_leaderboard.overall_hallucination_error_pct (Mar 12, 2026)

Vals CorpFin v2: overall_accuracy_pct

Value 78.4% · Conf 100.0% · Weight 1.9%

vals_corp_fin_v2.overall_accuracy_pct (Mar 12, 2026)

Vals MedCode: overall_accuracy_pct

Value 73.5% · Conf 100.0% · Weight 1.8%

vals_medcode.overall_accuracy_pct (Mar 12, 2026)

Ranking Diagnostics & Missing Models

Source Lift

Ranked

52

Sources

8

Quality

Insufficient

Vals CorpFin v2

vals_corp_fin_v2

42 rows

1.5% avg lift

Vals GPQA

vals_gpqa

42 rows

1.8% avg lift

Vals Legal Bench

vals_legal_bench

30 rows

0.4% avg lift

Vals MedQA

vals_medqa

30 rows

0.4% avg lift

Missing Strong Models

gpt-4o

external/openai/gpt-4o

Rank #22

15.2%

Thin evidence after weighting

qwen-2.5-72b-instruct

external/qwen/qwen-2-5-72b-instruct

Rank #27

14.2%

Thin evidence after weighting

deepseek/deepseek-r1

external/deepseek/deepseek-r1

Rank #54

10.5%

Thin evidence after weighting
Taxonomy Details

Core Tasks

task.contradiction_detectiontask.claim_check_with_evidence

Required Modes

mode.citationsmode.long_context

Domains

domain.biomed_literature

Related Use Cases