Long ContextActive
Trust Score:
93

MLCR-AA (Medical Long Context Reasoning)

Medical long-context reasoning benchmark built on healthcare-claims style documents.

Launched: —Refresh: static
Status Assessment (active):

Actively evaluated by Artificial Analysis on current frontier models; no saturation evidence recorded.

Performance Timeline

Longitudinal progression of model scores against human baselines.

Performance & Historical Trajectory

Empirical score progression across model release dates and evaluation rounds.

Independent Vendor-Reported
ModelScoreDateSource TypeProvenance
Qwen3.7 Plus8.9%2026-11-15independentSource ↗
Kimi K3.1 Pro38.3%2026-11-15independentSource ↗
Ministral 3 3B0%2026-11-15independentSource ↗
Beam 1.017.8%2026-11-15independentSource ↗
Ling-3.0-flash-Fin3.9%2026-11-15independentSource ↗
Ministral 3 14B2.2%2026-11-15independentSource ↗
Inkling Small7.8%2026-11-15independentSource ↗
Gemini 3.8 Flash10.6%2026-11-15independentSource ↗
Muse Ember18.3%2026-11-15independentSource ↗
Ling 3.0 Tiny10%2026-11-15independentSource ↗
Gemini 4 Hydrogen21.1%2026-11-15independentSource ↗
Kimi K2.7 Code12.2%2026-11-15independentSource ↗
Qwen3.8 2.4T A95B30.6%2026-11-15independentSource ↗
Ministral 3 8B0%2026-11-15independentSource ↗
Qwen3.8-Flash-Next20.6%2026-11-15independentSource ↗
Ling 3.1 Flash12.2%2026-11-15independentSource ↗
Solar Mini 46.7%2026-11-15independentSource ↗
Qwen3.8 Max (0902)20%2026-11-15independentSource ↗
K2 Horizon 375B A23B12.2%2026-11-15independentSource ↗
Gemini 3.1 Pro11.1%2026-11-15independentSource ↗
GPT-5.5 Instant31.7%2026-11-15independentSource ↗
gpt-oss-120b1.7%2026-11-15independentSource ↗
GLM 5.317.2%2026-11-15independentSource ↗
Hy432.8%2026-11-15independentSource ↗
MiniMax M3.1 Flash16.7%2026-11-15independentSource ↗
Mistral Medium 3.510%2026-11-15independentSource ↗
Nemotron 3.5 Super VL13.3%2026-11-15independentSource ↗
Inkling 2 Small20.6%2026-11-15independentSource ↗
Muse Wonder Full22.2%2026-11-15independentSource ↗
Ring-2.6-1T3.9%2026-11-15independentSource ↗
Quasar 438B9.4%2026-11-15independentSource ↗
Grok 4.720%2026-11-15independentSource ↗
GPT-6.1 Sol38.3%2026-11-15independentSource ↗
Gemini 3.5 Flash-Lite10.6%2026-11-15independentSource ↗
Step 518.3%2026-11-15independentSource ↗
Claude Fable 5.243.9%2026-11-15independentSource ↗
GPT-6.1 Terra30.6%2026-11-15independentSource ↗
Gemini 4 Carbon19.4%2026-11-15independentSource ↗
Inkling6.1%2026-11-15independentSource ↗
Qwen3.5 397B A17B8.9%2026-11-15independentSource ↗
MiMo-V2.6-Flash11.7%2026-11-15independentSource ↗
Mistral Small 41.7%2026-11-15independentSource ↗
GLM 5.3 Flash11.7%2026-11-15independentSource ↗
GPT-6 Luna26.1%2026-11-15independentSource ↗
North Mini Code4.4%2026-11-15independentSource ↗
GPT-6 Astra41.7%2026-11-15independentSource ↗
Claude Haiku 5.535%2026-11-15independentSource ↗
Claude Sonnet 5.541.7%2026-11-15independentSource ↗
MiMo-V2.6-Pro12.8%2026-11-15independentSource ↗
MiniMax M34.4%2026-11-15independentSource ↗
Mistral Large 417.8%2026-11-15independentSource ↗
Nemotron 3.5 Lightning2.2%2026-11-15independentSource ↗
Command A+1.7%2026-11-15independentSource ↗
Claude Opus 5.527.8%2026-11-15independentSource ↗
Qwen3.8 27B22.8%2026-11-15independentSource ↗
Muse Spark Open7.8%2026-11-15independentSource ↗
Muse Wonder22.2%2026-11-15independentSource ↗
Ling-3.0-flash-VL5%2026-11-15independentSource ↗
Gemini 4 Argon13.9%2026-11-15independentSource ↗
Ling 3.1 Flash 101915%2026-11-15independentSource ↗
DeepSeek V4.1 Flash 101726.7%2026-11-15independentSource ↗
Nemotron 3 Ultra5.6%2026-11-15independentSource ↗

Human Baseline & Difficulty Horizon

Calibrated human reference points, specialist benchmarks, and ceiling thresholds.
Measured Human Scoren/aNot established
Baseline Protocol & Interpretation

No peer-reviewed human baseline has been published for this benchmark.

Metric & Scoring Methodology

Verification protocols, aggregation formulas, and specialized metric variants.
Primary Metric:accuracy (%)
Scoring Engine:llm-judge

Dataset & Compute Cost

Evaluation volume, public availability, API pricing, and local hardware requirements.
Total Dataset Size—Annotated evaluation items
Public Test SetPublicOpenly mirrored on repositories
Access GatingOpen AccessUnrestricted download
Evaluation LicenseOpen AccessDataset usage and redistribution terms
Frontier API Compute Cost:

$5 – $20 USD for full benchmark evaluation run on frontier APIs.

Recommended Local GPU Setup:

1x NVIDIA RTX 4090 (24GB) or A100 (40GB/80GB) via vLLM / SGLang

How to Run & Reproduce

Standardized evaluation protocols, CLI commands, and reproducible runner templates.
Prompt Regimezero-shot
Reasoning Modedirect
Sampling Temp0
Pass@k Budgetk = 1
Tools & SandboxPure Text
Scoring Verifierllm-judge
Option AEleutherAI LM-Evaluation-Harness (Open-Weight Models)
lm_eval --model hf --model_args pretrained=<model_path> --tasks mlcr-aa --batch_size auto
Option BOpenCompass Evaluation Framework
opencompass --datasets mlcr-aa --models <model_config>
Python APIDeterministic Inference Loop Snippet
# Standard API Evaluation Loop
from openai import OpenAI

client = OpenAI()
response = client.chat.completions.create(
    model="gpt-4o",
    messages=[{"role": "user", "content": prompt}],
    temperature=0.0,
)
Standardized Reporting Requirement:

When publishing results for MLCR-AA (Medical Long Context Reasoning), always report the exact prompt template, few-shot exemplar ordering, sampling temperature (temperature=0), maximum reasoning budget tokens, and the precise timestamped model snapshot ID.

Contamination & Memorization Analysis

Audit of pretraining exposure risks, memorization vectors, and refresh policies.
Overall Contamination Risk:MEDIUM
Refresh Cadence:

Static fixed snapshot

Test Set Exposure:

Public on web / HuggingFace