Deepthink-Reasoning-7B: benchmark results

Provider: prithivMLmods. Access: Open.

Aggregate Rating 1461 ± 51, rank #3393 of 10552 rated models, from 8 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
Open LLM Leaderboard - MATH Level 533.46Score84.2
Open LLM Leaderboard - MuSR44.32Score79.9
Open LLM Leaderboard - MMLU-Pro43.49Score79
Open LLM Leaderboard - BBH55.05Score73.6
Open LLM Leaderboard - IFEval48.4Score56.5
Open LLM Leaderboard - GPQA29.95Score55.6
AEPC-QA (RAG)25.14Accuracy (%) with the authors' retrieval-augmented pipeline 4.7
AEPC-QA (Closed-Book)30.33Accuracy (%) answering from the model's own knowledge on AEP4

Interactive version: theaggregate.ai/model?slug=deepthink-reasoning-7b · How It Works · Data refreshed daily, snapshot 2026-11-16.