flan-t5-base: benchmark results
Provider: Google. Released 2022-10-21. Access: API.
Unified ELO 1335 ± 1, rank #1294 of 1392 rated models, from 13 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| ConvRe - Text2Re Hard | 50.2 | Accuracy (%) | 88.9 |
| ConvRe - Re2Text Easy | 84.6 | Accuracy (%) | 55.6 |
| InstructEval - Problem Solving | 31.6 | Average (%, MMLU/BBH/DROP/CRASS/HumanEval) | 53.1 |
| InstructEval - Alignment (HHH) | 59 | Average (%, harmless/helpful/honest) | 44 |
| ConvRe | 50.8 | Average Score (%) | 22.2 |
| ConvRe - Re2Text Hard | 17 | Accuracy (%) | 22.2 |
| Open LLM Leaderboard - BBH | 11.34 | Score | 20.9 |
| Open LLM Leaderboard - MuSR | 3.22 | Score | 15.4 |
| Open LLM Leaderboard - MMLU-Pro | 3.97 | Score | 11.4 |
| ConvRe - Text2Re Easy | 51.2 | Accuracy (%) | 11.1 |
| Open LLM Leaderboard - IFEval | 18.91 | Score | 10.4 |
| Open LLM Leaderboard - MATH Level 5 | 1.06 | Score | 7.6 |
Interactive version: theaggregate.ai/model?slug=flan-t5-base · How It Works · Data refreshed daily, snapshot 2026-09-05.