PaLM-2 (Bison) — benchmark results
Google's mid-size PaLM 2 tier behind Vertex AI's chat-bison API, announced at I/O in May 2023 and since retired. Provider: Google. Released 2023-05-10. Access: API.
Unified ELO 1495 ± 22, rank #820 of 1776 rated models, from 10 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| HELM NaturalQuestions (Open) | 81.26 | F1 (%) | 97.8 |
| HELM WMT 2014 | 24.08 | BLEU-4 (%) | 97.8 |
| HELM v2 Lite - LegalBench | 62.96 | Exact Match (%) | 85 |
| HELM v2 Lite - HumanEval (Code) | 29 | Pass@1 (%) | 76.5 |
| HELM Lite | 56.77 | Mean win rate (self-reported) | 63.6 |
| HELM NaturalQuestions (Closed) | 39 | F1 (%) | 57.8 |
| HELM v2 Lite - MedQA | 67.31 | Exact Match (%) | 57.5 |
| HELM (Stanford) | 52.64 | Mean Win Rate (%) | 56.7 |
| HELM NarrativeQA | 71.8 | F1 (%) | 36.7 |
| LLM-AggreFact | 66.13 | Balanced Accuracy (%) | 23.7 |
Interactive version: theaggregate.ai/model?slug=palm-2-bison · How the rankings work · Data refreshed daily, snapshot 2026-07-22.