WhaVax: leaderboard

WhaVax (2026) asks whether a model can tell vaccine misinformation from other vaccine talk in Portuguese WhatsApp messages from Brazilian public groups, labelled by medical specialists. LLMs classify each message with eight in-context examples; the score is macro-F1 on a 0-100 scale, averaged over repeated five-fold cross-validation; higher is better.

Metric: Macro-F1 (0-100) for misinformation versus non-misinformation over expert-labelled WhatsApp vaccine messages in Portuguese, eight in-context examples, mean over five-fold cross-validation repeated eight times; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 6 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1GPT-5.178.6#240
2GPT-5.278#185
3DeepSeek V3.274.4#419
4Llama 3.1 8B73.4#5196
5Llama 3.2 3B50.3#7877
6Qwen 3 30B A3B38.3#1156

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=whavax · How It Works · Data refreshed daily, snapshot 2026-11-21.