Llama 3.3 70B: how it answers moral dilemmas
Across 117 dilemmas, Llama 3.3 70B sided with the majority of models 79% of the time, and answered “yes” to 45% of the 67 yes/no questions.
- Maker
- Meta
- Released
- December 2024
Llama 3.3 70B Instruct is Meta's open-weights, text-only instruction-tuned model released December 2024, delivering performance near the larger Llama 3.1 405B at lower cost. The weights are publicly available for self-hosting and fine-tuning.
Agreement with other models, by category
| Category | Dilemmas | Sided with majority |
|---|---|---|
| Day-to-day | 20 | 85% |
| Political | 20 | 75% |
| Fictional | 20 | 80% |
| Historical | 7 | 100% |
| Classic | 20 | 95% |
| Personality | 30 | 60% |
Where Llama 3.3 70B breaks from the pack
- A family entered the country illegally 10 years ago, works, pays taxes, and has children… — Give legal status without punishment “I think they've earned a right to stay”
- A country has frequent gun violence. What should it do about private firearms? — Ban most private firearms “I think public safety should be the priority”
- Which film would you rather rewatch tonight? — Pulp Fiction
- Which music mood appeals to you most? — Warm jazz/blues
- Which vehicle would you choose for a long journey? — Camper van
- Which holiday would you choose? — Tropical island with no plans
Which model thinks like you?
Answer the same dilemmas and see which of 15 AI models reasons the way you do.
Take the quizCompare Llama 3.3 70B
- Claude 4.8 Opus vs Llama 3.3 70B
- DeepSeek v4 Pro vs Llama 3.3 70B
- Gemini 3.1 Pro Preview vs Llama 3.3 70B
- GPT 5.5 vs Llama 3.3 70B
- Grok 4.3 vs Llama 3.3 70B
Other models