Llama 4
Meta · v2026-05 · Open-weight
Capabilities
- Context window
- 128K tokens
- Modalities
- Text, Code
- Tool use
- Yes
- Structured output
- Yes
- API available
- Yes
- Local available
- Yes
Pricing
- Input
- Self-hosted or $0.20–$1.00 / 1M tokens
- Output
- Self-hosted or $0.60–$3.00 / 1M tokens
- Cached input
- Varies by provider
Providers
- Self-hosted
- Together AI
- Fireworks AI
- Groq
Best use cases
- Local deployment
- Fine-tuning
- Cost-sensitive production
- Privacy-critical applications
Known weaknesses
- Weaker than frontier closed models on complex reasoning
- Licence restrictions on commercial scale
Last verified: 2026-07-11 · See corrections →