Llama 3.2 3B Instruct
Meta · Released Sep 2024
A 3-billion-parameter instruction-tuned language model from Meta with an 80,000-token context window, released in 2024.
- Strengths
- Offers a balance between model capability and computational efficiency, making it suitable for resource-constrained deployments where larger models are impractical.
- Best for
- Lightweight applications requiring fast inference and lower computational overhead, such as edge devices, real-time systems, or scenarios where inference latency is critical.
- Limitations
- Smaller parameter count means reduced reasoning depth and knowledge capacity compared to Llama 3.2 11B Vision Instruct or larger Llama models in the lineup; lacks multimodal capabilities and vision understanding.
Input / 1M
$0.05
Output / 1M
$0.33
Cached input / 1M
—
Context window
131K
Price history
Input (solid)Output (dashed)
Price change
- 30d
- in decreased 1.8% out decreased 1.5%
- 90d
- in decreased 1.8% out decreased 1.5%
- 1y
- in decreased 1.8% out decreased 1.5%
- Since launch
- in decreased 1.8% out decreased 1.5%
Snapshots
| Effective | Input | Output | Cached in | Note | Source |
|---|---|---|---|---|---|
| 23 Jul 2026 | $0.05 | $0.33 | — | Imported from OpenRouter | openrouter.ai |
| 16 Jul 2026 | $0.0509 | $0.335 | — | Imported from OpenRouter | openrouter.ai |
| 15 Jul 2026 | $0.05 | $0.33 | — | Imported from OpenRouter | openrouter.ai |
| 7 Jul 2026 | $0.05 | $0.33 | — | Imported from OpenRouter | openrouter.ai |
| 11 Jun 2026 | $0.0509 | $0.335 | — | Imported from OpenRouter | openrouter.ai |