Meta · Efficient · Open weights
Llama 3.3 8B
A small open-weight model that runs comfortably on modest hardware. The default starting point for local experimentation.
Strong at
- -Small footprint
- -Open weights
- -Easy to self-host
Typical use
- -Local prototyping
- -Edge deployment
Same tier
What else to look at
| Model | Developer | Context | In / 1M | Out / 1M |
|---|---|---|---|---|
| Claude Haiku 4.5 | Anthropic | 200K | $1 | $5 |
| GPT-5.6 Luna | OpenAI | 1M | $0.20 | $1.20 |
| DeepSeek V4 Pro | DeepSeek | 1M | $0.43 | $0.87 |
| DeepSeek V4 Flash | DeepSeek | 1M | $0.14 | $0.28 |
| Qwen 3.7 Flash | Alibaba | - | $0.03 | $0.13 |
Verified 2026-08-07.