llama-3-8b

llama-3-8b is a budget-tier language model from Meta with 8B parameters. llama-3-8b is an open-weights model, meaning the weights are publicly available for self-hosting, fine-tuning, and deployment on your own infrastructure — not just through Meta's API. Llama 3 is Meta's open-weights model family, available in 8B and 70B sizes — the most widely deployed open-weights models, with strong community support and broad framework compatibility.

llama-3-8b is priced at $0.04/1M input tokens and $0.04/1M output tokens. Output tokens are 1× more expensive than input tokens — this ratio reflects the sequential nature of autoregressive generation, which requires a separate forward pass per output token versus parallel processing of all input tokens simultaneously. At a typical usage pattern of 1,000 input tokens and 500 output tokens per call, llama-3-8b costs approximately $0.06 per 1,000 API calls. Meta does not publish a prompt caching rate for this model, so repeated long contexts are billed at the standard $0.04/1M input rate.

Its 8K-token context window is standard for this tier — adequate for most chat and completion tasks, but may require chunking for long-document workloads. Median latency (P50 time-to-first-token) is approximately 110ms — fast enough for real-time interactive applications. Throughput is approximately 450 tokens/second — fast enough to stream responses at reading speed with headroom for concurrent requests. Maximum output length is 4K tokens — adequate for most generation tasks including essays, code functions, and detailed responses.

llama-3-8b is best suited to cost, speed. Supported capabilities include streaming token output for real-time UIs. When evaluating this model for a specific workload, the most important factors are: context window (does your input fit?), output price (how long are your generations?), and latency (is this interactive or batch?). The pricing table above shows how llama-3-8b compares to similar models across these dimensions.

llama-3-8b sits in the budget tier — the most cost-efficient models, best for high-volume simple tasks, classification, extraction, and applications where cost per call is the primary constraint. As an open-weights model, llama-3-8b can be self-hosted on GPU compute (see the GPU rental comparison on this site) to eliminate per-token API costs entirely — the break-even point versus API pricing depends on your volume and the GPU hourly rate. When comparing llama-3-8b to alternatives, use the similar models section on this page to find models at comparable price points with different capability tradeoffs. The intelligence score (46/100) reflects aggregate benchmark performance — a useful proxy for general capability, though real-world performance varies significantly by task type. Always test with your specific prompts and use case before committing to a model for production.