1 million Llama 3.1 405B Instruct tokens cost $1.75 as input or $3.50 as output. Most workloads land in between — a 70/30 input/output split costs about $2.275. Verified Aug 19, 2026
Interactive calculator
1,000,000 in · 0 out
Input price
$1.75 / 1M tokens
Output price
$3.50 / 1M tokens
Uncached pricing. Add caching and monthly projections in the full calculator.
What 1 million tokens buys in practice
$2.10
80% input / 20% output — short replies over conversation history.
$1.925
90% input / 10% output — long retrieved context, grounded answers.
$1.838
95% input / 5% output — full documents in, structured extraction out.
Same 1 million tokens on other models
Related calculations
FAQ
1 million input tokens cost $1.75; 1 million output tokens cost $3.50. A typical 70/30 input/output mix costs about $2.275. Rates: $1.75/M input, $3.50/M output.
A chat-style request of ~4,800 tokens (4,000 in + 800 out) means 1 million tokens is roughly 208 requests. For a RAG workload with ~8,500 tokens per request, about 118 requests.
Llama 3.1 405B Instruct has no published cached-input discount on its official pricing page; all input is billed at $1.75/M.