PlanPulse
English

Llama 3.3 70B API Pricing

Short answer

Llama 3.3 70B costs $0.90 per 1M input tokens and $0.90 per 1M output tokens (cached input $0.45/1M). Context window: 131K. Last verified 2026-06-17 from fireworks.ai. Confidence: Auto.

Llama 3.3 70B is Meta's popular open-weight workhorse, hosted across many inference providers with a wide price spread.

Input
$0.90 /1M
Output
$0.90 /1M
Cached input
$0.45 /1M
Input pricing$0.90
Cached input$0.45
Output pricing$0.90

Limits & capabilities

Context window131K
Max output—
Modalitytext input/output
ChannelFireworks
ProviderMeta

Best for

Open weightGeneralCheap
Open-weight; price ranges ~3-10x across hosts. Groq is fastest; OpenRouter routes to the cheapest.
Open provider → Pricing source: fireworks.ai →
Last checked Auto

More from Meta

Llama 4 Scout

$0.10 /1M in
$0.30 /1M out

Open weight, Ultra-long context, Multimodal

Llama 4 Maverick

$0.15 /1M in
$0.60 /1M out

Open weight, Reasoning, Multimodal

Related comparisons

OpenAI vs Anthropic API

Compare pricing, channels, and trade-offs →