Llama 3.1 405B: Pricing, Context, and Benchmarks

Llama 3.1 405B is Meta's Llama 3 model released on 2024-07-23. It runs on a 128K-token context window, priced at $3.50/M input and $3.50/M output. Largest open-weight model from Meta. Strong for self-hosting but Llama 3.3 70B is the more practical default.

Input
$3.50/M
Output
$3.50/M
Context
128K
Max output
4K

Specifications

ProviderMeta
FamilyLlama 3
Released2024-07-23
Statusactive
Context window128K tokens
Max output4K tokens
Modalitiestext
Capabilities function calling json mode streaming
Tokenizerllama-3
API endpointhttps://api.together.xyz/v1/chat/completions
DocumentationOfficial docs ↗
Last updated2026-05-29

Cost calculator

For 1,000 tokens of input and 1,000 tokens of output per call, this model costs $0.0070 per call. For typical usage:

WorkloadTokens / call (in/out)$/call$/1K calls$/1M calls
Short Q&A 200 / 100 $0.001050 $1.05 $1,050
Standard chat 1K / 500 $0.005250 $5.25 $5,250
RAG with retrieval 4K / 500 $0.015750 $15.75 $15,750
Long doc summary 20K / 1K $0.073500 $73.50 $73,500
Long context (100K input) 100K / 2K $0.355250 $355.25 $355,250

For more precise estimates including caching discounts, use the LLM API Cost Calculator.

Other Meta models

Common comparisons

API setup

curl https://api.together.xyz/v1/chat/completions \
  -H "Authorization: Bearer $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "llama-3-1-405b", "messages": [{"role": "user", "content": "Hello"}]}'

See the official documentation for the full request/response schema. For pre-flight token counting before paying, use the appropriate token counter.