open source-host

Meta (Llama)

Llama 4 open weights; served across many partner clouds.

live

Uptime (30d)

99.9%

Median tps

165

Median TTFT

150 ms

Price index

0.5×

vs. market median

Overview

HQ: Menlo Park, US
Founded: 2004
Models: 7
Modalities: text, vision
Regions: us-east, us-west, eu-west

Strengths

open-weightsthroughput

Route via llmcloud

curl https://api.llmcloud.ai/v1/chat/completions \
  -H "Authorization: Bearer $LLMCLOUD_KEY" \
  -H "X-Provider: meta" \
  -d '{
    "model": "auto:quality",
    "messages": [{"role":"user","content":"hi"}]
  }'

Models served

llama-4-405b

q 87.6165 tps$2.4/1M

llama-4-maverick

q 86.9178 tps$1.8/1M

llama-3.1-405b-instruct

q 85.2130 tps$2.7/1M

llama-4-scout

q 82.1240 tps$0.6/1M

llama-4-70b

q 81.1220 tps$0.9/1M

llama-3.3-70b-instruct

q 80.4225 tps$0.7/1M

llama-3.2-90b-vision

q 79.4150 tps$0.9/1M

llama-3.1-70b-instruct

q 78.5240 tps$0.6/1M

llama-3.2-11b-vision

q 72.1260 tps$0.2/1M

llama-3.1-8b-instruct

q 68.2380 tps$0.05/1M

llama-guard-3-8b

q 65.0340 tps$0.05/1M