Skip to content

Llama 3.3 70B Instruct

by MetaRank #180Score 66.8

The Meta Llama 3.3 multilingual large language model (LLM) is a pretrained and instruction tuned generative model in 70B (text in/text out). The Llama 3.3 instruction tuned text only model...

Performance Overview

Score
66.8
Rank
#180
24h Change
0
7d Change
-59
State
stable
Confidence
high

Signal Scores

SignalNormalizedWeightContributionFreshness
Capabilities
capability
50.020%10.02026-08-08T12:22:49.281Z
Benchmarks
benchmark
70.930%21.32026-08-08T12:22:49.281Z
Pricing
pricing_tier
99.715%15.02026-08-08T12:22:49.281Z
Context Window
context_window
81.210%8.12026-08-08T12:22:49.281Z
Recency
recency
21.915%3.32026-08-08T12:22:49.281Z
Output Capacity
output_capacity
70.010%7.02026-08-08T12:22:49.281Z

Top Drivers

positive
Benchmarks
positive
Pricing
$0.32/M output tokens
$0.32
neutral
Capabilities
Supports tools, JSON mode, streaming
3/7
positive
Context Window
131K token context window
131K

Capabilities

CapabilitySupported
VisionNo
ReasoningNo
JSON ModeYes
StreamingYes
Function CallingYes
Web SearchNo

Pricing

Input / 1M tokens
$0.10
Output / 1M tokens
$0.32
Context Window
131K
Max Output
16K

Related Models

Related

Llama 3.3 70B Instruct Tracker - Performance History | LM Market Cap