modelbenchmark.io

Llama-3.3-70B-Instruct

Meta · llama-3-3-70b

Compare

Open Llama instruction model for multilingual chat, reasoning, and coding

Specification

most-agreed values

Context
128K
Max output
4K
Released
2024-12-06
Knowledge cutoff
2023-12
Retires
Open weights
yes
Input
text
Output
text

Price

US dollars per million tokens · most-agreed

Input
free
Output
free

Quality

3 benchmarks · 1 source

Composite
25th
percentile of 334 scored models
Rank
252 / 334
±0.06 sd
Evidence
3 × 1
one source, or fewer than 3 benchmarks
Effort range
one configuration only
reasoning29th1/4 bench
math21st2/7 bench

Every score

one row per source and configuration — nothing averaged away

BenchmarkScoreConfigurationSourceRun
MATH level 541.6 ±1.2defaultEpoch AI2025-01-27
GPQA diamond47.4 ±2.9defaultEpoch AI2025-01-27
OTIS Mock AIME 2024-20255.1 ±2.4defaultEpoch AI2025-02-25

The leading figure for a benchmark is the median across its configurations, so one heroic high-effort run cannot set the number.

Available from 5 hosts

HostIn $/MOut $/MCache rdCache wrContextOutputRetires
vercelfreefree128K4K
venice$0.70$2.80128K4K
snowflake$0.72$0.72128K16K
vercel_ai_gateway · vercel-ai-gateway$0.72$0.72128K8K
cerebras$0.85$1.20128K128K

Other listings of this model

this page is one of them

The corpus files Llama-3.3-70B-Instruct under several keys. This page is the llama-3-3-70b listing. The full record — every host, every price and every benchmark score — is on the main Llama-3.3-70B-Instruct page.

More from Meta

most-hosted first