Meta AI

Llama 3.3 70B Instruct

Type
language model
Context
131K tokens
Max output
16K
Released
6 December 2024
Knowledge cutoff
October 2024
API string
llama-3.3-70b-instruct
Weights
open

Prices by provider

Provider Input, $ per 1M Output, $ per 1M In our data since
NanoGPT $0.05 $0.23 31 Jul 2026
Cortecs $0.089 $0.275 31 Jul 2026
meganova $0.1 $0.3 31 Jul 2026
Kilo Code $0.1 $0.32 31 Jul 2026
Hyperbolic $0.12 $0.3 31 Jul 2026
Helicone $0.13 $0.39 31 Jul 2026
LLM Gateway $0.13 $0.4 31 Jul 2026
Nebius Token Factory $0.13 $0.4 31 Jul 2026
OpenRouter $0.13 $0.4 31 Jul 2026
io.net $0.13 $0.38 31 Jul 2026
novita-ai $0.135 $0.4 31 Jul 2026
Novita AI $0.135 $0.4 31 Jul 2026
crusoe $0.2 $0.2 31 Jul 2026
Nscale $0.2 $0.2 31 Jul 2026
Merge Gateway $0.22 $0.5 31 Jul 2026
DeepInfra $0.23 $0.4 31 Jul 2026
neon $0.5 $1.5 1 Aug 2026
Databricks $0.5 $1.5 1 Aug 2026
hyper $0.513 $1.045 31 Jul 2026
Abacus.AI $0.59 $0.79 1 Aug 2026
Hugging Face $0.59 $0.79 31 Jul 2026
regolo-ai $0.6 $2.7 31 Jul 2026
SambaNova $0.6 $1.2 1 Aug 2026
OVHCloud AI Endpoints ovhcloud/Meta-Llama-3_3-70B-Instruct $0.67 $0.67 1 Aug 2026
Azure Cognitive Services $0.71 $0.71 31 Jul 2026
IBM watsonx.ai $0.71 $0.71 31 Jul 2026
Microsoft Foundry $0.71 $0.71 31 Jul 2026
Weights & Biases $0.71 $0.71 31 Jul 2026
Amazon Bedrock $0.72 $0.72 2 Aug 2026
Google Vertex AI $0.72 $0.72 31 Jul 2026
OCI Generative AI $0.72 $0.72 31 Jul 2026
OVHCloud AI Endpoints meta-llama-3_3-70b-instruct $0.74 $0.74 1 Aug 2026
Pioneer $0.9 $0.9 31 Jul 2026
Scaleway $0.9 $0.9 31 Jul 2026
berget $0.99 $0.99 31 Jul 2026
evroc $1.15 $1.15 31 Jul 2026
greenpt $1.254 $1.254 31 Jul 2026
cloudferro-sherlock $2.92 $2.92 31 Jul 2026

Only the base tier and only per-token prices are shown. Batch, discounted and cached rates, as well as prices per image or per second of video, do not go into this table: they cannot stand in the same column as a price per million tokens.

A single provider appears several times if it sells this model under different identifiers and at different prices — for example, at different weight precision. The identifier is shown next to the name. Identical offers that arrived from two sources under different spellings are merged into one row.

The date in the last column is the day this price first entered our collection. The price may well be older: before that day we simply were not recording it. It has not changed since — otherwise a new row with a new date would stand in its place.

Measurement results

Task set Result Run conditions Measured by
Arena Score in English 1,304.16 1,300–1,308
Arena Score, multi-turn dialogue 1,280.78 1,274–1,287
Arena Score in French 1,278.96 1,253–1,305
Arena Score, overall 1,275.06 1,272–1,279
Arena Score, programming 1,268.88 1,262–1,275
Arena Score, mathematics 1,267.37 1,260–1,275
Arena Score in Spanish 1,266.47 1,242–1,291
Arena Score, hard prompts 1,258.2 1,253–1,263
Arena Score, long queries 1,256.62 1,250–1,263
Arena Score, creative writing 1,251.98 1,245–1,259
Arena Score in Russian 1,250.26 1,242–1,258
Arena Score, instruction following 1,242.89 1,238–1,248
Arena Score, expert questions 1,224.53 1,213–1,236
MATH, difficulty level five 41.6 % · with a tuned harness Epoch evaluations
Fiction.LiveBench — holding a long context 33.3 % Fiction.live leaderboard
GPQA Diamond — graduate-level questions 29.92 % · with a tuned harness Epoch evaluations
BALROG — game environments 23 % Balrog Leaderboard
WeirdML — unusual machine learning tasks 14.44 % WeirdML Leaderboard
Mock AIME 2024–2025 — olympiad problems 5.04 % · with a tuned harness Epoch evaluations
SimpleBench — trick questions 3.88 % SimpleBench Leaderboard
CritPt — physics problems 0 %