WeavePrism

Llama 3.3 70B vs Mistral Small 3.2

Mistral Small 3.2 costs 74% less at typical usage ($0.15 vs $0.58 per 1M tokens). Llama 3.3 70B scores higher on 7 of 11 tasks.

SpecLlama 3.3 70BMistral Small 3.2
Input /1M$0.54$0.10
Output /1M$0.68$0.30
Typical /1M$0.58$0.15
Context window128k128k
SpeedBlazingBlazing
PrivacySelf-hostSelf-host
Open weightsYesYes
Llama 3.3 70B
Steady since we started tracking (Jul 13, 2026)
Steady since we started tracking (Jul 13, 2026)
Mistral Small 3.2
Steady since we started tracking (Jul 13, 2026)
Steady since we started tracking (Jul 13, 2026)

Quality by task

Coding76/74
Agents72/68
Reasoning77/74
Writing80/78
Chat84/82
Classify82/82
Extract80/80
Summarize82/80
Translate82/84
Vision25/76
General78/76

Price your own prompt against both

The free estimator ranks Llama 3.3 70B, Mistral Small 3.2 and every other catalog model by cost and quality for your exact workload.

Open the estimator

Get notified when either model's price changes

We watch the catalog for list-price changes. Opt in with just an email, no account, and unsubscribe in one click anytime.

Also compare Llama 3.3 70B

Also compare Mistral Small 3.2

Prices are list $/1M tokens, hand-maintained from public provider pricing. Task scores are benchmark-informed 0-100 estimates, not gospel — validate on your own evals before moving production traffic. Not affiliated with any model provider.