Compare · ModelsLive · 2 picked · head to head
Grok-3 mini vs Llama 3.3 70B Instruct
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
Grok-3 mini wins on 5/5 benchmarks
Grok-3 mini wins 5 of 5 shared benchmarks. Leads in knowledge · math · coding.
Category leads
knowledge·Grok-3 minimath·Grok-3 minicoding·Grok-3 mini
Hype vs Reality
Attention vs performance
Grok-3 mini
#93 by perf·no signal
Llama 3.3 70B Instruct
#218 by perf·#18 by attention
Best value
Llama 3.3 70B Instruct
Grok-3 mini
n/a
no price
Llama 3.3 70B Instruct
171.9 pts/$
$0.21/M
Vendor risk
Who is behind the model
xAI
$250.0B·Tier 1
Meta AI
$1.87T·Tier 1
Head to head
5 benchmarks · 2 models
Grok-3 miniLlama 3.3 70B Instruct
Fiction.LiveBench
Grok-3 mini leads by +33.4
Fiction.LiveBench · a continuously updated benchmark using recently published fiction to test reading comprehension and reasoning, preventing data contamination.
Grok-3 mini
66.7
Llama 3.3 70B Instruct
33.3
GPQA diamond
Grok-3 mini leads by +38.4
Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs.
Grok-3 mini
68.3
Llama 3.3 70B Instruct
29.9
MATH level 5
Grok-3 mini leads by +49.3
MATH Level 5 · the hardest tier of the MATH benchmark, featuring competition-level problems from AMC, AIME, and Olympiad-style mathematics.
Grok-3 mini
90.9
Llama 3.3 70B Instruct
41.6
OTIS Mock AIME 2024-2025
Grok-3 mini leads by +72.7
OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills.
Grok-3 mini
77.8
Llama 3.3 70B Instruct
5.0
WeirdML
Grok-3 mini leads by +28.1
WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns.
Grok-3 mini
42.6
Llama 3.3 70B Instruct
14.4
Full benchmark table
| Benchmark | Grok-3 mini | Llama 3.3 70B Instruct |
|---|---|---|
Fiction.LiveBench Fiction.LiveBench · a continuously updated benchmark using recently published fiction to test reading comprehension and reasoning, preventing data contamination. | 66.7 | 33.3 |
GPQA diamond Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs. | 68.3 | 29.9 |
MATH level 5 MATH Level 5 · the hardest tier of the MATH benchmark, featuring competition-level problems from AMC, AIME, and Olympiad-style mathematics. | 90.9 | 41.6 |
OTIS Mock AIME 2024-2025 OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills. | 77.8 | 5.0 |
WeirdML WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns. | 42.6 | 14.4 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
| Model | Input | Output | Context | Projected $/mo |
|---|---|---|---|---|
| — | — | — | — | |
| $0.10 | $0.32 | 131K tokens (~66 books) | $1.55 |