Compare · ModelsLive · 2 picked · head to head
GPT-4o-mini vs Llama 3.2 90B
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
GPT-4o-mini wins on 4/6 benchmarks
GPT-4o-mini wins 4 of 6 shared benchmarks. Leads in math.
Category leads
knowledge·Llama 3.2 90Bmath·GPT-4o-mini
Hype vs Reality
Attention vs performance
GPT-4o-mini
#146 by perf·no signal
Llama 3.2 90B
#164 by perf·no signal
Vendor risk
Who is behind the model
OpenAI
$840.0B·Tier 1
Meta AI
$1.50T·Tier 1
Head to head
6 benchmarks · 2 models
GPT-4o-miniLlama 3.2 90B
Balrog
Llama 3.2 90B leads by +9.9
Balrog · benchmarks AI agents on text-based adventure games, testing language understanding, strategic planning, and long-horizon reasoning.
GPT-4o-mini
17.4
Llama 3.2 90B
27.3
GeoBench
GPT-4o-mini leads by +12.0
GeoBench · tests geographic knowledge and spatial reasoning across countries, landmarks, coordinates, and geopolitical understanding.
GPT-4o-mini
64.0
Llama 3.2 90B
52.0
GPQA diamond
Llama 3.2 90B leads by +4.4
Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs.
GPT-4o-mini
17.0
Llama 3.2 90B
21.4
MATH level 5
GPT-4o-mini leads by +13.2
MATH Level 5 · the hardest tier of the MATH benchmark, featuring competition-level problems from AMC, AIME, and Olympiad-style mathematics.
GPT-4o-mini
52.6
Llama 3.2 90B
39.4
MMLU
GPT-4o-mini leads by +2.0
Massive Multitask Language Understanding · 57 subjects spanning STEM, humanities, social sciences, and more. The standard benchmark for broad knowledge.
GPT-4o-mini
75.7
Llama 3.2 90B
73.7
OTIS Mock AIME 2024-2025
GPT-4o-mini leads by +4.3
OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills.
GPT-4o-mini
6.8
Llama 3.2 90B
2.5
Full benchmark table
| Benchmark | GPT-4o-mini | Llama 3.2 90B |
|---|---|---|
Balrog Balrog · benchmarks AI agents on text-based adventure games, testing language understanding, strategic planning, and long-horizon reasoning. | 17.4 | 27.3 |
GeoBench GeoBench · tests geographic knowledge and spatial reasoning across countries, landmarks, coordinates, and geopolitical understanding. | 64.0 | 52.0 |
GPQA diamond Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs. | 17.0 | 21.4 |
MATH level 5 MATH Level 5 · the hardest tier of the MATH benchmark, featuring competition-level problems from AMC, AIME, and Olympiad-style mathematics. | 52.6 | 39.4 |
MMLU Massive Multitask Language Understanding · 57 subjects spanning STEM, humanities, social sciences, and more. The standard benchmark for broad knowledge. | 75.7 | 73.7 |
OTIS Mock AIME 2024-2025 OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills. | 6.8 | 2.5 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
| Model | Input | Output | Context | Projected $/mo |
|---|---|---|---|---|
| $0.15 | $0.60 | 128K tokens (~64 books) | $2.62 | |
| — | — | — | — |