Compare · ModelsLive · 2 picked · head to head
Gemini 1.5 Pro (May 2024) vs Llama 2-13B
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
Gemini 1.5 Pro (May 2024) wins on 6/6 benchmarks
Gemini 1.5 Pro (May 2024) wins 6 of 6 shared benchmarks. Leads in reasoning · general · knowledge.
Category leads
reasoning·Gemini 1.5 Pro (May 2024)general·Gemini 1.5 Pro (May 2024)knowledge·Gemini 1.5 Pro (May 2024)math·Gemini 1.5 Pro (May 2024)
Hype vs Reality
Attention vs performance
Gemini 1.5 Pro (May 2024)
#208 by perf·no signal
Llama 2-13B
#231 by perf·#19 by attention
Best value
Pricing unknown
Gemini 1.5 Pro (May 2024)
n/a
no price
Llama 2-13B
n/a
no price
Vendor risk
Who is behind the model
Google DeepMind
$4.27T·Tier 1
Meta AI
$1.89T·Tier 1
Head to head
6 benchmarks · 2 models
Gemini 1.5 Pro (May 2024)Llama 2-13B
BBH
Gemini 1.5 Pro (May 2024) leads by +41.3
BIG-Bench Hard · a curated subset of 23 challenging tasks from BIG-Bench where language models previously failed to outperform average humans.
Gemini 1.5 Pro (May 2024)
85.6
Llama 2-13B
44.3
Dtbench
Gemini 1.5 Pro (May 2024) leads by +23.7
Gemini 1.5 Pro (May 2024)
27.4
Llama 2-13B
3.7
GPQA diamond
Gemini 1.5 Pro (May 2024) leads by +26.1
Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs.
Gemini 1.5 Pro (May 2024)
27.8
Llama 2-13B
1.8
MATH level 5
Gemini 1.5 Pro (May 2024) leads by +37.5
MATH Level 5 · the hardest tier of the MATH benchmark, featuring competition-level problems from AMC, AIME, and Olympiad-style mathematics.
Gemini 1.5 Pro (May 2024)
40.8
Llama 2-13B
3.3
MMLU
Gemini 1.5 Pro (May 2024) leads by +40.4
Massive Multitask Language Understanding · 57 subjects spanning STEM, humanities, social sciences, and more. The standard benchmark for broad knowledge.
Gemini 1.5 Pro (May 2024)
81.2
Llama 2-13B
40.8
OTIS Mock AIME 2024-2025
Gemini 1.5 Pro (May 2024) leads by +6.7
OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills.
Gemini 1.5 Pro (May 2024)
6.7
Llama 2-13B
0.0
Full benchmark table
| Benchmark | Gemini 1.5 Pro (May 2024) | Llama 2-13B |
|---|---|---|
BBH BIG-Bench Hard · a curated subset of 23 challenging tasks from BIG-Bench where language models previously failed to outperform average humans. | 85.6 | 44.3 |
Dtbench | 27.4 | 3.7 |
GPQA diamond Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs. | 27.8 | 1.8 |
MATH level 5 MATH Level 5 · the hardest tier of the MATH benchmark, featuring competition-level problems from AMC, AIME, and Olympiad-style mathematics. | 40.8 | 3.3 |
MMLU Massive Multitask Language Understanding · 57 subjects spanning STEM, humanities, social sciences, and more. The standard benchmark for broad knowledge. | 81.2 | 40.8 |
OTIS Mock AIME 2024-2025 OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills. | 6.7 | 0.0 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
| Model | Input | Output | Context | Projected $/mo |
|---|---|---|---|---|
| — | — | — | — | |
| — | — | — | — |