Compare · ModelsLive · 2 picked · head to head
LLaMA-13B vs Llama 2-13B
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
Llama 2-13B wins on 10/11 benchmarks
Llama 2-13B wins 10 of 11 shared benchmarks. Leads in knowledge · reasoning · math.
Category leads
knowledge·Llama 2-13Breasoning·Llama 2-13Bmath·Llama 2-13B
Hype vs Reality
Attention vs performance
LLaMA-13B
#168 by perf·no signal
Llama 2-13B
#126 by perf·no signal
Vendor risk
Who is behind the model
Meta AI
$1.50T·Tier 1
Meta AI
$1.50T·Tier 1
Head to head
11 benchmarks · 2 models
LLaMA-13BLlama 2-13B
ARC AI2
Llama 2-13B leads by +10.1
AI2 Reasoning Challenge · tests grade-school level science knowledge with multiple-choice questions requiring reasoning beyond simple retrieval.
LLaMA-13B
36.9
Llama 2-13B
47.1
BBH
Llama 2-13B leads by +27.1
BIG-Bench Hard · a curated subset of 23 challenging tasks from BIG-Bench where language models previously failed to outperform average humans.
LLaMA-13B
17.2
Llama 2-13B
44.3
GSM8K
Llama 2-13B leads by +16.3
Grade School Math 8K · 8,500 linguistically diverse grade-school math word problems that require multi-step reasoning to solve.
LLaMA-13B
20.6
Llama 2-13B
36.9
HellaSwag
Llama 2-13B leads by +2.0
HellaSwag · tests commonsense reasoning by asking models to predict the most plausible continuation of everyday scenarios.
LLaMA-13B
72.3
Llama 2-13B
74.3
LAMBADA
Llama 2-13B leads by +1.3
LAMBADA · measures the ability to predict the final word of a passage, requiring broad contextual understanding across long text spans.
LLaMA-13B
75.2
Llama 2-13B
76.5
MMLU
Llama 2-13B leads by +10.5
Massive Multitask Language Understanding · 57 subjects spanning STEM, humanities, social sciences, and more. The standard benchmark for broad knowledge.
LLaMA-13B
30.3
Llama 2-13B
40.8
OpenBookQA
Llama 2-13B leads by +0.8
OpenBookQA · science questions that require combining a given core fact with broad common knowledge, mimicking an open-book exam setting.
LLaMA-13B
41.9
Llama 2-13B
42.7
PIQA
Llama 2-13B leads by +1.4
PIQA (Physical Interaction QA) · tests intuitive physical reasoning by asking models to select the correct approach for everyday physical tasks.
LLaMA-13B
60.2
Llama 2-13B
61.6
ScienceQA
Llama 2-13B leads by +16.6
ScienceQA · multimodal science questions spanning natural science, social science, and language science with diverse question formats and image context.
LLaMA-13B
24.4
Llama 2-13B
41.0
TriviaQA
Llama 2-13B leads by +1.7
TriviaQA · reading comprehension benchmark with trivia questions, requiring models to find and reason over evidence from provided documents.
LLaMA-13B
77.9
Llama 2-13B
79.6
Winogrande
LLaMA-13B leads by +0.4
WinoGrande · large-scale commonsense reasoning benchmark where models must resolve ambiguous pronouns in carefully constructed sentence pairs.
LLaMA-13B
46.0
Llama 2-13B
45.6
Full benchmark table
| Benchmark | LLaMA-13B | Llama 2-13B |
|---|---|---|
ARC AI2 AI2 Reasoning Challenge · tests grade-school level science knowledge with multiple-choice questions requiring reasoning beyond simple retrieval. | 36.9 | 47.1 |
BBH BIG-Bench Hard · a curated subset of 23 challenging tasks from BIG-Bench where language models previously failed to outperform average humans. | 17.2 | 44.3 |
GSM8K Grade School Math 8K · 8,500 linguistically diverse grade-school math word problems that require multi-step reasoning to solve. | 20.6 | 36.9 |
HellaSwag HellaSwag · tests commonsense reasoning by asking models to predict the most plausible continuation of everyday scenarios. | 72.3 | 74.3 |
LAMBADA LAMBADA · measures the ability to predict the final word of a passage, requiring broad contextual understanding across long text spans. | 75.2 | 76.5 |
MMLU Massive Multitask Language Understanding · 57 subjects spanning STEM, humanities, social sciences, and more. The standard benchmark for broad knowledge. | 30.3 | 40.8 |
OpenBookQA OpenBookQA · science questions that require combining a given core fact with broad common knowledge, mimicking an open-book exam setting. | 41.9 | 42.7 |
PIQA PIQA (Physical Interaction QA) · tests intuitive physical reasoning by asking models to select the correct approach for everyday physical tasks. | 60.2 | 61.6 |
ScienceQA ScienceQA · multimodal science questions spanning natural science, social science, and language science with diverse question formats and image context. | 24.4 | 41.0 |
TriviaQA TriviaQA · reading comprehension benchmark with trivia questions, requiring models to find and reason over evidence from provided documents. | 77.9 | 79.6 |
Winogrande WinoGrande · large-scale commonsense reasoning benchmark where models must resolve ambiguous pronouns in carefully constructed sentence pairs. | 46.0 | 45.6 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
| Model | Input | Output | Context | Projected $/mo |
|---|---|---|---|---|
| LLaMA-13B | — | — | — | — |
| Llama 2-13B | — | — | — | — |