Key Takeaways
Gemma 2 9B wins:
- Higher intelligence benchmark
Llama 3 8B Instruct wins:
- Cheaper output tokens
- Faster response time
- Better at coding
Price Advantage
Llama 3 8B Instruct
Benchmark Advantage
Gemma 2 9B
Context Window
Llama 3 8B Instruct
Speed
Llama 3 8B Instruct
Pricing Comparison
Price Comparison
| Metric | Gemma 2 9B | Llama 3 8B Instruct | Winner |
|---|---|---|---|
| Input (per 1M tokens) | $0.03 | $0.03 | Tie |
| Output (per 1M tokens) | $0.09 | $0.04 | Llama 3 8B Instruct |
Using a 3:1 input/output ratio, Llama 3 8B Instruct is 28% cheaper overall.
Gemma 2 9B Providers
Nebius $0.03 (Cheapest)
Novita $0.08
Llama 3 8B Instruct Providers
DeepInfra $0.03 (Cheapest)
Novita $0.04
Together $0.10
Cloudflare $0.28
Benchmark Comparison
7
Benchmarks Compared
1
Gemma 2 9B Wins
2
Llama 3 8B Instruct Wins
Benchmark Scores
| Benchmark | Gemma 2 9B | Llama 3 8B Instruct | Winner |
|---|---|---|---|
Intelligence Index Overall intelligence score | 32.1 | 8.7 | |
Coding Index Code generation & understanding | - | 4.0 | - |
MMLU Pro Academic knowledge | 31.9 | 40.5 | |
GPQA Graduate-level science | 14.8 | 29.6 | |
LiveCodeBench Competitive programming | - | 9.6 | - |
AIME Competition math | - | 0.0 | - |
BBH Big-Bench Hard | 42.1 | - | - |
Llama 3 8B Instruct wins 2 out of 7 benchmarks.
Cost vs Quality
X-axis:
Y-axis:
Loading chart...
Gemma 2 9B
Other models
Context & Performance
Context Window
Gemma 2 9B
8,192
tokens
Llama 3 8B Instruct
8,192
tokens
Max output: 16,384 tokens
Speed Performance
| Metric | Gemma 2 9B | Llama 3 8B Instruct | Winner |
|---|---|---|---|
| Tokens/second | N/A | 66.3 tok/s | |
| Time to First Token | N/A | 0.39s |
Capabilities
Feature Comparison
| Feature | Gemma 2 9B | Llama 3 8B Instruct |
|---|---|---|
| Vision (Image Input) | ||
| Tool/Function Calls | ||
| Reasoning Mode | ||
| Audio Input | ||
| Audio Output | ||
| PDF Input | ||
| Prompt Caching | ||
| Web Search |
License & Release
| Property | Gemma 2 9B | Llama 3 8B Instruct |
|---|---|---|
| License | Open Source | Open Source |
| Author | Meta-llama | |
| Released | Jun 2024 | Apr 2024 |
Gemma 2 9B Modalities
Input
text
Output
text
Llama 3 8B Instruct Modalities
Input
text
Output
text