Key Takeaways
Llama 3.2 3B Instruct wins:
- Cheaper input tokens
- Cheaper output tokens
- Supports tool calls
GPT-OSS-20b wins:
- Faster response time
- Higher intelligence benchmark
- Better at coding
- Better at math
Price Advantage
Llama 3.2 3B Instruct
Benchmark Advantage
GPT-OSS-20b
Context Window
GPT-OSS-20b
Speed
GPT-OSS-20b
Pricing Comparison
Price Comparison
| Metric | Llama 3.2 3B Instruct | GPT-OSS-20b | Winner |
|---|---|---|---|
| Input (per 1M tokens) | $0.02 | $0.03 | Llama 3.2 3B Instruct |
| Output (per 1M tokens) | $0.02 | $0.14 | Llama 3.2 3B Instruct |
Using a 3:1 input/output ratio, Llama 3.2 3B Instruct is 65% cheaper overall.
Llama 3.2 3B Instruct Providers
DeepInfra $0.02 (Cheapest)
Novita $0.03
Cloudflare $0.05
Together $0.06
Hyperbolic $0.10
GPT-OSS-20b Providers
DeepInfra $0.03 (Cheapest)
Parasail $0.04
Chutes $0.04
Novita $0.04
Clarifai $0.04
Benchmark Comparison
7
Benchmarks Compared
0
Llama 3.2 3B Instruct Wins
5
GPT-OSS-20b Wins
Benchmark Scores
| Benchmark | Llama 3.2 3B Instruct | GPT-OSS-20b | Winner |
|---|---|---|---|
Intelligence Index Overall intelligence score | 9.7 | 24.5 | |
Coding Index Code generation & understanding | - | 18.5 | - |
Math Index Mathematical reasoning | 3.3 | 89.3 | |
MMLU Pro Academic knowledge | 34.7 | 74.8 | |
GPQA Graduate-level science | 25.5 | 68.8 | |
LiveCodeBench Competitive programming | 8.3 | 77.7 | |
AIME Competition math | 6.7 | - | - |
GPT-OSS-20b significantly outperforms in coding benchmarks.
Cost vs Quality
X-axis:
Y-axis:
Loading chart...
Llama 3.2 3B Instruct
Other models
Context & Performance
Context Window
Llama 3.2 3B Instruct
131,072
tokens
Max output: 16,384 tokens
GPT-OSS-20b
131,072
tokens
Speed Performance
| Metric | Llama 3.2 3B Instruct | GPT-OSS-20b | Winner |
|---|---|---|---|
| Tokens/second | 44.1 tok/s | 307.3 tok/s | |
| Time to First Token | 0.43s | 0.52s |
GPT-OSS-20b responds 597% faster on average.
Capabilities
Feature Comparison
| Feature | Llama 3.2 3B Instruct | GPT-OSS-20b |
|---|---|---|
| Vision (Image Input) | ||
| Tool/Function Calls | ||
| Reasoning Mode | ||
| Audio Input | ||
| Audio Output | ||
| PDF Input | ||
| Prompt Caching | ||
| Web Search |
License & Release
| Property | Llama 3.2 3B Instruct | GPT-OSS-20b |
|---|---|---|
| License | Open Source | Open Source |
| Author | Meta-llama | OpenAI |
| Released | Sep 2024 | Aug 2025 |
Llama 3.2 3B Instruct Modalities
Input
text
Output
text
GPT-OSS-20b Modalities
Input
text
Output
text