Key Takeaways
Devstral 2 2512 wins:
- Larger context window
- Higher intelligence benchmark
- Better at coding
- Better at math
Llama 3.3 Nemotron Super 49B V1.5 wins:
- Cheaper input tokens
- Cheaper output tokens
- Faster response time
- Has reasoning mode
Price Advantage
Llama 3.3 Nemotron Super 49B V1.5
Benchmark Advantage
Devstral 2 2512
Context Window
Devstral 2 2512
Speed
Llama 3.3 Nemotron Super 49B V1.5
Pricing Comparison
Price Comparison
| Metric | Devstral 2 2512 | Llama 3.3 Nemotron Super 49B V1.5 | Winner |
|---|---|---|---|
| Input (per 1M tokens) | $0.40 | $0.10 | Llama 3.3 Nemotron Super 49B V1.5 |
| Output (per 1M tokens) | $0.90 | $0.40 | Llama 3.3 Nemotron Super 49B V1.5 |
| Cache Read (per 1M) | $0.45 | N/A | Devstral 2 2512 |
Using a 3:1 input/output ratio, Llama 3.3 Nemotron Super 49B V1.5 is 67% cheaper overall.
Devstral 2 2512 Providers
No provider data available
Llama 3.3 Nemotron Super 49B V1.5 Providers
No provider data available
Benchmark Comparison
7
Benchmarks Compared
6
Devstral 2 2512 Wins
0
Llama 3.3 Nemotron Super 49B V1.5 Wins
Benchmark Scores
| Benchmark | Devstral 2 2512 | Llama 3.3 Nemotron Super 49B V1.5 | Winner |
|---|---|---|---|
Intelligence Index Overall intelligence score | 22.0 | 14.6 | |
Coding Index Code generation & understanding | 23.7 | 10.5 | |
Math Index Mathematical reasoning | 36.7 | 8.0 | |
MMLU Pro Academic knowledge | 76.2 | 69.2 | |
GPQA Graduate-level science | 59.4 | 48.1 | |
LiveCodeBench Competitive programming | 44.8 | 29.0 | |
AIME Competition math | - | 13.7 | - |
Devstral 2 2512 significantly outperforms in coding benchmarks.
Cost vs Quality
X-axis:
Y-axis:
Loading chart...
Devstral 2 2512
Other models
Context & Performance
Context Window
Devstral 2 2512
262,144
tokens
Llama 3.3 Nemotron Super 49B V1.5
131,072
tokens
Devstral 2 2512 has a 50% larger context window.
Speed Performance
| Metric | Devstral 2 2512 | Llama 3.3 Nemotron Super 49B V1.5 | Winner |
|---|---|---|---|
| Tokens/second | 81.2 tok/s | 82.7 tok/s | |
| Time to First Token | 0.39s | 0.24s |
Capabilities
Feature Comparison
| Feature | Devstral 2 2512 | Llama 3.3 Nemotron Super 49B V1.5 |
|---|---|---|
| Vision (Image Input) | ||
| Tool/Function Calls | ||
| Reasoning Mode | ||
| Audio Input | ||
| Audio Output | ||
| PDF Input | ||
| Prompt Caching | ||
| Web Search |
License & Release
| Property | Devstral 2 2512 | Llama 3.3 Nemotron Super 49B V1.5 |
|---|---|---|
| License | Open Source | Proprietary |
| Author | Mistral AI | Nvidia |
| Released | Dec 2025 | Oct 2025 |
Devstral 2 2512 Modalities
Input
text
Output
text
Llama 3.3 Nemotron Super 49B V1.5 Modalities
Input
text
Output
text