Key Takeaways
DeepSeek V3.2 wins:
- Cheaper input tokens
- Cheaper output tokens
- Larger context window
- Faster response time
- Higher intelligence benchmark
- Better at coding
- Better at math
- Has reasoning mode
Phi-3 Medium 128K Instruct wins:
- No clear advantages in compared metrics
Price Advantage
DeepSeek V3.2
Benchmark Advantage
DeepSeek V3.2
Context Window
DeepSeek V3.2
Speed
DeepSeek V3.2
Pricing Comparison
Price Comparison
| Metric | DeepSeek V3.2 | Phi-3 Medium 128K Instruct | Winner |
|---|---|---|---|
| Input (per 1M tokens) | $0.26 | $1.00 | DeepSeek V3.2 |
| Output (per 1M tokens) | $0.38 | $1.00 | DeepSeek V3.2 |
| Cache Read (per 1M) | $0.03 | N/A | DeepSeek V3.2 |
Using a 3:1 input/output ratio, DeepSeek V3.2 is 71% cheaper overall.
DeepSeek V3.2 Providers
No provider data available
Phi-3 Medium 128K Instruct Providers
No provider data available
Benchmark Comparison
7
Benchmarks Compared
2
DeepSeek V3.2 Wins
0
Phi-3 Medium 128K Instruct Wins
Benchmark Scores
| Benchmark | DeepSeek V3.2 | Phi-3 Medium 128K Instruct | Winner |
|---|---|---|---|
Intelligence Index Overall intelligence score | 32.1 | 32.0 | - |
Coding Index Code generation & understanding | 34.6 | - | - |
Math Index Mathematical reasoning | 59.0 | - | - |
MMLU Pro Academic knowledge | 83.7 | 41.2 | |
GPQA Graduate-level science | 75.1 | 11.5 | |
LiveCodeBench Competitive programming | 59.3 | - | - |
BBH Big-Bench Hard | - | 48.5 | - |
DeepSeek V3.2 significantly outperforms in coding benchmarks.
Cost vs Quality
X-axis:
Y-axis:
Loading chart...
DeepSeek V3.2
Other models
Context & Performance
Context Window
DeepSeek V3.2
163,840
tokens
Phi-3 Medium 128K Instruct
128,000
tokens
DeepSeek V3.2 has a 22% larger context window.
Speed Performance
| Metric | DeepSeek V3.2 | Phi-3 Medium 128K Instruct | Winner |
|---|---|---|---|
| Tokens/second | 24.4 tok/s | N/A | |
| Time to First Token | 1.43s | N/A |
Capabilities
Feature Comparison
| Feature | DeepSeek V3.2 | Phi-3 Medium 128K Instruct |
|---|---|---|
| Vision (Image Input) | ||
| Tool/Function Calls | ||
| Reasoning Mode | ||
| Audio Input | ||
| Audio Output | ||
| PDF Input | ||
| Prompt Caching | ||
| Web Search |
License & Release
| Property | DeepSeek V3.2 | Phi-3 Medium 128K Instruct |
|---|---|---|
| License | Open Source | Open Source |
| Author | Deepseek | Microsoft |
| Released | Dec 2025 | Unknown |
DeepSeek V3.2 Modalities
Input
text
Output
text
Phi-3 Medium 128K Instruct Modalities
Input
Output