Key Takeaways
DeepSeek V3.1 wins:
- Higher intelligence benchmark
- Better at coding
Qwen3 VL 30B A3B Instruct wins:
- Cheaper input tokens
- Cheaper output tokens
- Larger context window
- Faster response time
- Better at math
- Supports vision
Price Advantage
Qwen3 VL 30B A3B Instruct
Benchmark Advantage
DeepSeek V3.1
Context Window
Qwen3 VL 30B A3B Instruct
Speed
Qwen3 VL 30B A3B Instruct
Pricing Comparison
Price Comparison
| Metric | DeepSeek V3.1 | Qwen3 VL 30B A3B Instruct | Winner |
|---|---|---|---|
| Input (per 1M tokens) | $0.15 | $0.13 | Qwen3 VL 30B A3B Instruct |
| Output (per 1M tokens) | $0.75 | $0.52 | Qwen3 VL 30B A3B Instruct |
Using a 3:1 input/output ratio, Qwen3 VL 30B A3B Instruct is 24% cheaper overall.
DeepSeek V3.1 Providers
SambaNova $0.15 (Cheapest)
Chutes $0.20
DeepInfra $0.21
SiliconFlow $0.27
Novita $0.27
Qwen3 VL 30B A3B Instruct Providers
Alibaba $0.13 (Cheapest)
DeepInfra $0.15
Phala $0.20
Novita $0.20
SiliconFlow $0.29
Benchmark Comparison
6
Benchmarks Compared
5
DeepSeek V3.1 Wins
1
Qwen3 VL 30B A3B Instruct Wins
Benchmark Scores
| Benchmark | DeepSeek V3.1 | Qwen3 VL 30B A3B Instruct | Winner |
|---|---|---|---|
Intelligence Index Overall intelligence score | 28.0 | 16.0 | |
Coding Index Code generation & understanding | 28.4 | 14.3 | |
Math Index Mathematical reasoning | 49.7 | 72.3 | |
MMLU Pro Academic knowledge | 83.3 | 76.4 | |
GPQA Graduate-level science | 73.5 | 69.5 | |
LiveCodeBench Competitive programming | 57.7 | 47.6 |
DeepSeek V3.1 significantly outperforms in coding benchmarks.
Cost vs Quality
X-axis:
Y-axis:
Loading chart...
DeepSeek V3.1
Other models
Context & Performance
Context Window
DeepSeek V3.1
32,768
tokens
Max output: 7,168 tokens
Qwen3 VL 30B A3B Instruct
131,072
tokens
Max output: 32,768 tokens
Qwen3 VL 30B A3B Instruct has a 75% larger context window.
Speed Performance
| Metric | DeepSeek V3.1 | Qwen3 VL 30B A3B Instruct | Winner |
|---|---|---|---|
| Tokens/second | 0.0 tok/s | 105.4 tok/s | |
| Time to First Token | 0.00s | 0.92s |
Qwen3 VL 30B A3B Instruct responds Infinity% faster on average.
Capabilities
Feature Comparison
| Feature | DeepSeek V3.1 | Qwen3 VL 30B A3B Instruct |
|---|---|---|
| Vision (Image Input) | ||
| Tool/Function Calls | ||
| Reasoning Mode | ||
| Audio Input | ||
| Audio Output | ||
| PDF Input | ||
| Prompt Caching | ||
| Web Search |
License & Release
| Property | DeepSeek V3.1 | Qwen3 VL 30B A3B Instruct |
|---|---|---|
| License | Open Source | Open Source |
| Author | Deepseek | Qwen |
| Released | Aug 2025 | Oct 2025 |
DeepSeek V3.1 Modalities
Input
text
Output
text
Qwen3 VL 30B A3B Instruct Modalities
Input
textimage
Output
text