Key Takeaways
Qwen3 Max wins:
- Higher intelligence benchmark
- Better at coding
- Better at math
Qwen3 VL 235B A22B Instruct wins:
- Cheaper input tokens
- Cheaper output tokens
- Larger context window
- Faster response time
- Supports vision
Price Advantage
Qwen3 VL 235B A22B Instruct
Benchmark Advantage
Qwen3 Max
Context Window
Qwen3 VL 235B A22B Instruct
Speed
Qwen3 VL 235B A22B Instruct
Pricing Comparison
Price Comparison
| Metric | Qwen3 Max | Qwen3 VL 235B A22B Instruct | Winner |
|---|---|---|---|
| Input (per 1M tokens) | $1.20 | $0.20 | Qwen3 VL 235B A22B Instruct |
| Output (per 1M tokens) | $6.00 | $0.88 | Qwen3 VL 235B A22B Instruct |
| Cache Read (per 1M) | $240000.00 | $110000.00 | Qwen3 VL 235B A22B Instruct |
Using a 3:1 input/output ratio, Qwen3 VL 235B A22B Instruct is 85% cheaper overall.
Qwen3 Max Providers
Alibaba $1.20 (Cheapest)
Qwen3 VL 235B A22B Instruct Providers
DeepInfra $0.20 (Cheapest)
Parasail $0.21
Fireworks $0.22
AtlasCloud $0.30
SiliconFlow $0.30
Benchmark Comparison
6
Benchmarks Compared
6
Qwen3 Max Wins
0
Qwen3 VL 235B A22B Instruct Wins
Benchmark Scores
| Benchmark | Qwen3 Max | Qwen3 VL 235B A22B Instruct | Winner |
|---|---|---|---|
Intelligence Index Overall intelligence score | 31.3 | 20.6 | |
Coding Index Code generation & understanding | 26.4 | 16.5 | |
Math Index Mathematical reasoning | 80.7 | 70.7 | |
MMLU Pro Academic knowledge | 84.1 | 82.3 | |
GPQA Graduate-level science | 76.4 | 71.2 | |
LiveCodeBench Competitive programming | 76.7 | 59.4 |
Qwen3 Max significantly outperforms in coding benchmarks.
Cost vs Quality
X-axis:
Y-axis:
Loading chart...
Qwen3 Max
Other models
Context & Performance
Context Window
Qwen3 Max
256,000
tokens
Max output: 32,768 tokens
Qwen3 VL 235B A22B Instruct
262,144
tokens
Qwen3 VL 235B A22B Instruct has a 2% larger context window.
Speed Performance
| Metric | Qwen3 Max | Qwen3 VL 235B A22B Instruct | Winner |
|---|---|---|---|
| Tokens/second | 27.1 tok/s | 45.9 tok/s | |
| Time to First Token | 2.20s | 1.10s |
Qwen3 VL 235B A22B Instruct responds 70% faster on average.
Capabilities
Feature Comparison
| Feature | Qwen3 Max | Qwen3 VL 235B A22B Instruct |
|---|---|---|
| Vision (Image Input) | ||
| Tool/Function Calls | ||
| Reasoning Mode | ||
| Audio Input | ||
| Audio Output | ||
| PDF Input | ||
| Prompt Caching | ||
| Web Search |
License & Release
| Property | Qwen3 Max | Qwen3 VL 235B A22B Instruct |
|---|---|---|
| License | Open Source | Open Source |
| Author | Qwen | Qwen |
| Released | Sep 2025 | Sep 2025 |
Qwen3 Max Modalities
Input
text
Output
text
Qwen3 VL 235B A22B Instruct Modalities
Input
textimage
Output
text