Key Takeaways
Qwen2.5 72B Instruct wins:
- Cheaper input tokens
- Cheaper output tokens
- Faster response time
Qwen3 Max wins:
- Larger context window
- Higher intelligence benchmark
- Better at coding
- Better at math
Price Advantage
Qwen2.5 72B Instruct
Benchmark Advantage
Qwen3 Max
Context Window
Qwen3 Max
Speed
Qwen2.5 72B Instruct
Pricing Comparison
Price Comparison
| Metric | Qwen2.5 72B Instruct | Qwen3 Max | Winner |
|---|---|---|---|
| Input (per 1M tokens) | $0.12 | $1.20 | Qwen2.5 72B Instruct |
| Output (per 1M tokens) | $0.39 | $6.00 | Qwen2.5 72B Instruct |
| Cache Read (per 1M) | N/A | $240000.00 | Qwen3 Max |
Using a 3:1 input/output ratio, Qwen2.5 72B Instruct is 92% cheaper overall.
Qwen2.5 72B Instruct Providers
DeepInfra $0.12 (Cheapest)
Chutes $0.30
Novita $0.38
Hyperbolic $0.40
Qwen3 Max Providers
Alibaba $1.20 (Cheapest)
Benchmark Comparison
8
Benchmarks Compared
0
Qwen2.5 72B Instruct Wins
6
Qwen3 Max Wins
Benchmark Scores
| Benchmark | Qwen2.5 72B Instruct | Qwen3 Max | Winner |
|---|---|---|---|
Intelligence Index Overall intelligence score | 15.6 | 31.3 | |
Coding Index Code generation & understanding | 11.9 | 26.4 | |
Math Index Mathematical reasoning | 14.0 | 80.7 | |
MMLU Pro Academic knowledge | 72.0 | 84.1 | |
GPQA Graduate-level science | 49.1 | 76.4 | |
LiveCodeBench Competitive programming | 27.6 | 76.7 | |
Aider Real-world code editing | 65.4 | - | - |
AIME Competition math | 16.0 | - | - |
Qwen3 Max significantly outperforms in coding benchmarks.
Cost vs Quality
X-axis:
Y-axis:
Loading chart...
Qwen2.5 72B Instruct
Other models
Context & Performance
Context Window
Qwen2.5 72B Instruct
32,768
tokens
Max output: 16,384 tokens
Qwen3 Max
262,144
tokens
Max output: 65,536 tokens
Qwen3 Max has a 88% larger context window.
Speed Performance
| Metric | Qwen2.5 72B Instruct | Qwen3 Max | Winner |
|---|---|---|---|
| Tokens/second | 45.6 tok/s | 27.0 tok/s | |
| Time to First Token | 1.11s | 2.37s |
Qwen2.5 72B Instruct responds 69% faster on average.
Capabilities
Feature Comparison
| Feature | Qwen2.5 72B Instruct | Qwen3 Max |
|---|---|---|
| Vision (Image Input) | ||
| Tool/Function Calls | ||
| Reasoning Mode | ||
| Audio Input | ||
| Audio Output | ||
| PDF Input | ||
| Prompt Caching | ||
| Web Search |
License & Release
| Property | Qwen2.5 72B Instruct | Qwen3 Max |
|---|---|---|
| License | Open Source | Open Source |
| Author | Qwen | Qwen |
| Released | Sep 2024 | Sep 2025 |
Qwen2.5 72B Instruct Modalities
Input
text
Output
text
Qwen3 Max Modalities
Input
text
Output
text