Qwen 2.5 Max vs Qwen3 72B Instruct

Capability, benchmark, and verified hosting price comparison.

Quick verdict

Cheapest hosting: Qwen3 72B Instruct— $0.900 vs $2.800 / 1M blended tokens
Context window: Qwen3 72B Instruct— 131,072 tokens vs 32,768
Weights available: Qwen3 72B Instruct— open-weight (Apache 2.0) vs closed

Derived from public benchmarks and verified hosting prices. Benchmark wins count only where both models have a score; direction respects each benchmark’s higher-is-better flag.

Qwen 2.5 Max

by Alibaba Cloud · Qwen2.5

Context: 32,768 tokens
Open weight: No
Modalities: text

Qwen3 72B Instruct

by Alibaba · Qwen

Context: 131,072 tokens
Open weight: Yes
License: Apache 2.0
Params: 72.0B
Modalities: text

Benchmarks

Benchmark	Qwen 2.5 Max	Qwen3 72B Instruct
HumanEval	—	0.821
MMLU	—	0.859

Cheapest verified hosting

Qwen 2.5 Max cheapest

Alibaba Cloud DashScope

$2.800/ 1M tokens (blended)

Input: $1.600
Output: $6.400
Context: 32,768 tok

Qwen3 72B Instruct cheapest

Together.ai

$0.900/ 1M tokens (blended)

Input: $0.900
Output: $0.900
Context: 131,072 tok
Latency p50: 220 ms