Qwen3.7-Max vs DeepSeek-V4-Pro

Alibaba's Qwen3.7-Max and DeepSeek's DeepSeek-V4-Pro compared on specs, pricing, and capabilities — as configured in AI Crucible ensembles.

DeepSeek-V4-Pro input tokens are 5.7x cheaper than Qwen3.7-Max ($0.52 vs $3.00 per 1M tokens).
DeepSeek-V4-Pro output tokens are 8.6x cheaper than Qwen3.7-Max ($1.04 vs $9.00 per 1M tokens).
Qwen3.7-Max accepts image inputs; DeepSeek-V4-Pro is text-only in AI Crucible.

Specification	Qwen3.7-Max	DeepSeek-V4-Pro
Provider	Alibaba	DeepSeek
API model ID	qwen3.7-max	deepseek-v4-pro
Description	Alibaba's flagship Qwen3.7 model with 1M context and native extended-thinking, leading on SWE-Pro and Terminal-Bench agentic benchmarks.	DeepSeek's most capable V4 model with dual thinking/non-thinking modes and 1M context.
Context window	1M tokens	1M tokens
Max output tokens	65,536	8,192
Input cost ($/1M tokens)	$3.00	$0.52
Output cost ($/1M tokens)	$9.00	$1.04
Cache read cost ($/1M tokens)	$0.30	$0.00
Latency	medium	medium
Reasoning model	Yes	Yes
Vision (image input)	Yes	No
Tool use	Yes	Yes