Qwen 2.5 Max vs TypeSafe AI Jev

A direct, empirical evaluation of Qwen 2.5 Max (Alibaba Cloud) against TypeSafe AI Jev (TypeSafe AI (Diogo Almeida)).

Qwen 2.5 Max

Developed by Alibaba Cloud · Proprietary (~700B) parameters

Cross-lingual document extraction and structured structured output generation.

TypeSafe AI Jev

Developed by TypeSafe AI (Diogo Almeida) · Non-Autoregressive System 1 Transformer parameters

Sub-15ms operational tasks, 100% schema-valid JSON extraction, deterministic tool dispatch, and high-throughput agent routing.

Metric Qwen 2.5 Max TypeSafe AI Jev
SWE-bench Verified 51.4% 68.4%
MMLU-Pro 77.2% 84.6%
MATH-500 89.4% 91.2%
Context Window 128,000 tokens 32,000 tokens
Input Pricing (per 1M) $1.60 $0.06
Output Pricing (per 1M) $6.40 $0.12