modelbenchmark.io

Qwen3.8 Max Prime

Alibaba · qwen3-8-max-prime

Compare

High-throughput edition of Qwen3.8 Max for coding, professional work, multimodal understanding, and long-running agent workflows

Specification

most-agreed values

Context
1M
Max output
131K
Released
2026-09-23
Knowledge cutoff
—
Retires
—
Open weights
no
Input
text, image, video
Output
text

Price

US dollars per million tokens · most-agreed

Input
$4.00
Output
$12.00
Cache read
$0.50
Cache write
$5.00

Available from 4 hosts

HostIn $/MOut $/MCache rdCache wrContextOutputRetires
kilo$4.00$12.00$0.50—1M131K—
nano-gpt$4.00$12.00$0.50—1M131K—
openrouter$4.00$12.00$0.50—1M131K—
vercel$4.00$12.00$0.50$5.001M131K—

More from Alibaba

most-hosted first