25 lines
787 B
Markdown
25 lines
787 B
Markdown
|
---
|
||
|
|
type: model
|
||
|
|
name: Qwen 3.8
|
||
|
|
vendor: Alibaba
|
||
|
|
date: 2026-07-19
|
||
|
|
status: announced
|
||
|
|
tags: [qwen, alibaba, open-source, moe, china-ai]
|
||
|
|
---
|
||
|
|
|
||
|
|
# Qwen 3.8
|
||
|
|
|
||
|
|
## Spec
|
||
|
|
- **Parameters:** 2.4T (trillion)
|
||
|
|
- **Release:** Open-weight (open source)
|
||
|
|
- **Date:** Announced 2026-07-19
|
||
|
|
- **Vendor claim:** "One of the most powerful model available today, second only to Fable 5"
|
||
|
|
- **Early access:** Alibaba Token Plan, Qoder, QoderWork
|
||
|
|
|
||
|
|
## Context
|
||
|
|
- Follows Kimi-K3 (2.8T) announced same week — two massive Chinese open-source MoE releases in rapid succession
|
||
|
|
- Continues trend: Chinese labs releasing open-weight models at scales US labs keep closed
|
||
|
|
- "Fable 5" referenced as #1 — not a widely established benchmark reference
|
||
|
|
|
||
|
|
## Related
|
||
|
|
- [[../concepts/llm/kimi-k3.md]] — 2.8T, open-weight, same week
|