Is Qwen3.6 or Qwen3.8 better?
Is Qwen3.6 or Qwen3.8 better?
Solved Qwen3.8 27B
qwen3.8-27b-nvfp4
- Views
- 1
- Replies
- 3
- Created
- 2026-08-25 16:15
- Last reply
- 2026-08-25 22:34
For a new general-purpose workload, I would usually start by testing Qwen3.8 because it is the newer generation and the one AxForge currently exposes serverlessly.
Qwen3.6 35B-A3B is still interesting as a managed deployment because its MoE architecture can be very fast on GB10.
So 'better' depends on quality, latency and your task. Qwen3.8 is the simpler API starting point; Qwen3.6 can be a strong dedicated-throughput choice.
Sign in to reply.
Posting guidelines
- Be helpful and respectful
- Search before you post
- Use a clear title and tags
- Add code or a screenshot when it helps
- Keep to the topic
Accounts that break these can lose forum access — paid plans included. Read the full guidelines →
Related topics
- Model requests — ask for a model we don't have yet 0 replies · 2 views
- Request: Qwen3 30B A3B 0 replies · 2 views
- Request: LTX-Video 0 replies · 2 views
- Request: MiniMax H3 0 replies · 2 views
- How much context does Qwen3.8 have on AxForge? 2 replies · 1 views