Models
What the club runs now, what members are voting on, and the seats, all in the open.
0
seats filled
1
model at opening
–
seats until the next vote
At opening
| Model | Context (tokens) | Longest reply (tokens) | Thinking |
|---|---|---|---|
| qwen3.8-27b | 262,144 | 32,768 | on: slower, more careful |
| qwen3.8-27b-fast | 262,144 | 32,768 | off: faster replies |
Use the -fast model for quick edits and chat, and the other for harder problems. Switching between them in one conversation starts its cache afresh.
The next vote
No vote is open right now. The next one will be announced in our Discord.
On the shortlist so far:
qwen3.8-flash-next
Much faster per token; needs a bigger GPU
qwen3.6-35b-a3b
750k context, about 2.5x faster
Anyone can read along in our Discord; members can post.