Qwen3.7-Max
1M context, model=qwen3.7-max, it is on the list
Official qwen.ai/blog?id=qwen3.7 (May 2026). Call it as qwen3.7-max, 1M context, max output often cited as 65536, thinking via extra_body.enable_thinking. Alibaba’s list is commonly $2.50/$7.50 per 1M (Artificial Analysis cites this official Alibaba API). QCode /models has this id. Newer 3.8-Max is not in the catalog.
What you actually send
API id
Set model="qwen3.7-max". QCode /models lists it.
Context
Official 1M-token window.
Max out (often cited)
Docs and aggregators often cite a 65536 max output. We do not invent another number.
Alibaba list
Common official Alibaba API: $2.50/$7.50 per 1M. Artificial Analysis cites that table. QCode bills official list × service rate.
What 3.7-Max is
Qwen3.7-Max is the hosted Qwen flagship that came before 3.8-Max. Official write-up: qwen.ai/blog?id=qwen3.7 (May 2026). Public id qwen3.7-max, 1M context. Thinking is commonly on by default; flip it with extra_body.enable_thinking.
3.8 is newer. It is not here.
Qwen3.8-Max (2026-08-03) is newer and the published hosted rate is lower ($2/$6). QCode /models does not list qwen3.8-max. On this site, Qwen is still qwen3.7-max (and qwen3.7-plus). “The industry has 3.8” is not “you can call 3.8 here.”
Timeline
Official blog qwen.ai/blog?id=qwen3.7. 3.7-Max ships as the hosted flagship.
QCode /models lists qwen3.7-max and qwen3.7-plus.
3.8-Max ships. Newer. Not listed on QCode.
Confirmed vs treat carefully
Confirmed
Official blog May 2026; id qwen3.7-max; 1M context; enable_thinking in extra_body; Alibaba list commonly $2.50/$7.50 (AA cites the official Alibaba API). QCode /models has qwen3.7-max.
Treat carefully
Do not treat 3.8-Max as a QCode id. 65536 is a commonly cited output cap, not a rerun here. QCode = official list × service rate; the dashboard wins.
Stay on 3.7 or chase 3.8
Stay on 3.7-Max
The id is listed, 1M, thinking is switchable, the list price is known. Lock production here.
Chase 3.8-Max
Newer, but not on QCode. Use official hosting or your own weights. Do not look for qwen3.8-max here.
How to call it
Set model=qwen3.7-max. Put the thinking switch on extra_body.enable_thinking. Plan output as if 65536 is the usual ceiling. Same QCode key, swap this id.
On QCode
QCode /models lists qwen3.7-max. This id is callable. Billing is official list × service rate. Newer Qwen3.8-Max (qwen3.8-max) is not in the catalog — do not order as if 3.8 were here.
Qwen3.7-Max FAQ
Is this model on QCode?
Yes. /models lists qwen3.7-max.
How big is context?
1M tokens.
How do I toggle thinking?
extra_body.enable_thinking.
Official price?
Alibaba’s list is commonly $2.50/$7.50 per 1M. Artificial Analysis cites the official Alibaba API. QCode = list × rate.
Max output?
Often cited as 65536. Confirm on the endpoint you actually hit.
Is 3.8-Max on QCode?
No. Newer, but qwen3.8-max is not listed. Here you call qwen3.7-max.
Sources
qwen.ai/blog?id=qwen3.7 (May 2026); Alibaba API list commonly $2.50/$7.50, cited by Artificial Analysis. Live QCode availability is /models.
Related
Qwen3.8-Max guide
The newer flagship. Not listed on QCode.
GLM-5.2 guide
Another China-lab flagship line to compare.
2026 model radar
Which families are actually on sale.
Not affiliated with Alibaba / Qwen. QCode offers qwen3.7-max, not qwen3.8-max. Prices follow the official table and this site’s dashboard.