Qwen 3.8 27B
loaded
top
Hosted on our own DGX Spark in Austria, available through an OpenAI-compatible API.
80
Logic Index · budget 2,048
38 t/s
tokens per second here
€ 0.50 / € 2
euros per M tokens, input / output
10M free tokens
Try Qwen 3.8 27B for free
Every new account starts with 10M free tokens, usable for this and every other model. No card, no subscription.
Fact sheet
- Model name in the API
- qwen3.8-27b
- Size
- 27B dense (NVFP4)
- Context
- 256K
- Input
- Text, Image
- Capabilities
- ToolsImageReasoning
- License
- Apache 2.0
- Tier
- Standard
- Hosting
- always loaded
Call
OpenAI-compatible. The alias works as model name too.
{
"model": "top",
"thinking_token_budget": 2048,
"messages": [ … ]
}
This thinking budget is the one the Logic Index was reached with. Without the field, our default for short answers applies.
Logic Index
All models in the Logic Index →By test area
Reasoning chains
93
Logic vs. world knowledge
100
Logic grid puzzles
46
Consistency: same answer in every run for 97% of tasks.
With a budget of 2,048 the thinking is not enough for the larger puzzles. With a budget of 8,192 the same model solves far more, see its own entry.
Index by thinking budget
Speed over time
Operations of the Sparks →Tokens per second from our nightly measurement. Latest: 38 t/s.