Qwen3.8-27B API, 140 tok/s on one GPU

infrastructure / Show HN

SCORE

A hosted API providing access to the Qwen3.8-27B large language model. The service delivers inference at 140 tokens per second on a single GPU. Users can access the API after logging in through the provided endpoint.

VISIT PRODUCT →

Sources (1)

Show HN

https://news.ycombinator.com/item?id=49320595

1 PTS

Score Breakdown

Traction

raw 1.00 · weight 35%

3.5pts

Novelty

0 days old · weight 20%

20.0pts

Source diversity

1 source · weight 10%

3.3pts

AI quality

raw 22.00 · weight 35%

7.7pts

Minimal differentiation—just another hosted LLM API with a speed claim, zero traction, and no clear unique value beyond commodity inference.

Final Score35/100

Similar Products

Marzban-Node

INFRASTRUCTURE · GITHUB

SCORE

Qwen3.8-27B-GGUF

DEVELOPER-TOOLS · HUGGING FACE

SCORE

Qwen3.8-Flash-Next

DEVELOPER-TOOLS · HUGGING FACE

SCORE

K3Flight

INFRASTRUCTURE · GITHUB

SCORE

Marzban-Panel

INFRASTRUCTURE · GITHUB

SCORE

3x-ui-multi

INFRASTRUCTURE · GITHUB

SCORE