Runware and LLM Gateway launch 30% off open-model inference for 30 days

smakosh · reddit · 2026-07-28

LLM Gateway says it has partnered with Runware to serve several open-weight models at 30% off for the first 30 days.

Models listed include gpt-oss-120b, Gemma 4 31B IT, DeepSeek V4 Flash/Pro, Kimi K2.6, and GLM 5.2.
Example discounted prices per million tokens are provided, such as:
gpt-oss-120b: about $0.022 input / $0.098 output
DeepSeek V4 Flash: about $0.053 / $0.107
GLM 5.2: about $0.56 / $1.79
The discount is applied automatically at billing time when the router picks Runware or the user pins a Runware model.
If a request falls back to another provider, standard pricing applies.
Runware says it does not train on API traffic, but it does log prompts.
The promo ends August 26.

More from Venture