Back to changelog

Runware Launch: 30% Off Open-Source Models

We partnered with Runware.ai to serve six open-source flagships — gpt-oss-120b, Gemma 4, DeepSeek V4 Pro and Flash, Kimi K2.6, and GLM 5.2 — at 30% off for 30 days. The discount applies automatically to everything routed through Runware until August 26.

LLM Gateway and Runware logos beside a large 30% and a one-month calendar, announcing the launch discount

Open-source flagships now carry serious production workloads, which makes the per-token price of serving them the number that matters. We partnered with Runware.ai to bring their fast, OpenAI-compatible inference for open models to the gateway — and to launch it, every Runware model is 30% off for 30 days.

Six Models, 30% Off Until August 26

Model Context Input $/M Output $/M
gpt-oss-120b 131K $0.032 $0.14
Gemma 4 31B IT 262K $0.102 $0.297
DeepSeek V4 Pro 1M $0.961 $1.922
DeepSeek V4 Flash 1M $0.076 $0.153
Kimi K2.6 262K $0.60 $3.05
GLM 5.2 1M $0.80 $2.55

Those are the list prices — the 30% comes off on top at billing time. gpt-oss-120b, already one of the cheapest ways to run a 120B model, lands at about $0.022/M input during the promo.

How the Discount Works

Nothing to configure: the discount applies automatically to any usage billed through Runware, whether the router picked it for you or you pinned it yourself:

1curl https://api.llmgateway.io/v1/chat/completions \2  -H "Authorization: Bearer $LLM_GATEWAY_API_KEY" \3  -d '{4    "model": "runware/glm-5.2",5    "messages": [{ "role": "user", "content": "Hello" }]6  }'
  • Discounted pricing shows directly on the model pages while the promo runs.
  • Automatic fallback keeps working as usual — only requests actually served by Runware get the discount.
  • The promo ends August 26, 2026; list prices apply after that.

Runware serves these models through the same unified API as every other provider, with streaming support and no training on your API traffic.

Runware on LLM Gateway → | Browse all models →