Blog
Authors
Runware Model Router: set the rules, and let the API choose the model
The Runware Model Router picks a model for each request from Runware's official catalog, ranked by quality, speed, and cost, within the requirements and price ceiling you set.
Engineering leaders from NVIDIA, OpenAI, and NetApp join Runware
Senior hires from NVIDIA, OpenAI, and NetApp take over Runware's infrastructure and security, with further senior additions from Amazon Web Services and Juniper Networks.
What's new on Runware: October 2026
The Runware AI factory, Runware Serverless and its first updates, Kling VIDEO 4.0, FLUX 3 Image and Klein, Nano Banana 2.1, and 18 more new models, plus October's model updates, platform updates, and deprecations.
Runware unveils AI factory to build modular AI data centers just like cars
Runware's new European factory will build complete, AI-native data centers on an assembly line: 1MW and 2MW Sonic Pods, up to 30MW a day, 10GW a year, from Q1 2027.
From FLUX.1 [dev] to FLUX 3 Image: four generations of Black Forest Labs models
FLUX 3 Image, Black Forest Labs' first image model built on the FLUX 3 multimodal backbone, is live on Runware. A look back at four generations of BFL image models.
Runware Serverless: run your own AI from $0.63 per GPU-hour
Runware Serverless is live, with prices from $0.63 per GPU-hour. Run your own code, models and containers on Runware's GPU infrastructure, built on our own Sonic Pods, while we manage the platform.
Zero Data Retention on Runware, now available
Runware's Zero Data Retention (ZDR) lets enterprise teams run AI image, video, and language model requests without retaining prompts or generated outputs.
Content moderation for AI pipelines: gpt-oss-safeguard-20b and Shieldstral 1.0 3B
OpenAI's gpt-oss-safeguard-20b and Mistral AI's Shieldstral 1.0 3B are live on the Runware API: safety moderation for prompts, text, and generated images.
Jev, Laya, and the emerging role of decision models
TypeSafe's Jev and Convai's Laya package a familiar task, making a bounded decision from messy input, into a dedicated alternative to an LLM. Here's how they differ and where each might fit into your workflows.
