Hyperstack Weekly Rundown 47

Welcome to Hyperstack Monthly Update

We've been busy this month. Some updates you'll notice immediately, while others are quietly making your experience faster and more reliable. Either way, they're all worth knowing about. Here's what we've been working on.

Check out the latest tutorials and blogs on Hyperstack:

How to Use GLM-5.2:

On Hyperstack AI Studio

Hyperstack AI Studio now supports GLM-5.2, the latest open-weight reasoning and coding model from Z.ai (formerly Zhipu AI) with a one million token context window. You can access it through the existing serverless API or the point-and-click Playground—no GPU provisioning, downloads, or server management required. The Playground is ideal for quick trials, while the API is designed for integrating GLM-5.2 into products, coding assistants, and automated pipelines. Our latest tutorial covers both, with a primary focus on the API using live endpoint examples and real responses.

Read the full tutorial →

Generate and Edit Stunning AI Images:

On Hyperstack AI Studio

AI Studio now generates images. Alongside its language models, the platform serves a growing line-up of text-to-image and image-to-image models, including FLUX.1, FLUX.2, Qwen-Image, Stable Diffusion 3.5 and more, behind a single serverless API and a point-and-click Playground. There are two ways to use it. The Playground is the fastest way to experiment by hand, and the API is how you put image generation into a product or an automated pipeline. Our latest tutorial covers both.

Read the full tutorial→

GLM-5.2 vs Claude Fable 5:

An Open-Weight Frontier Model Meets a Closed One

The sudden suspension of Claude Fable 5 highlighted a reality many AI teams overlook: access to a powerful model can disappear overnight. At the same time, GLM-5.2 has emerged as one of the strongest open-weight alternatives, combining competitive performance with the freedom to self-host. In this blog, we compare their capabilities, benchmarks, pricing, hardware requirements and deployment trade-offs.

Read the full comparison→

Help Shape the Future of Hyperstack

Great products are built with the people who use them. If there’s something you would like to see on Hyperstack, whether it is a new feature, workflow improvement or integration that would make your work easier, we would love to hear about it.

Your feedback helps us prioritise what matters most and build a platform that works better for the community.

Share Feature Request →

Welcome to Hyperstack Weekly Rundown

It's that time again. Weekly Rundown is here and there's a lot to catch up on. A brand-new Image Playground just landed in AI Studio, new tutorials dropped on the blog and a round of platform improvements shipped quietly in the background. Whether you're generating images, optimising inference costs or running regulated research workloads, this week had something for you. 

Take a few minutes to catch up on everything.

Check out the latest tutorials and blogs on Hyperstack:

How to Optimse KV Cache:

Guide to Faster and Cheaper LLM Inference

If you run LLMs at scale, optimising the KV cache is one of the most impactful ways to reduce cost per token, extend context length, and increase batch size. Our latest guide explains what the KV cache is, why it becomes a bottleneck and practical optimisation techniques, from a one-line quantisation flag you can use today to the architectural changes behind frontier models with reproducible numbers and runnable code.

Read the full tutorial →

Running a Multi-Program AI Research Environment:

On Secure Private Cloud

Our case study follows a regulated AI research institute running five concurrent genomics, molecular design and bioinformatics workloads. After public cloud and on-premises setups failed to meet performance and compliance needs, the team moved to a dedicated Secure Private Cloud on Hyperstack. The blog examines the migration decision and the impact on reproducibility, compliance, and research output.

Read the full case study→

Deploy Gemma Diffusion on Hyperstack:

A Step-by-Step Guide

DiffusionGemma is an open-weights, diffusion-based language model from Google DeepMind built on the 26B-A4B Gemma 4 Mixture-of-Experts architecture. Our latest tutorial shows how to deploy Google DeepMind's DiffusionGemma on Hyperstack using vLLM and a single NVIDIA H100-80G-PCIe, exposing an OpenAI-compatible API with the Entropy-Bounded diffusion sampler. Built on the Gemma 4 MoE architecture, DiffusionGemma generates text by denoising 256-token blocks in parallel, delivering over 1,000 tokens per second on a single NVIDIA H100 while supporting text, image, and video inputs.

Read the full tutorial→

Help Shape the Future of Hyperstack

Great products are built with the people who use them. If there’s something you would like to see on Hyperstack, whether it is a new feature, workflow improvement or integration that would make your work easier, we would love to hear about it.

Your feedback helps us prioritise what matters most and build a platform that works better for the community.

Share Feature Request →

Welcome to Hyperstack Weekly Rundown

Your weekly digest of the latest updates, tutorials and improvements from Hyperstack. This week features a Kubernetes version update, deployment guides for DeepSeek V4 and Kimi K2.6, and a chance to shape what we build next.

Take a few minutes to catch up on everything.

Check out the latest tutorials on new AI models on Hyperstack:

Deploy DeepSeek-V4 on Hyperstack:

A Step-by-Step Guide

DeepSeek-V4 is DeepSeek AI’s latest open-weight LLM family, optimised for efficiency and long-context reasoning. The lineup spans 284B–1.6T parameters with only 13B active per step, a native 1M-token context window, and FP4+FP8 mixed precision for efficient single-node deployment. This tutorial shows how to deploy DeepSeek-V4-Flash on Hyperstack using vLLM and Docker with an OpenAI-compatible API.

Read the full tutorial →

Deploy DeepSeek-V4 Pro on Hyperstack:

A Step-by-Step Guide

DeepSeek-V4 Pro is a 1.6 trillion parameter sparse MoE with 49 billion active per token, a 1M token context window, and an efficient hybrid attention stack. It delivers strong coding benchmark performance but exceeds single-node limits, making it ideal for multi-node deployment on 8x H100. This tutorial covers running it on Hyperstack using vLLM and Docker.

Read the full tutorial→

Deploy Kimi K2.6 on Hyperstack:

A Step-by-Step Guide

Kimi K2.6 is an open-weight, native multimodal agentic model from Moonshot AI, built for advanced coding and autonomous agent workflows. Using a sparse Mixture-of-Experts architecture with 1T parameters and only 32B active per step, it matches or exceeds leading closed models on coding and agentic benchmarks while remaining practical to self-host. This tutorial covers deploying Kimi K2.6 on Hyperstack using a GPU VM, vLLM and an OpenAI-compatible API.

Read the full tutorial→

Help Shape the Future of Hyperstack

Great products are built with the people who use them. If there’s something you would like to see on Hyperstack whether it is a new feature, workflow improvement or integration that would make your work easier, we would love to hear about it.

Your feedback helps us prioritise what matters most and build a platform that works better for the community.

Share Feature Request →