GreenNode’s latest update is now available on AI Platform. No, we haven’t launched a rocket to Mars (yet, we will leave that to Musk for now) — but with these new updates, who knows, maybe we’ll help you get there someday.
In this latest GreenNode release, we have made extra efforts to support more models, enhance security, and automate processes, so you can save a coffee break every day (your coffee budget, however, is still up to you).
Let’s dive in!
What’s New in GreenNode AI Platform 25.08?
The 25.08 release brings a fresh wave of upgrades designed to make your AI workloads more secure, more automated, and easier to manage. From fine-grained IP controls to hands-off scheduling, faster notebook/model workflows, and a direct feedback loop, this release is built to help developers and enterprises scale smarter without extra overhead.
Highlights at a glance:
- Serverless Model Management: Expanded support for GPT-OSS (120B, 20B) and Qwen3 models
- Strengthen security with IP Whitelisting for both Notebooks and Model Endpoints.
- AI Workload Automation: Automate scheduling for Notebooks and Model Endpoints, adding the edit/clone capabilities for both Notebooks and Model Registry.
- User experience enhancements with a new in-product feedback channel.
Serverless AI Model Enhancements
GPT-OSS 120B & 20B
When it comes to large language models, the conversation is no longer limited to closed-source giants. GPT-OSS (the open-source successor to OpenAI’s earlier GPT-2 work) is quickly becoming the model family of choice for enterprises seeking more flexibility, scalability, and cost efficiency than traditional OpenAI models.
With this release, GreenNode AI Serverless now supports GPT-OSS 120B and 20B, available for:
- Playground experimentation—test ideas, prompts, and workflows instantly.
- Full deployment—production-grade serving on GreenNode’s GPU-powered infrastructure.
You can now experience this new stuff directly on GreenNode Serverless AI Model
Qwen3 models
Claimed to outperform even Claude Sonnet 4 on select benchmarks, Qwen3 models combine reasoning power, coding skills, and multilingual adaptability in ways that make them highly competitive against commercial alternatives.
GreenNode AI Serverless now supports multiple Qwen3 variants, giving you expanded options for both playground testing and full-scale deployment:
- Qwen3-235B-Thinking-2507: A state-of-the-art reasoning model, excellent for logic, math, science, and code. Perfect for chain-of-thought workflows and structured problem-solving.
- Qwen3-Coder-30B-FP8-Instruct: Optimized for coding tasks with instruction-following capabilities, efficient yet powerful for developers experimenting at scale.
- Qwen3-Coder-480B-Instruct: A massive instruction-tuned model for enterprise-grade deployment. Ideal for large-context applications and advanced AI assistants.
Swagger API availability
In this release, the serverless AI service also introduces a fully documented Swagger API for GreenNode Model-as-a-Service (MaaS), giving developers a unified interface to explore, test, and integrate models more efficiently.
At its core, GreenNode MaaS provides an OpenAI-compatible API — meaning you can plug it into existing workflows with minimal effort, whether you’re coding in Python, JavaScript, or just testing directly from your terminal with cURL. Learn more: GreenNode Model-as-a-service Swagger API
- Note: You’ll need an API key to start. Don’t have one yet? Generate yours directly in the GreenNode AI Platform Portal.
Security enhancements with IP Whitelist for Notebooks and Model Endpoints
The latest release of the GreenNode AI Platform introduces an IP Whitelist for both Notebook Instances and Model Endpoints, providing tighter control over who (and what) can access your resources.
With this feature, you can now view, add, edit, and remove the IP addresses of your choice:
- See the full list of trusted client IP addresses authorized to connect.
- Add, remove, or edit entries in real time with no need to restart or redeploy notebooks and endpoints.
Whether you’re locking down your Notebook environment or protecting your Model Endpoints, GreenNode lets you manage IP whitelists directly from the console using CIDR notation or via API for seamless automation.
- Note: Notebook and Model Endpoint whitelists are managed independently on the GreenNode AI Platform portal. Learn more:
AI Workload Automation with Auto-config and Scheduling
Scheduling for Notebooks and Model Endpoints
AI workloads don’t need to run 24/7. But until now, too many of them did. A single notebook left open overnight or an endpoint serving no traffic can quietly rack up unnecessary costs. With the 25.08 release, GreenNode makes sure your infrastructure works when you need it and rests when you don’t.
Model endpoints
For model endpoints, you can now set precise schedules to automatically bring services online during business hours and power them down when traffic drops.
A financial services team, for example, might keep an endpoint live from 8 a.m. to 6 p.m. to handle client requests, then shut it off overnight to eliminate idle GPU time.
Schedules can be configured to recur daily, weekly, or monthly, all based on UTC. Learn more: Automated Scheduling for Model Endpoint.
Notebook instances
Notebook instances follow the same logic. Instead of watching GPU resources burn long after an experiment ends, teams can set schedules to automatically shut notebooks down outside of working hours, then spin them back up the next morning. The result is not only cost savings, but also a cleaner, more predictable usage pattern across the entire team. Learn more: Automated Scheduling for Notebook Instance.
To ensure the security of the workload, only users with the appropriate permissions — developer or administrator roles —can create, modify, or remove schedules for notebooks and model endpoints. Every action is automatically logged with full detail, ensuring traceability and accountability at all times.
Scheduling itself can be carried out either through the web interface or programmatically via API. Technically, this feature removes the burden of manual lifecycle management: No more “Did anyone remember to shut that down?” messages in Slack at midnight. From a business perspective, the impact is significant: lower cloud bills, predictable resource allocation, and stronger governance through complete auditability.
Notebook and Model Registry Edit/Clone
When setting up notebooks or managing model registries, teams often spend a significant amount of time reconfiguring frameworks, adjusting environments, and uploading datasets. These repetitive tasks can consume both developer effort and GPU resources. From this update, GreenNode AI Platform removes much of that friction by enabling users to edit and clone Notebook Instances and Model Registry entries with ease.
Notebook instances
GreenNode AI Platform allows users to duplicate an existing configuration in seconds, make adjustments, and launch a new experiment without starting from scratch. A proven setup can be replicated and tweaked for a different dataset or hypothesis, dramatically cutting the time it takes to move from idea to execution.
The result is faster prototyping, reproducible workflows, and a cleaner audit trail for how experiments evolve over time. Learn more: Edit/Clone Notebook instances.
Model Registry
When configuring an existing registered model, the engineering team can now update it with new weights or run comparative benchmarks without disrupting production deployments. This allows the team to test variations side by side, manage versioning, and streamline the entire model lifecycle.
By reducing friction in experimentation, teams are free to focus on iterating and innovating rather than wrangling configurations. Learn more: Edit/Clone Model registry.
You can now directly feedback for our product via the GreenNode AI platform portal
With an aim to provide you with instant support, GreenNode AI Platform's direct feedback is now available so you can share ideas, requests, or issues right where you work.
This shortens the loop between your experience and our roadmap. For us, it means insights in real time. For you, it means faster improvements and features shaped by real needs. The feedback you leave today could become tomorrow’s upgrade. You can always contact us for instant feedback and support here: Contact us.
Conclusion & Call to Action
The 25.08 release isn’t just about new features — it’s about moving closer to the vision of AI infrastructure that is secure by default, efficient at scale, and open to innovation. GreenNode AI Platform is built for developers who want faster iteration and for businesses that demand predictable, cost-effective operations.
As we continue to expand support for leading open models and streamline the way teams manage their workloads, our goal remains the same: to give you the freedom to build and deploy AI with confidence.
Now is the time to see what this means for your own projects. Whether you’re experimenting in the playground or running AI in production, GreenNode is ready to power the next step of your journey. Try it today on the GreenNode AI Platform.