Ray
ray.io
Distributed computing framework for scaling AI workloads.
Quick Information
- Category:
- MLOps & Infrastructure
- Pricing:
- Freemium
- Platforms:
- Python, Kubernetes, Cloud
- Free Plan:
- Yes
- API Available:
- Yes
- Last Verified:
- October 2026
What Is Ray?
Ray is a distributed computing framework for scaling AI workloads, and Anyscale is the company that sells a hosted platform on top of it. Ray itself is free and open source under Apache 2.0, and you can self-host it on your own Kubernetes cluster via KubeRay. Anyscale's hosted version is priced separately, though I found no official pricing page, so the figures are secondhand. It is usage-based, billed per instance-hour in Anyscale Credits, and one aggregator lists rates as of August 24, 2026 of 0.0135 for CPU-only, 0.5682 for a T4, 0.9542 for an L4, 1.3635 for an A10G and 4.9591 for an A100, with H100 and B200 pricing being a sales conversation. A $100 starter credit is mentioned. In bring-your-own-cloud setups you pay your cloud provider for compute and Anyscale bills platform usage on top, invoiced directly or through AWS, Azure or GCP marketplaces, and committed contracts use prepaid credits with discounts and enterprise SLAs.
One aggregator lists the entry price as $0.0135 a month rather than per hour, which looks like a mislabel. A GPU-cloud vendor's blog claims the Anyscale markup adds 50% to 100% on top of bare-metal GPU rates, but that source is a competitor, so treat it as unverified. Several sites report that Nscale agreed to acquire Anyscale for about $1.65 billion, announced July 30, 2026, with one saying no pricing change accompanied the announcement. Check anyscale.com/pricing before relying on any number.
Ray suits ML teams that need to scale training, tuning, data processing and serving beyond a single machine. Compare KServe, Lightning AI and vLLM. Browse the MLOps & Infrastructure tools directory for more.
What Can You Use Ray For?
Distributed training and tuning
Scale-out.
Model serving and data processing
Ray Serve and Data.
Ray Features
- Open-source Apache 2.0 — Framework.
- Anyscale BYOC billing — Platform fee on top of cloud.
- Nscale acquisition — Announced July 30, 2026.
Who Is Ray For?
- ML engineers — Scale beyond one machine.
- AI platform teams — Unified compute layer.
Ray Pros & Limitations
What We Like
- Free and widely adopted
- Flexible self-host or managed
- Broad library ecosystem
Limitations
- Anyscale prices not from an official page
- Reported markup unverified
- Ownership change pending
Ray Pricing
| Plan | Price | Main Features |
|---|---|---|
| Ray (open source) | Free (Apache 2.0) | Self-host, including KubeRay |
| Anyscale (usage-based) | Per instance-hour in credits: CPU 0.0135, T4 0.5682, L4 0.9542, A10G 1.3635, A100 4.9591 | Aggregator, August 2026; H100/B200 via sales |
| Committed contracts | Prepaid credits with discounts | Enterprise SLAs, 24x7 support |
Pricing was last verified October 2026. Check the official website for the latest pricing.
AI Listing Jungle Ranking
- Global Rank:
- #265
- Overall Score:
- 72.6/100
- Monthly Users:
- –
- Rank Change:
- ▲ 1296
This ranking reflects AI Listing Jungle's own independent scoring methodology and is not an endorsement by Ray or its company. How Our Rankings Work
Frequently Asked Questions
Is Ray free?
Yes, Apache 2.0.
How is Anyscale priced?
Usage-based credits per instance-hour.
Is Anyscale being acquired?
Nscale agreed to buy it for about $1.65B, announced July 30, 2026.
Is there a starter credit?
$100, per one aggregator.
Our Take
Ray is the default for distributed Python AI.
Compare the Anyscale markup to self-hosting.
Ray Alternatives
Ray Comparisons
Ready to try Ray?
Visit the official website to get started.
Visit Official WebsiteAI Listing Jungle is an independent AI tools directory. Product names, logos, and trademarks belong to their respective owners.