The home of open models, datasets and Spaces.
Anyscale
Ray-based platform for AI training and inference.
SharkScore
#12 in Model APIs & inference
Why it scores 70.5
Anyscale has a SharkScore of 70.5/100, ranked #12 of 262 in Model APIs & inference (updated 2026-11-19).
Each part compares Anyscale with other listings in Model APIs & inference. Evidence covers 87% of the score's weight.
- Adoption
- 53 · 30%
- User reviews
- no data · 25%
- Maintenance
- 100 · 20%
- Trust & transparency
- 80 · 15%
- Momentum
- no data · 10%
Paying never changes this score. How ranking works · method v2026-11-18.2
RankShark review of Anyscale
Automated evaluation by RankShark · 18 Nov 2026 · method v1.0 · How we review
We were not paid for this review.
As of 2026-11-18, Anyscale is a Ray-based platform for distributed AI workloads, with pay-as-you-go compute and BYOC deployment options. [E1, E2]
Anyscale provides a platform for building, running, and scaling data-intensive AI workloads using Ray. Its pages describe distributed training, data curation, batch inference, and deployment on Anyscale-hosted infrastructure or a customer's cloud. [E1, E2]
Best for
- Teams running distributed model training, multimodal data curation, or batch inference workloads. [E1] · home page
- Teams that want to deploy Ray workloads in their own cloud or on-premises. [E2] · pricing page
Not for
- Teams that need published fixed monthly plan prices; the pricing page describes usage-based billing and hourly compute rates. [E2] · pricing page
Key features
- Distributed AI workloads. The vendor says the platform supports distributed training, multimodal data curation, batch inference, and post-training workloads. [E1] · home page
- Ray and Python APIs. The vendor says Ray lets users execute Python functions and classes on a distributed cluster, and that it can scale libraries including PyTorch, vLLM, SGLang, and XGBoost. [E1] · home page
- Hosted and BYOC deployment. The vendor says Hosted uses Anyscale-managed infrastructure, while BYOC can run in a customer's cloud or on-premises. [E2] · pricing page
- Multi-cloud execution. The vendor says the same code can run across AWS, GCP, Azure, Nebius, or CoreWeave. [E1] · home page
- Access controls and audit logs. The vendor says its access controls and authentication include SSO, SAML, SCIM, and audit logs. [E1] · home page
Pros
- The pages describe multiple distributed AI workload types, including training, data curation, batch inference, and post-training. [E1] · home page
- The pricing page offers BYOC deployment in a customer's cloud or on-premises, with VM or Kubernetes deployment listed. [E2] · pricing page
- The vendor lists multiple cloud providers for multi-cloud execution. [E1] · home page
Cons
- Hosted deployment is listed as having limited regions and business-hours support, with 5 case submissions. [E2] · pricing page
- The published compute rates use the label “AC” and are hourly; the evidence does not state a USD monthly plan price. [E2] · pricing page
- Whether customer data is used for model training is not stated on the pages we checked.
Pricing
| Plan | Price | Limits |
|---|---|---|
| Pay as you go | Usage-based; listed Hosted compute rates include CPU Only AC 0.0135 /hr, NVIDIA T4 AC 0.5682 /hr, NVIDIA L4 AC 0.9542 /hr, NVIDIA A10G AC 1.3635 /hr, and NVIDIA A100 AC 4.9591 /hr. NVIDIA H/B/GB GPU families: Contact Us. [E2] | The vendor says customers pay for compute used on demand, with no monthly fixed fees. Hosted support is business hours only with 5 case submissions; BYOC offers enterprise SLAs with 24x7 coverage and unlimited case submissions. [E2] |
| Committed contracts | Talk to Sales. [E2] | The vendor describes volume discounts and benefits that scale with usage. [E2] |
Checked 18 Nov 2026 · source
- Free plan
- The pricing page says to get started with $100 credit; the homepage says up to $500 credit. [E1, E2] · home page, pricing page
- Free trial
- The homepage advertises a free trial for KubeRay Connect. [E1] · home page
- Trains on customer data
- Not stated (checked 18 Nov 2026)
- Data retention
- Not stated (checked 18 Nov 2026)
- Data regions
- Not stated (checked 18 Nov 2026)
- Data processing agreement
- Not stated (checked 18 Nov 2026)
- AI models used
- The vendor's example code names Llama 3.1 70B and BGE-large-en-v1.5; the pages do not state these are a hosted model catalog. [E1] · home page
- API
- The terms define Platform Services as including associated APIs. [E5] · terms
- SDKs
- The terms define Platform Services as including associated software development kits. [E5] · terms
- Webhooks
- Not stated (checked 18 Nov 2026)
- Platforms
- The pricing page lists deployment on VMs for Hosted, and on VMs or Kubernetes for BYOC; BYOC can run in any cloud or region and on-premises. [E2] · pricing page
What we measured
- GitHub stars · now
- 44,315
- Latest release · now
- ray-2.60.0 on 2026-11-04
- Package downloads · 30 days
- 15,526,535
- Tranco traffic rank · list of the day
- #165,499
How we tested · sources
- Hourly uptime and response-time checks of the product's URL.
- Repository, release and download data from GitHub, npm, PyPI and Hugging Face where they exist.
- We did not sign up, pay, or use the product as a customer.
- home page: https://www.anyscale.com/ · fetched 18 Nov 2026
- pricing page: https://www.anyscale.com/pricing · fetched 18 Nov 2026
- docs: https://docs.anyscale.com/?utm_source=anyscale&utm_medium=website&utm_campaign=footer · fetched 18 Nov 2026 · blocked our checker
- privacy page: https://www.anyscale.com/privacy-policy · fetched 18 Nov 2026
- terms: https://www.anyscale.com/terms · fetched 18 Nov 2026
Something wrong? Report an error with a public link that shows it; corrections re-run the review.
Top alternatives to Anyscale
See all- 79.3ToolFreemium · from $9/mo
- 77.5
API for building with Claude models.
ToolPaid - 76.1
Proxy and SDK to call 100+ LLM APIs in the OpenAI format.
ToolOpen source - 75.7
Google Cloud's platform to build with Gemini and other models.
ToolPaid