Anyscale Free Signup Credits - $100
Source: https://www.anyscale.com/pricing
Description
Create account to comment on specific lines or Sign in
+ 1 Anyscale (the company behind Ray, run by Ray's original creators) offers $100 in free credits to every new account on its Unified Compute Platform — a fully managed, serverless Ray environment for distributed training, batch inference, and online inference. Credits cover any compute on the platform, including GPU instances (T4 → H200) and CPU clusters. Note: the seed entry's $50 amount is outdated — the official signup CTA on anyscale.com is currently "Get Started with $100 Credit" (verified May 2026).
No comments on this line yet.
+ 2
No comments on this line yet.
+ 3 For reference, that $100 launches roughly:
No comments on this line yet.
+ 4 • ~7,400 CPU-hours ($0.0135/hr), or
No comments on this line yet.
+ 5 • ~176 T4 GPU-hours, or
No comments on this line yet.
+ 6 • ~73 L4 GPU-hours, or
No comments on this line yet.
+ 7 • ~73 A10G GPU-hours, or
No comments on this line yet.
+ 8 • ~20 A100 GPU-hours, or
No comments on this line yet.
+ 9 • ~10 H100 GPU-hours, or
No comments on this line yet.
+ 10 • ~9 H200 GPU-hours
No comments on this line yet.
+ 11
No comments on this line yet.
+ 12 The headline starter templates (multimodal AI, LLM training/inference, VLA fine-tuning) are designed to launch for $3–$5 each, so $100 is enough to run dozens of full template walkthroughs end-to-end.
No comments on this line yet.
+ 13
No comments on this line yet.
+
14
No comments on this line yet.
+ 15
No comments on this line yet.
+ 17
No comments on this line yet.
+ 18 1. Go to anyscale.com/signup (or click "Get Started with $100 Credit" on anyscale.com)
No comments on this line yet.
+ 19 2. Sign up with Google, GitHub, or email + password
No comments on this line yet.
+ 20 3. Verify your email address
No comments on this line yet.
+ 21 4. Create or join an organization (Anyscale is org-scoped — a personal account is still wrapped in an org)
No comments on this line yet.
+ 22 5. Land on the Anyscale Console at console.anyscale.com — the $100 credit is applied automatically to the org
No comments on this line yet.
+ 23 6. (Optional) Add a payment method later if you want to keep using the platform after credits run out — no card is required upfront for the free credit per Anyscale's pricing page
No comments on this line yet.
+ 24 7. Pick a starter template from the gallery and click Launch — a Ray cluster spins up on Anyscale-managed cloud and starts billing against your credit
No comments on this line yet.
+ 25
No comments on this line yet.
+ 26 Important:
No comments on this line yet.
+ 27 • No credit card required to claim the $100 — you can sign up, run templates, and burn through the credit without ever entering payment details.
No comments on this line yet.
+ 28 • Anyscale is org-based; the credit attaches to the organization, not the user. Don't create multiple orgs trying to stack credits — one grant per signup.
No comments on this line yet.
+ 29 • The credit is for Anyscale platform fees + underlying compute. Once exhausted, the same workload keeps running and billing at PAYG rates if a card is on file, or fails to launch if not.
No comments on this line yet.
+ 30 • No published expiration window — prior generations of the offer were rumored to be ~30–90 days, but as of May 2026 Anyscale's own pricing/signup pages do not specify an expiry. Treat the credit as best-used in your first few weeks and check the Billing tab in the console for the exact expiry on your account.
No comments on this line yet.
+ 31
No comments on this line yet.
+
32
No comments on this line yet.
+ 33
No comments on this line yet.
+ 35
No comments on this line yet.
+ 36 Anyscale is a fully managed Ray — you write standard Ray / Ray Serve / Ray Train / Ray Data Python code, and the platform handles:
No comments on this line yet.
+ 37
No comments on this line yet.
+ 38 • Cluster lifecycle (autoscaling, spot fallback, head-node HA)
No comments on this line yet.
+ 39 • Job scheduling (one-shot training jobs, recurring batch jobs)
No comments on this line yet.
+ 40 • Anyscale Services (online inference endpoints with auto-scaling + canary deploys)
No comments on this line yet.
+ 41 • Workspaces (cloud IDE / Jupyter on top of a live Ray cluster, no local setup)
No comments on this line yet.
+ 42 • Cost tracking per job / cluster / user, plus pre-built Grafana dashboards
No comments on this line yet.
+ 43
No comments on this line yet.
+ 44 It's the right pick if you already use Ray (or want to) and don't want to babysit Kubernetes or roll your own KubeRay setup.
No comments on this line yet.
+ 45
No comments on this line yet.
+
46
No comments on this line yet.
+ 47
No comments on this line yet.
+ 49
No comments on this line yet.
+ 50 These are the rates the $100 credit burns down against. Cloud-instance underlying cost is included in the figure (for hosted Anyscale Cloud), not billed separately.
No comments on this line yet.
+ 51
No comments on this line yet.
+ 52 InstanceHourly$100 buys you
No comments on this line yet.
+ 53 CPU only$0.0135~7,400 hr
No comments on this line yet.
+ 54 NVIDIA T4$0.5682~176 hr
No comments on this line yet.
+ 55 NVIDIA L4$0.9542~104 hr
No comments on this line yet.
+ 56 NVIDIA A10G$1.3635~73 hr
No comments on this line yet.
+ 57 NVIDIA A100$4.9591~20 hr
No comments on this line yet.
+ 58 NVIDIA H100$9.2880~10.7 hr
No comments on this line yet.
+ 59 NVIDIA H200$10.6812~9.3 hr
No comments on this line yet.
+ 60
No comments on this line yet.
+ 61 Volume discounts and committed contracts (with GPU reservations, billing through AWS/GCP/Azure marketplaces) are available for higher usage — not relevant for the free-credit tier.
No comments on this line yet.
+ 62
No comments on this line yet.
+ 63 Source: anyscale.com/pricing
No comments on this line yet.
+ 64
No comments on this line yet.
+
65
No comments on this line yet.
+ 66
No comments on this line yet.
+ 68
No comments on this line yet.
+ 69 The console exposes a template gallery designed to be one-click launchable against your free credit. Featured starter templates (verified May 2026):
No comments on this line yet.
+ 70
No comments on this line yet.
+ 71 TemplateUse caseApprox. launch cost
No comments on this line yet.
+ 72 Multimodal AI workloadsImage search + classification end-to-end~$3
No comments on this line yet.
+ 73 LLM training and inferenceFine-tune + serve an entity-recognition LLM~$5
No comments on this line yet.
+ 74 Fine-tune PI0.5 VLAVision-Language-Action model fine-tuning~$5
No comments on this line yet.
+ 75
No comments on this line yet.
+ 76 Additional templates in the gallery cover Ray Data ETL, Ray Train distributed training (PyTorch/Lightning/HuggingFace), Ray Tune hyperparameter sweeps, vLLM serving, and Stable Diffusion inference. Browse the full set at console.anyscale.com under Templates.
No comments on this line yet.
+ 77
No comments on this line yet.
+
78
No comments on this line yet.
+ 79
No comments on this line yet.
+ 81
No comments on this line yet.
+ 82 • No always-free tier. Once $100 is gone, you pay PAYG or stop. There is no "free instance always running" like Oracle / GCP have.
No comments on this line yet.
+ 83 • Hosted only. The free credit applies to Anyscale-managed cloud. If you want Anyscale on your own AWS/GCP account (BYOC), that's a separate (paid) deployment model.
No comments on this line yet.
+ 84 • Limited regions. Anyscale Cloud runs in select AWS regions; not every GPU SKU is available everywhere. Check region availability at instance-launch time.
No comments on this line yet.
+ 85 • No model-API freebie. Unlike Together / Fireworks / OpenRouter, Anyscale doesn't give you free LLM API tokens — the credit is for compute to run your own models. (Anyscale's old hosted Endpoints for Llama/Mistral product was retired.)
No comments on this line yet.
+ 86 • GPU quotas. New accounts may hit per-org GPU caps before they exhaust $100. Request a quota bump via the console or Slack if you need to spin up large H100 clusters.
No comments on this line yet.
+ 87
No comments on this line yet.
+
88
No comments on this line yet.
+ 89
No comments on this line yet.
+ 91
No comments on this line yet.
+ 92 For a vibe coder / indie builder — do this in order:
No comments on this line yet.
+ 93
No comments on this line yet.
+ 94 1. Run all three featured templates (~$13) to see how the workspace + jobs flow works.
No comments on this line yet.
+ 95 2. Try a Ray Train fine-tune of a small open LLM (Qwen3-1.7B, Llama-3.2-3B) on an A10G — ~2–3 hours, ~$3–$4.
No comments on this line yet.
+ 96 3. Deploy a Ray Serve endpoint with vLLM for a 7B model on L4 — leave it running for an afternoon, that's ~$8.
No comments on this line yet.
+
97
4. Bigger one-shot job: an H100 fine-tune for 5 hours ($46) on a 7B–13B model.
No comments on this line yet.
+ 98
No comments on this line yet.
+ 99 Do NOT leave a cluster running idle overnight — Anyscale autoscales workers down, but the head node keeps billing. Use Workspaces with idle timeout, or Jobs which terminate on completion.
No comments on this line yet.
+ 100
No comments on this line yet.
+
101
No comments on this line yet.
+ 102
No comments on this line yet.
+ 104
No comments on this line yet.
+ 105 • Ray learning resources: ray.io and docs.ray.io — the platform assumes you can write Ray code; the credit is wasted if you can't.
No comments on this line yet.
+ 106 • Anyscale Academy — free Ray courses at courses.anyscale.com (separate signup from the platform).
No comments on this line yet.
+ 107 • Set a budget alert in the console before launching anything on H100/H200 — a forgotten 8×H100 cluster can melt $100 in under 90 minutes.
No comments on this line yet.
+ 108 • Ray Summit attendees historically get larger top-up credits — worth checking if you're going to a future Ray event.
No comments on this line yet.
+ 109 • Startup deals: Anyscale has appeared in partner perk packs (e.g., Antler Disrupt's $650K perk bundle) — if you're VC-backed, ask your platform team for a larger Anyscale grant before burning the public $100.
No comments on this line yet.
+ 110 • For pure inference (don't want to run Ray yourself), Together AI / Fireworks / OpenRouter / DeepInfra are cheaper for token-based usage. Anyscale is the right pick when you specifically want distributed Ray jobs on managed infra.
No comments on this line yet.
+ 111
No comments on this line yet.
+
112
No comments on this line yet.
+ 113
No comments on this line yet.
+ 114 Sources:
No comments on this line yet.
+ 115 • Anyscale Pricing
No comments on this line yet.
+ 116 • Anyscale Signup
No comments on this line yet.
+ 117 • Anyscale Product Overview
No comments on this line yet.
+ 118 • Anyscale Console
No comments on this line yet.
+ 119 • Anyscale Docs – Get Started
No comments on this line yet.
+ 120 • Anyscale Docs – Budgets
No comments on this line yet.
+ 121 • Ray.io
No comments on this line yet.
+ 122 • Anyscale Academy / Courses
No comments on this line yet.