Cost across cloud + local
Paste anything — a prompt, source file, transcript, email draft. We count tokens, then show you the cost across 11 cloud providers and your local rig, side by side. Same data for the same input. Decide where each workload belongs.
11 providers across 3 tiers. Prices verified May 2026. Token counts are content-aware approximations (±15%) — provider price differences dwarf tokenizer accuracy, so this is enough to make a decision. URL updates as you change — share by copy.
Calculator
Start with the sample text or paste your own. Pick a local rig + model from your catalog and the cloud providers you want to compare against. Everything is reactive.
Price figures are our planning estimates (launch MSRP, or a used-market estimate for older cards), not live prices. Check the current price before you buy.
| Provider | Model | $/M in | $/M out | Per run | per month | vs cheapest |
|---|---|---|---|---|---|---|
| DeepSeek | DeepSeek V3 (API) Mid-tier | $0.27 | $1.10 | $0.00012 | $0.177 | cheapest |
| Together AI | Llama 3.3 70B (Together) Open-weight | $0.88 | $0.88 | $0.00019 | $0.284 | 1.6× |
| OpenAI | GPT-5 Frontier | $5.00 | $20.00 | $0.00216 | $3.23 | 18.3× |
Rent the GPU, not the API
Third option: rent a GPU by the hour and run open-weight models yourself. You pay only while the instance is up, and you can test the exact model and quant before you buy hardware. Privacy depends on the host you pick, so read its terms before you send anything sensitive.
RunPod
billed per GPU-hourHourly GPU pods, in community-cloud and secure-cloud tiers.
The two tiers run on different hosts and terms. Check which tier a pod is on before a long job.
Rent on RunPodVast.ai
billed per GPU-hourMarketplace of host-listed GPUs; price and reliability vary by host.
Host quality varies. Filter for verified hosts and check the host's reliability before a long run.
Rent on Vast.aiReferral links: RunLocalAI may earn a commission or account credits if you sign up and rent, at no extra cost to you. Hourly rates and which GPUs are listed change all the time, so check the provider's page before you start. Editorial opinions are independent of the referral relationship; see how we make money.
ClaudeStore: an unofficial reseller of Claude API credits
If the calculator above says a hosted model is the right call for your workload, you can buy API access from Anthropic directly. ClaudeStore is an unofficial third-party reseller of Claude API credits. It is not Anthropic. Its terms and prices are set by ClaudeStore and can change, so read them before you pay. Requests you send through a reseller pass through its servers; do not send anything you would not share with a third party.
Referral disclosure: RunLocalAI receives a referral commission on sign-ups through the ClaudeStore link, at no extra cost to you. The calculator above, not the referral, is what tells you whether a hosted model is the right answer for your workload. How we make money.
Where to go from here
Once you know it's worth going local: pick the whole rig + runtime + model + install script in one wizard.
The local cost above assumes Q4_K_M. Drill into which quant fits your specific hardware × model × context.
The break-even number above is volume-driven. Switch view to tune utilization / electricity / amortization assumptions.
Cost is one side of the trade — speed is the other. Watch tokens stream at the rate your rig would produce them, side-by-side with the cloud provider.
Step-by-step guide to swap Claude Sonnet for a local model in your Claude Code workflow. Real commands.