Verify your email. Claim DREAM1 in Usage & billing. Once per account and verified email. No payment needed.
25% off Anthropic & GPUsLaunch discount applied automatically to eligible usage.
AI models,
tools & GPUs.
One account.
Access 750+ AI models from 90+ providers, 2,700+ tools across 90+ services, and 22 GPU configurations with one API key and prepaid balance. We manage provider access. Pay for what you use.
No payment required to create an account. No separate vendor accounts for supported services.
Missing a tool? Email us.
- Try a model or tool in Playground first.
- Connect your agent through MCP, the SDK or CLI.
- Use your Dreamtune key for supported services.
We’re developers too. We’ve topped up one API, subscribed to another, and lost track of where we still had money left.
Give your agent access to models, web search, voice and GPU compute with one API key. Fund one balance and track every request and charge in your workspace.
One place to fund what you build.
Catalog checked . GPU capacity varies by region.
Models for the thinking
Choose a supported model for writing, coding or reasoning.
Tools for the doing
Search the web, read pages, generate media and call APIs.
GPUs for the heavy work
Rent available compute and pay for the runtime you consume.
One place to pay
Fund your Dreamtune balance. See requests and charges together.
Usage at a glance.
Model tokens, tool calls and GPU time, together in one workspace.
Daily token usage
Oct 15, 2026 – Oct 21, 2026
Resource breakdown
Anthropic modelsModel284M tokens
HiggsfieldImage & video24,800 generations
ElevenLabsAudio7,200 minutesFirecrawlTool172,000 calls
NVIDIA A100Compute4,800 GPU hours
NVIDIA H100Compute2,400 GPU hours
Featured on Dreamtune.
Models, web search, voice, image generation and GPUs. Pick a service and start building.
Abliterated (uncensored)
GLM, Qwen, DeepSeek
Flux Search
Search, Lineage, Citations
Higgsfield
Image, Video, Creative tools
ElevenLabs
Audio, Speech, Music
OpenAI
Text, Vision, Code
Anthropic
Text, Vision, Reasoning
Google Gemini
Text, Vision, Multimodal
DeepSeek
Text, Vision, Code
Firecrawl
Search, Scraping, Extraction
Exa
Search, Research, Web content
Browserbase
Browser, Automation, Sessions
Apify
Scraping, Data, Automation
DataForSEO
SEO, Search data, Keywords
Lambda
GPU, Inference, Training
Qwen
Text, Vision, Code
Mistral
Text, Code, Multilingual
Meta Llama
Text, Chat, Code
Grok
Text, Vision, Code
Kimi
Text, Vision, Reasoning
Your first request.
From setup to result.
Start with a small task in Playground. Then use the same workspace balance for your application’s supported models, tools and compute.
Create your account
Sign up and verify your email. You can start in your browser without installing an SDK.
Claim your welcome credit
Open Usage & billing and claim $1.00 with the prefilled welcome code. No payment needed.
Run a real request
Choose a model in Playground and send a short prompt. Check the result and actual charge in Logs.
Connecting an existing agent? Get setup help from the founder.
Frequently
asked questions
How access and billing work.
What exactly do I get with Dreamtune?
Pay-as-you-go access to supported AI models, developer tools and GPU compute. Create a Dreamtune account, fund its prepaid balance and use services in the browser or through your API key. Dreamtune manages upstream access and deducts usage from that balance.
How many models, tools and GPU configurations are available?
The catalog includes 750+ AI models from 90+ model providers, 2,700+ API tools across 90+ tool services, and 22 GPU configurations. Counts describe the supported catalog. Check individual listings for available operations, rates and live GPU capacity.
Do I get my own account with every vendor?
You get a Dreamtune workspace and API key, not a dedicated account with every vendor. Some services use shared upstream accounts managed by Dreamtune. Your Dreamtune balance, keys and run history are scoped to your workspace. You do not receive vendor login credentials.
Can you combine the subscriptions I already pay for?
Dreamtune pays for supported usage made through Dreamtune. It does not import your existing subscriptions, pay their invoices or move the balances you already hold with other providers. Check the catalog for the specific models and operations available.
Can my existing AI agent use it?
Yes. Connect through MCP, the SDK or the CLI using a workspace API key with the permissions your agent needs. Hosted MCP requires Streamable HTTP and the X-Dreamtune-API-Key request header; it does not use browser OAuth. Local MCP is available for clients that cannot send custom headers. Follow the client-specific setup in Docs.
Can I add credit with crypto, or let my agent request a top-up?
Yes. Choose a supported asset and network in Usage & billing. An agent with billing permissions can also request payment instructions and check a top-up through the API or MCP. Your wallet separately authorizes and sends the transfer; the top-up API cannot sign transfers or withdraw funds. Credit is added after payment and settlement are verified. Check the quoted amount, network, fees and expiry before sending.
Can I rent any listed GPU immediately?
GPU capacity varies by configuration and region. Check live availability in the catalog before launching; unavailable configurations cannot be launched. Hosted MCP launches require your SSH public key. Compute is billed for consumed runtime, so stop the job when you finish.
Is this an unlimited subscription bundle?
One shared credit balance pays for supported services. Each tool call or compute job has its own usage cost. It doesn’t mean unlimited access or the same price for every task.
Can I bring my own API keys?
That’s part of the plan. You’ll be able to use supported provider keys alongside platform-managed access.
How will the credit marketplace work?
Sellers will be able to list eligible unused credits. Buyers will compare the provider, available value, price, and expiry before buying. Transfers will depend on provider approval and supported settlement. Join the wishlist to help us prioritize Marketplace.
Can I use it today?
Yes. Create an account to use supported models, tools and GPU compute with your workspace wallet. Marketplace is coming soon.
What happens if a provider fails during a task?
Failed or undelivered model and tool results do not debit your wallet. Successful calls completed earlier in your workflow remain billable, even if a later step fails. GPU compute is different: runtime already consumed is billable. Check Runs and Logs for the recorded outcome and Usage & billing for your balance.
Will Dreamtune automatically retry or switch models?
We do not automatically switch your request to another model or provider. A timeout does not prove that upstream work failed or stopped. Inspect the existing run or GPU job before submitting again; do not blindly retry a paid operation with a new idempotency key.
Can I set a hard spending ceiling on each task?
Not yet. Per-key budgets control admission of new work; they are not guaranteed per-task ceilings. A completed call can settle above the remaining wallet or key allowance, blocking further work. We do not currently reserve a maximum task budget and release the unused amount. Stop GPU jobs explicitly when you finish.
How do I see which project is spending the most?
Create a separate named API key for each project. In API keys, select View usage to see actual credits consumed, services used, recorded input/output tokens and GPU billing minutes for that key. Token and runtime details depend on recorded provider data. Dashboard activity without an API key is not assigned to a project key.
Can I try Dreamtune with free credit?
Create an account, verify your primary email, then select Claim credit in Usage & billing for USD 1.00 in welcome credit. The DREAM1 welcome code is already filled in. Available once per account and verified email. Contact the founder using the support email below if you need help testing a workflow.