DevPromptAI
A free browser-based workspace that connects your own API keys to nine AI providers, letting you run one prompt across several models at once, compare answers side by side, and build a reusable prompt library.
What is DevPromptAI?
DevPromptAI is a browser-based workspace that sits on top of the commercial AI APIs rather than providing a model of its own. It follows a bring your own key model: you supply an API key from OpenAI, Anthropic, Google Gemini, DeepSeek, Perplexity, OpenRouter, Hugging Face, Box AI or a local Ollama server, and the application becomes the interface to all of them. The FAQ is explicit about what that means technically: the key is stored encrypted in your browser's local storage and never sent to DevPromptAI, and requests go straight from the browser window to the provider's API. Around sixty-six predefined models ship with the app, from GPT-4.1 and Claude Opus 4 to DeepSeek R1 and Perplexity's web-searching Sonar family, and you can add your own.
The workspace is organised around a handful of objects. Chats can be saved, pinned, searched and grouped into projects. Assistants are reusable personas with their own system prompt, tone preset, temperature, top-p, frequency and presence penalties, avatar and output language, chosen from ninety-five. Snippets are prompt fragments recalled by typing a slash in the message bar. A multi-model mode sends one prompt to several models and lays the answers out side by side, with Run All and Stop All controls; this is the feature the product leads with. A prompt generator rewrites a rough instruction into a more precise one. A token counter and a cost estimate sit at the foot of the panel, both flagged as approximate.
An account is optional. Without one the tool works but saves nothing; with one, opened by email and password, Google or GitHub, chats, assistants, snippets and layouts synchronise through Firebase Realtime Database. A saved chat can be shared by link, publicly or restricted.
Beyond text, the app reaches DALL-E 3 and DALL-E 2 for images and OpenAI for audio transcription, and it bundles Tesseract for on-device OCR. It renders Markdown and highlights code. There is no mobile version: phone and tablet visitors are routed to a dedicated not-supported page.
What it does
- Query nine AI providers from a single interface
- Run one prompt across several models and compare the answers side by side
- Build a reusable library of prompts, snippets and projects
- Configure specialised assistants with their own system prompt, tone and sampling settings
- Have the AI rewrite and optimise a rough prompt
- Estimate the input tokens and the cost of an exchange
- Share a saved chat through a link
When to use DevPromptAI / When not to
A quick filter to help you decide if DevPromptAI is the right fit.
When to use DevPromptAI
- Developers who already hold API keys from OpenAI, Anthropic or Google and want one interface for all of them
- Technical teams comparing several models on the same prompt before committing to one
- Prompt engineers building and reusing a structured library of prompts, snippets and assistants
- Freelancers and small teams who would rather pay providers per token than add another monthly subscription
- Privacy-minded users who want their API key to stay in their own browser and their prompts to bypass a third-party server
When not to use DevPromptAI
- Newcomers without an API key who expect a ready-to-use assistant: the app supplies the interface, never the model
- Anyone working from a phone or tablet, since mobile visitors are routed to a not-supported page
- Users under 13, excluded by the privacy policy
- Developers needing programmatic access: DevPromptAI consumes AI APIs but exposes none of its own
- Regulated organisations requiring a signed DPA, a published subprocessor list or SOC 2 evidence, none of which exists
How to use DevPromptAI
A typical end-to-end flow, from setup to results.
- Open devprompt.ai in a desktop browser; phones and tablets are redirected to a not-supported page
- Create an account with one of the AI providers and generate an API key
- Open Settings, go to the Models tab and paste the key, which the application validates immediately
- Adjust the API URL for that provider if you route through a proxy or a local Ollama server
- Optionally sign up with an email address, Google or GitHub to save and synchronise your work
- Type your prompt in the message bar, or press slash to insert a saved snippet
- Press Run for a single answer, or Run All to query several models at once
- Create an assistant under Settings to fix a system prompt, a tone and an output language
- Save the chat, assign it to a project and pin the ones you come back to
- For local models, start the Ollama server with OLLAMA_ORIGINS set to https://devprompt.ai
Pros & Cons
Pros
- Genuinely free, with no paid tier, paywall or payment processor anywhere in the shipped code
- Breadth of choice: nine providers and around sixty-six models without leaving the interface
- The API key stays encrypted in the browser and, per the FAQ, is never transmitted to the publisher
- Spending stays under your control, since the provider bills you directly at token rates
- No sign-up required to start; an account only adds saving and synchronisation
- Fine-grained assistant settings usually reserved for API playgrounds
- Ollama support, so local open-source models can be run at no cost at all
Cons
- You must obtain an API key yourself, a real barrier for non-technical users
- No mobile version of any kind
- No public pricing, features, about or contact page: everything lives inside application modals
- Fully JavaScript-rendered and unindexable, with neither robots.txt nor sitemap.xml served
- The publisher is barely documented: AiDevs, with no legal form, postal address or registry entry
- A single support channel, and it is a Gmail address
- Terms and privacy policy unchanged since 1 April 2024, while the model catalogue is kept current
Pricing & Plans
DevPromptAI is free of charge. A permanent free plan is the only plan, there is no paid tier and therefore no entry price point in any currency, and no payment processor appears anywhere in the delivered code. The expense sits elsewhere: the user pays the AI provider directly for the tokens consumed. OpenAI, Anthropic and Perplexity bill through prepaid credits that must be purchased before use; Gemini and Hugging Face offer a limited free daily allowance; OpenRouter works on prepaid credits and exposes some free models; models run locally through Ollama cost nothing. The application displays a cost estimate that it explicitly presents as approximate.
- full access to every feature
- with your own API keys
- at no charge
- the application works
- but nothing is saved or synchronised
- adds cloud saving and synchronisation of chats
- assistants
- snippets and layouts
Data, GDPR & hosting
A consolidated view of how DevPromptAI handles your data.
GDPR overview
DevPromptAI never claims GDPR compliance in so many words, and never denies it either. The privacy policy does grant rights that mirror the regulation: access, correction and deletion at any time, data portability, and objection to certain uses, all exercised by emailing devprompt.ai@gmail.com. A section on international transfers acknowledges that EU/EEA data may move outside the EEA and says processors are bound by a data processing agreement. Beyond that the file is thin: no processor is named, no lawful basis is given, no standard contractual clauses are cited, no Article 27 representative is designated and no data protection officer is appointed. The stated minimum age is 13, below the 16-year default the regulation sets for member states that have not lowered it. The policy carries a single date, 1 April 2024.
Who owns the data?
The terms are explicit: you keep ownership of everything you put into the service, and AiDevs claims no intellectual property rights over your content. The publisher also states that it does not sell or share your data with third parties and that you retain all rights to it. Two carve-outs matter. Any feedback you send can be reused by the company with no obligation towards you, and the DevPromptAI name, logo and trademarks remain the publisher's. Architecturally the point is close to moot: prompts travel from your browser straight to the provider's API, so the material never reaches an AiDevs server in the first place.
Reuse rights
You may reuse whatever you generate without asking permission. The publisher claims no rights over your content and imposes no licence on outputs, so the code, text and images produced through the service can be used commercially, published or redistributed at your discretion, subject only to the terms of the AI provider whose model produced them. What DevPromptAI itself collects is narrow: an email address for the account, a first-party session cookie, device information gathered at sign-up, and Google Analytics on the public site. It is used to run, maintain and improve the service and to send confirmations, technical notices, security alerts and support messages. Saved chats, assistants and snippets sit in Firebase Realtime Database once an account exists, while the API key never leaves the browser. Nothing published states whether customer content feeds model training.
Data retention & training
Hosting summary
The privacy policy names Firebase Realtime Database as the storage for everything saved once an account exists, but gives no country and no region. The application's own code is more specific: it points at a Firebase instance in europe-west1, Google Cloud's Belgian region. That is a technical fact read from the delivered JavaScript rather than a commitment published by the publisher, and it should be treated as such. The website itself resolves to 63.176.8.218, an Amazon address in Frankfurt, Germany. The policy adds that data belonging to EU/EEA users may in some cases be transferred outside the EEA, without naming a destination or a safeguard beyond a data processing agreement with unnamed processors. One structural point matters more than any of this: prompts and answers do not pass through DevPromptAI infrastructure at all. They travel from the browser to the API of whichever provider you have connected, so the hosting question for your actual content is answered by that provider, not by this tool.
Things to keep in mind
Risks and trade-offs to weigh before adopting DevPromptAI.
- Your API key sits in the browser's local storage: a shared, borrowed or compromised machine exposes it, and the bill is yours
- No spending cap exists on the DevPromptAI side, so budget guardrails have to be set with the provider; an unattended multi-model run burns tokens fast
- Putting four models on every question is a comfortable way to outsource judgement, and the habit erodes the ability to weigh an answer on its own merits
- Token and cost figures are estimates that the application explicitly declines to guarantee
- The service is supplied as is, with no uptime guarantee and best-effort email support from a single address
- The publisher may suspend or terminate an account at any time without notice; disputes fall under Polish law after sixty days of informal negotiation, with class actions waived
- A chat shared in public mode is reachable by anyone holding the link, prompts and answers included
Setup & Integrations
Technical difficulty
Low for a technical user, moderate for anyone else. Nothing is installed and nothing is downloaded: the tool runs in a desktop browser. One step decides the experience, namely opening an account with an AI provider and generating an API key, then pasting it into Settings, where the application validates it immediately. A DevPromptAI account is optional and free. The only genuinely advanced case is running local models, which requires starting the Ollama server with OLLAMA_ORIGINS set to the site's address.
Deployment
Integrations
Supported languages
Behind DevPromptAI
Resources
All the official URLs gathered for verification and reference.
Frequently asked questions
Is DevPromptAI free?
Do I need an API key to use it?
Where is my API key stored?
Which AI providers can I connect?
Do I have to create an account?
Can I compare several models on the same prompt?
How is the cost calculated?
Is there a mobile app?
What is the minimum age?
How do I delete my account?
Should you pick DevPromptAI?
DevPromptAI is best understood as an interface layer, not an AI service. It supplies no model and charges nothing; what it offers is the convenience of driving nine providers and some sixty-six models from a single window, with your own API keys and your own provider bill. For a developer who already holds those keys that is a genuinely useful proposition, and the side-by-side comparison of several models on one prompt is the part that most justifies the tool's existence. The assistant settings, the snippet library and the project grouping are refinements an API playground rarely bothers with.
The reservations concern the publisher rather than the product. AiDevs appears once, in the terms, with no legal form, no postal address and no registry trace; the single contact channel is a Gmail address; there is no about page, no pricing page and no contact page, because there are no pages at all beyond the application itself. The terms and the privacy policy have not moved since 1 April 2024, even though the model catalogue has been kept current. Nothing is published about security certification, subprocessors, or whether customer content feeds model training, the last of which the architecture arguably makes moot since prompts go from the browser straight to the provider.
That combination suits an individual or a technical team weighing convenience against a modest amount of trust. It suits a regulated organisation much less: without a data processing agreement, a subprocessor list or an identifiable legal entity, the paperwork is simply not there. Treat it as a well-built free client for APIs you already pay for and it delivers; expect a vendor relationship and it does not.
- Choosing a selection results in a full page refresh.
- Opens in a new window.