Cursor is now part of X. Flintr is staying independent.

Every frontier model.
One independent editor.

Q: Who owns Flintr?

Flintr is independent. Development is funded by backers, and after launch by subscriptions, not by a platform with its own models or social network to promote. Our charter is part of our bylaws: if ownership ever changes, model neutrality and list-price usage transfer with the company, or annual plans are refunded.

Q: How does on-demand billing work?

Every paid plan includes a monthly amount of model usage. Once you've used it, further requests are metered per token at the provider's published list price, with no markup and no credits to convert. You see the cost of each request in the editor, and everything appears as one line on your monthly invoice.

Q: How do Pro, Pro+ and Ultra differ?

They're the tiers you already know. Pro ($20) includes $20 of model usage, Pro+ ($60) includes $70 plus priority capacity and max context, and Ultra ($200) includes $400 plus the highest priority and unlimited parallel background agents. Change tiers whenever you like and the difference is prorated.

Q: Can I set a spending limit?

Yes. Set a hard monthly cap for yourself, or per user and org-wide on Teams. When you reach it, Flintr either pauses premium requests or falls back to fast models, whichever you choose. Email and in-editor alerts fire at thresholds you pick.

Q: Can I keep separate subscriptions for different projects?

Yes. One account can hold several plans, in any mix of Pro, Pro+ and Ultra. Pin a project to a plan to keep its billing separate, for example for a client, and let every other project draw from your plans in the order you set. Each plan has its own invoice and spend cap.

Q: Which models are included?

Frontier and open-weight models from Anthropic, OpenAI, Google, DeepSeek, Mistral, Qwen, Meta, Moonshot and xAI, among others. We aim to add new models within a day of release. Auto picks a model per request, or you can pin one.

Q: Can I bring my own API keys?

Yes, on every plan. Add keys per provider and choose whether each provider uses your key or your included usage. Requests on your own keys are never counted against your plan.

Q: Will my Cursor extensions, rules and settings work?

Flintr is built on open-source Code OSS, so VS Code-compatible extensions, themes and keybindings work as they are. flintr import --from cursor also converts project rules, MCP server configs and chat history.

Q: Do you train on my code?

No. Privacy mode is on by default and we have zero-data-retention agreements with every model provider. Your code is never stored or used for training, by us or by anyone else.

We're building the AI code editor Cursor used to be. Claude, GPT, Gemini and open-weight models in one subscription, with on-demand usage billed at the provider's list price, and no parent company's ecosystem to lock you into.

Back the launch

How switching will work

Coming to macOS · Windows · Linux/pairing withClaude Opus

acme-api · usage.tsPrivacy mode

usage.ts

pricing.ts

1import { db } from "@/db"

2import { pricing } from "./pricing"

4export async function recordUsage(user: User, event: UsageEvent) {

5 const rate = pricing[event.model]

6 const cost =

7 (event.input * rate.in + event.output * rate.out) / 1e6

9 await db.usage.insert({ userId: user.id, cost })

11 const spend = await db.usage.monthToDate(user.id)

12 if (spend >= user.spendLimit) {Tab to accept

13 await notify(user, "spend-limit-reached")

14 return { blocked: true, spend }

15 }

16 return { blocked: false, spend }

17}

mainLn 12, Col 3

TypeScriptUsage: $14.20 of $20 included

Model-neutral

Every frontier model. One subscription.

No house model pushed on you, and no exclusive deal deciding what you can pick. Use Auto routing, pin a favourite, or switch mid-conversation.

Claude Opus

Anthropic

Deep reasoning

Claude Sonnet

Anthropic

Agent default

GPT-5

OpenAI

Frontier

GPT-5 mini

OpenAI

Fast

Gemini Pro

Google

1M context

Gemini Flash

Google

Fast

Claude Opus

Anthropic

Deep reasoning

Claude Sonnet

Anthropic

Agent default

GPT-5

OpenAI

Frontier

GPT-5 mini

OpenAI

Fast

Gemini Pro

Google

1M context

Gemini Flash

Google

Fast

Claude Opus

Anthropic

Deep reasoning

Claude Sonnet

Anthropic

Agent default

GPT-5

OpenAI

Frontier

GPT-5 mini

OpenAI

Fast

Gemini Pro

Google

1M context

Gemini Flash

Google

Fast

Claude Opus

Anthropic

Deep reasoning

Claude Sonnet

Anthropic

Agent default

GPT-5

OpenAI

Frontier

GPT-5 mini

OpenAI

Fast

Gemini Pro

Google

1M context

Gemini Flash

Google

Fast

DeepSeek

Open weights

Codestral

Mistral AI

Autocomplete

Qwen Coder

Alibaba

Open weights

Llama

Meta

Self-hostable

Kimi

Moonshot AI

Long context

Grok

xAI

Yes, even this one

DeepSeek

Open weights

Codestral

Mistral AI

Autocomplete

Qwen Coder

Alibaba

Open weights

Llama

Meta

Self-hostable

Kimi

Moonshot AI

Long context

Grok

xAI

Yes, even this one

DeepSeek

Open weights

Codestral

Mistral AI

Autocomplete

Qwen Coder

Alibaba

Open weights

Llama

Meta

Self-hostable

Kimi

Moonshot AI

Long context

Grok

xAI

Yes, even this one

DeepSeek

Open weights

Codestral

Mistral AI

Autocomplete

Qwen Coder

Alibaba

Open weights

Llama

Meta

Self-hostable

Kimi

Moonshot AI

Long context

Grok

xAI

Yes, even this one

models across 9 providers

markup on on-demand usage

<0h

from model launch to Flintr

invoice, whatever you use

Everything you liked

The workflow you know, minus the new owner.

Tab, Agent, rules, MCP and codebase context, built on open-source Code OSS so your extensions and keybindings just work.

prompt

auto-route

Auto routes every request

Quick edits go to fast, cheap models. Gnarly refactors go to frontier reasoning. Or pin the model you trust.

See all models

function slugify(title: string) {

return title.toLowerCase()

.replace(/[^a-z0-9]+/g, "-")

.replace(/(^-|-$)/g, "")

}

Tab ⇥ accept · 38ms

Tab that reads ahead

Multi-line, multi-file completions in under 50ms.

Unlimited on Pro

Searched codebase23 files

Agents that finish the job

Plan, edit, run tests and open the PR. You review the diff.

How agents pick a model

October usage

of $20.00 included in Pro

alert at 80%hard cap $50

Claude Sonnet$6.53

GPT-5$3.98

Gemini Pro$2.56

DeepSeek$1.13

Usage you can actually read

Live spend per model, alerts when you choose, and a hard cap that really is hard. No credits, no multipliers.

Estimate your bill

zero data retention

Private by default

Zero-retention with every provider. Never used for training.

How we handle your code

.flintr/rules/api-style.md

.flintr/rules/testing.md

mcp · github

mcp · linear

mcp · postgres

Rules and MCP, carried over

Your project rules and MCP servers work on day one.

Import your setup

sk-ant-••••••••3f9a

sk-proj-••••••••a71c

AIza••••••••Qx2

Bring your own keys

Mix your own API keys with included usage, per provider.

Supported providers

The Flintr charter

Built so this can't happen to you twice.

Flintr is independent and funded by its users. These commitments are written into our bylaws. If ownership ever changes, they transfer with the company, or every annual plan gets refunded.

Model-neutral, forever

No exclusive deals and no house model ranked above the rest. When a better model ships on Tuesday, it's in Flintr by Wednesday.

Usage at list price

Anything beyond your included usage is billed at the provider's list price. No credits, no multipliers, no "fast request" maths.

Your code stays yours

Privacy mode is on by default, with zero-retention agreements across every provider. Your code never trains anyone's model.

Leave whenever you like

Export settings, rules, chats and history in one click. If you can't leave an editor, you never really chose it.

Switching from Cursor

Bring your whole setup. One command.

Flintr reads your Cursor profile and moves everything across in under a minute. Same muscle memory, same rules, same MCP tools, under different ownership.

Extensions, themes and keybindings
settings.json and workspace settings
Project rules (.cursor/rules → .flintr/rules)
MCP servers and their credentials
Chat history and saved prompts
Your own API keys, if you used them

Read the full migration guide

✔ Found Cursor profile at ~/.cursor
✔ Imported 42 extensions and 3 themes
✔ Imported keybindings and settings.json
✔ Converted 7 project rules → .flintr/rules
✔ Connected 3 MCP servers (github, linear, postgres)
✔ Migrated 1,284 chats to local history
Settings sync: on · Privacy mode: on

Pricing

Pro, Pro+ and Ultra. Same plans, honest usage.

Every plan includes model usage. Go past it and you pay exactly what the provider charges, up to a cap you set.

Planned pricing at launch. Backers from €50 get months of Pro included.

MonthlyYearly · save 20%

Pro

For individual developers.

$20/ mo

Model usage included$20 / mo

Back the launch

Unlimited Tab completions
Every frontier model, plus Auto routing
On-demand usage at list price
Hard spend caps and alerts
Background agents
Bring your own API keys

Pro+

Recommended

For people who live in Agent.

$60/ mo

Model usage included$70 / mo

Back the launch

Everything in Pro
3.5x the included usage of Pro
Priority capacity at peak hours
Max context windows
Early access to new models

Ultra

For running agents all day.

$200/ mo

Model usage included$400 / mo

Back the launch

Everything in Pro+
20x the included usage of Pro
Highest priority, even on launch day
Unlimited parallel background agents
Priority support from our engineers

Estimate your bill

On-demand usage is metered per token at the provider's list price. Here's roughly what that looks like.

Agent requests per workday80

10400

Model mix

Estimated model usage

80 × 21 days × ~$0.025 = $42.00

ProBest

$42.00

Pro+

$60.00

Ultra

$200.00

Pro plan (includes $20.00): $20.00
On-demand at list price: $22.00
You pay: $42.00 / mo

Pro is $18.00 a month cheaper than Pro+ at this usage.

Assumes an average request of ~8k input and ~1k output tokens. Set a hard cap at any amount and Flintr pauses or falls back to fast models when you reach it.

Hobby will be free. Limited Tab completions and agent requests for everyone, from the public beta onwards. Get in earlier.

Teams and Enterprise. Pooled usage, SSO, org-wide privacy and admin controls. Talk to us.

Compare every feature across Hobby, Pro, Pro+, Ultra and Teams.

Compare

How Flintr compares.

Public plans and pricing, checked October 2026. Flintr is still in development; the others ship today.

Flintr compared with other AI coding tools
Feature	Flintr	Cursor	Devin Desktop	GitHub Copilot	Zed	Claude Code
What it is	AI code editor	AI code editor	Agent board and editor	Assistant for your IDE	Native editor	Terminal and IDE agent
Paid plans from	$20	$8	$20	$10	$10	$20
Included usage	Dollars, any model	Grok and Cursor models	Daily and weekly quotas	Monthly AI credits	$5 of tokens	Session and weekly limits
Beyond your plan	List price, no markup	API list price	API pricing	Per-token rates	List price +10%	API rates
VS Code extensions						In your IDE
Owner	Independent	X	Cognition	GitHub	Zed Industries	Anthropic
Available today

Swipe sideways to see every column.

Full comparisons:Flintr vs Cursor Flintr vs Windsurf / Devin Desktop Flintr vs GitHub Copilot Flintr vs Zed Flintr vs Claude Code

Back the launch

Funded by developers, not investors.

Investors eventually want an exit, and exits are how editors end up owned by someone else. We'd rather answer to the people who'll use it. Back the launch and every euro goes into development.

Read the backer terms

€35,000raised of €150,000

Includes €35,000 committed by the founding team.

backer: 1
funded: 23%
from: €5

Beta tester · Access to the private beta before the public launch€25+
Early adopter · 1 month of Pro at launch€50+
Founding member · 3 months of Pro at launch, instead of 1€100+

Back the launch

FAQ

Questions, answered.

Anything else? Our team replies within a few hours on hello@flintr.ai.

Who owns Flintr?

Flintr is independent. Development is funded by backers, and after launch by subscriptions, not by a platform with its own models or social network to promote. Our charter is part of our bylaws: if ownership ever changes, model neutrality and list-price usage transfer with the company, or annual plans are refunded.

How does on-demand billing work?

Every paid plan includes a monthly amount of model usage. Once you've used it, further requests are metered per token at the provider's published list price, with no markup and no credits to convert. You see the cost of each request in the editor, and everything appears as one line on your monthly invoice.

How do Pro, Pro+ and Ultra differ?

They're the tiers you already know. Pro ($20) includes $20 of model usage, Pro+ ($60) includes $70 plus priority capacity and max context, and Ultra ($200) includes $400 plus the highest priority and unlimited parallel background agents. Change tiers whenever you like and the difference is prorated.

Can I set a spending limit?

Yes. Set a hard monthly cap for yourself, or per user and org-wide on Teams. When you reach it, Flintr either pauses premium requests or falls back to fast models, whichever you choose. Email and in-editor alerts fire at thresholds you pick.

Can I keep separate subscriptions for different projects?

Yes. One account can hold several plans, in any mix of Pro, Pro+ and Ultra. Pin a project to a plan to keep its billing separate, for example for a client, and let every other project draw from your plans in the order you set. Each plan has its own invoice and spend cap.

Which models are included?

Frontier and open-weight models from Anthropic, OpenAI, Google, DeepSeek, Mistral, Qwen, Meta, Moonshot and xAI, among others. We aim to add new models within a day of release. Auto picks a model per request, or you can pin one.

Can I bring my own API keys?

Yes, on every plan. Add keys per provider and choose whether each provider uses your key or your included usage. Requests on your own keys are never counted against your plan.

Will my Cursor extensions, rules and settings work?

Flintr is built on open-source Code OSS, so VS Code-compatible extensions, themes and keybindings work as they are. flintr import --from cursor also converts project rules, MCP server configs and chat history.

Do you train on my code?

No. Privacy mode is on by default and we have zero-data-retention agreements with every model provider. Your code is never stored or used for training, by us or by anyone else.

Your editor. Your models. Your bill.

Flintr is in development and funded by the people who'll use it. Back the launch from €5, or from €25 to get into the private beta.

Back the launch

See planned pricing

Coming to macOS 12+ · Windows 10+ · Linux (deb, rpm, AppImage)

Every frontier model.One independent editor.

Every frontier model. One subscription.

The workflow you know, minus the new owner.

Auto routes every request

Tab that reads ahead

Agents that finish the job

Usage you can actually read

Private by default

Rules and MCP, carried over

Bring your own keys

Built so this can't happen to you twice.

Model-neutral, forever

Usage at list price

Your code stays yours

Leave whenever you like

Bring your whole setup. One command.

Pro, Pro+ and Ultra. Same plans, honest usage.

Pro

Pro+

Ultra

Estimate your bill

How Flintr compares.

Funded by developers, not investors.

Questions, answered.

Who owns Flintr?

How does on-demand billing work?

How do Pro, Pro+ and Ultra differ?

Can I set a spending limit?

Can I keep separate subscriptions for different projects?

Which models are included?

Can I bring my own API keys?

Will my Cursor extensions, rules and settings work?

Do you train on my code?

Your editor. Your models. Your bill.

Every frontier model.
One independent editor.