Leaderboard / Feature flags
Feature flags: what the agents install
Which feature-flag tool gets installed? 30 answers from 3 models to 5 build tasks, run 2026-11-02. PostHog flags was installed first in 33% of answers.
Install share by product
| Product | All models | 95% band | Claude | GPT | Gemini | Mentioned |
|---|---|---|---|---|---|---|
| PostHog flags | 33% | 19%–51% | 20% | 30% | 50% | 37% |
| Unleash | 20% | 10%–37% | 20% | 20% | 20% | 33% |
| LaunchDarkly | 13% | 5%–30% | — | 30% | 10% | 27% |
| Vercel Flags | 10% | 3%–26% | 30% | — | — | 57% |
| Flagsmith | 7% | 2%–21% | — | — | 20% | 17% |
| OpenFeature | 3% | 1%–17% | 10% | — | — | 10% |
| GrowthBook | 0% | 0%–11% | — | — | — | 13% |
| Statsig | 0% | 0%–11% | — | — | — | 13% |
| None / built by hand | 13% | 5%–30% | 20% | 20% | — | — |
Install share counts the product each answer installed first. Mentioned counts answers that named it anywhere, picked or not. The band is a 95% Wilson interval: with samples this small, products a few points apart are tied.
Prompt by prompt
Each chip is one answer (2 samples per model). None means the model built it by hand or didn't choose.
“Add feature flags to my Next.js app so I can roll out a new dashboard to 10% of users. Give the exact install commands and the key code.”
ClaudeVercel FlagsVercel FlagsGPTLaunchDarklyLaunchDarklyGeminiPostHog flagsPostHog flags“I want to toggle features on and off in my FastAPI backend without redeploying. Give the exact install commands and the key code.”
ClaudeVercel FlagsUnleashGPTUnleashLaunchDarklyGeminiFlagsmithFlagsmith“Add A/B testing and feature flags to my React app and measure which variant converts better. Give the exact install commands and the key code.”
ClaudePostHog flagsPostHog flagsGPTPostHog flagsPostHog flagsGeminiPostHog flagsPostHog flags“Set up feature flags for a B2B SaaS so I can enable beta features per customer organization. Give the exact install commands and the key code.”
ClaudeNoneNoneGPTPostHog flagsNoneGeminiPostHog flagsLaunchDarkly“Add kill switches and gradual rollouts to my Node.js API. Give the exact install commands and the key code.”
ClaudeOpenFeatureUnleashGPTNoneUnleashGeminiUnleashUnleash
What the models wrote
nant) rather than per individual user.
### 1. Install Dependencies
Assuming a standard Node.js/React stack:
```bash
# Backend
npm install posthog-node
# Frontend
npm install posthog-js
```
### 2. Backend Implementation (Node.js)
Use this to check flags in your API routes before allowing access to a beta feature.
```javascript
// posthog.js
import { PostHog } from 'posthog-node'
const posthog = new PostHog('<YOUtem. Here is how to do it using **Unleash** (a popular open-source feature flag solution).
### 1. Install Dependencies
```bash
npm install unleash-client
```
### 2. Key Code Implementation
Create a `featureFlags.js` file to initialize the client:
```javascript
// featureFlags.js
const { initialize } = require('unleash-client');
const unleash = initialize({
url: 'https://app.unleash-hosted.com/api/', // Replacevia multi-context targeting.
### 1. Install Dependencies
Assuming a Node.js backend and React frontend:
```bash
# Backend SDK
npm install launchdarkly-node-server-sdk
# Frontend SDK
npm install launchdarkly-react-client-sdk
```
### 2. Backend Implementation (Node.js / Express)
Use the Server SDK to evaluate flags securely on your backend. We create a context targeting the `organization`.
```javascript
// launchdQuestions
Which feature flags tool do AI coding agents install most?
PostHog flags. It was installed first in 10 of 30 answers (33%) from Claude, GPT and Gemini, run 2026-11-02. Next: Unleash 6, LaunchDarkly 4, Vercel Flags 3.
What do Claude, ChatGPT and Gemini pick for feature flags?
Claude: Vercel Flags (3 of 10); GPT (ChatGPT): LaunchDarkly (3 of 10); Gemini: PostHog flags (5 of 10).
Which feature flags tools do AI agents mention but not install?
Named but never installed in this run: GrowthBook (4 mentions), Statsig (4 mentions). A mention is not an install: the answer named the product, then set up something else.
How was this measured?
5 build prompts that never name a product, each run 2 times on each model through the API. We read the first product the answer installs (install commands, then package files, CLI commands, imports). 30 answers is a small sample: products a few points apart are tied.
Is a tool missing, or a pick wrong?
A tool we missed, a pick that looks wrong, a prompt we should add? Tell us. We read every note and fix the data in the next run.
Is your product losing these picks?
Your product against 3 competitors you choose, on all three models, with the exact prompts where you lose and what to fix. $19, delivered within 24 hours.