PostHog and Mixpanel answer the same core question about what people do inside your product, and on paper they now look alike, with funnels, retention, session replay, feature flags, an AI agent and an official MCP server on both. They split on three things. One is the bill once you grow, another is self-hosting, which only PostHog allows, and the last is how much else comes in the box.
We make Spreelytics, a third tool, and say where it fits at the end. Every price and limit below was read from the two vendors' pricing pages on 12 November 2026. Check them before you commit.
The short answer
PostHog costs less at volume for anonymous traffic, and it's the only one of the two you can host yourself. Mixpanel's free plan keeps two years of data and gives you twice the replays. Signed-in users change the maths, because PostHog adds a surcharge for identified events that closes most of the gap.
At a glance
| PostHog | Mixpanel | |
|---|---|---|
| Free events a month | 1M | 1M |
| Data kept on the free plan | 1 year | 2 years |
| Session replays free each month | 5,000 | 10,000 |
| Flags and experiments on the free plan | Yes, 1M flag requests | Yes, up to 10 active flags |
| Saved reports on the free plan | Unlimited | Unlimited |
| 10M events, billed monthly | $324.40 anonymous, $1,080 identified | $1,176 |
| Open source and self-hosting | Yes, MIT | No |
| Data regions | US or EU | US or EU |
| Official MCP server | Yes | Every plan |
Free plans
PostHog gives you 1 million events a month in one project, kept for a year, plus separate free allowances: 5,000 session recordings, a million feature flag requests, 1,500 survey responses and 100,000 error exceptions. Go past the million on the free plan and the extra events are dropped.
Mixpanel also starts at 1 million events. It adds 10,000 session replays and up to 10 active feature flags, and saved reports are now unlimited. Cohorts are the catch. You can build one on the free plan, but you can't save it.
What they cost as you grow
Both price by events and keep the free million once you pay.
| Events a month | PostHog, anonymous | PostHog, identified | Mixpanel Growth, monthly |
|---|---|---|---|
| 1.5M | $25 | $124 | $140 |
| 10M | $324.40 | $1,080 | $1,176 |
Watch the identified column. PostHog charges a second rate, on top of the base one, for any event tied to a known person, and a logged-in app sends mostly those. At 10 million identified events the two monthly bills land within about $100 of each other. Paying Mixpanel yearly brings its 10 million price down to $840.
Autocapture matters here too. PostHog records clicks and form changes as events on the same meter as pageviews, so a page that gets a thousand visits can send several thousand events. The full working is in PostHog pricing and Mixpanel pricing.
Where they differ
Self-hosting. PostHog is MIT licensed, and you can run it on your own server. Its docs aim that version at hobby projects on one machine, up to a couple of hundred thousand events. Mixpanel is cloud only.
What else is bundled. PostHog also sells surveys, error tracking, a data warehouse, logs and a self-driving inbox that files reports and drafts code changes, each on its own meter. Mixpanel stays close to analytics, with replay, flags and experiments beside it and little else.
Revenue. PostHog's revenue analytics is in beta. It reads from Stripe only so far. Mixpanel records revenue as event properties, with attribution models on its paid plans.
AI. Each has an in-app agent and an MCP server. PostHog bills its AI features in credits, with 500 free a month, while Mixpanel lists its agent on every plan, Free included.
Which to pick
- You want to host it yourself: PostHog. Mixpanel has no self-hosted version.
- Mostly anonymous traffic, growing past a few million events: PostHog costs less.
- A logged-in product at around 10 million events: close on monthly billing, and Mixpanel's yearly price is lower.
- Replays on a free plan: Mixpanel gives twice as many.
- One vendor for surveys, errors and a warehouse too: PostHog.
When neither is the right tool
Both are product tools first. If the question you ask most is which channel brings in the people who pay, you end up building that answer yourself in either one.
That's the question Spreelytics is built around. Payments from Stripe, Paddle, Polar or Lemon Squeezy are credited to the channel that earned them, last-touch or linear, on every plan including Free. AI crawlers are counted on your server. Our MCP server works with Claude, ChatGPT, Claude Code and Cursor, and an event costs the same whether the person is known or not: $339 a month at 10 million.
There are no feature flags or experiments here, replay needs our Growth plan, and there's no self-hosted version. If any of those matter every week, pick one of these two. See Spreelytics vs PostHog and Spreelytics vs Mixpanel, or the wider field in the best product analytics tools.
Questions people ask
Is PostHog cheaper than Mixpanel?
For anonymous events, yes, by a wide margin. At 10 million a month, billed monthly, PostHog is $324.40 against Mixpanel's $1,176, though identified events bring PostHog to $1,080, close to Mixpanel's monthly price and above its yearly one.
Can I self-host Mixpanel?
No. It runs in Mixpanel's cloud only, in the US or the EU. PostHog can be self-hosted, though its docs pitch that at small setups.
Which has the bigger free plan?
It depends what you count. Mixpanel keeps data for two years and gives 10,000 replays. PostHog keeps a year and gives 5,000, but adds free allowances for surveys, error tracking and its data warehouse.