AI Girlfriend Industry Report 2026: Trends, Players, and What Changed

Every year, I re-audit our entire database of 129 AI girlfriend platforms: I revisit each one, check whether it still exists in the form we originally tested, retest the features, and update the pricing and scores. That process gives me something most coverage of this category doesn't have, a genuine before-and-after view of the same set of products over time, rather than a single snapshot. This report is what that comparison actually shows. I'm not naming specific platforms here, since the point is the pattern across the industry, not any one company's story.

The year in one paragraph

If I had to summarize the last year of re-testing in one sentence: the category got more feature-rich at the top end, more price-competitive in the middle, and more unstable at the edges. Platforms that were already investing seriously kept adding capability, especially video. Platforms in the middle of the pack restructured pricing to stay competitive. And a real chunk of platforms, roughly 1 in 6, simply didn't survive the year in their original form.

Trend 1: AI video generation went from a curiosity to a real feature category

The clearest structural shift this year is video. When we last tallied it, 22% of the 129 platforms in our database now offer some form of AI video generation, up from a much smaller base when we first started tracking the category. That's still a minority of platforms, and it's still the newest and least mature feature layer we track, but it's no longer a novelty limited to one or two experimental products. It's become a real axis platforms compete on.

What's notable is how uneven this adoption is. Video generation grew meaningfully this year, while voice interaction, a much older and arguably more foundational feature, barely moved and remains the industry's weakest category at 1.81 out of 5. Platforms appear to be prioritizing newer, more marketable features over shoring up long-standing gaps, which is a pattern worth watching rather than assuming will self-correct.

Trend 2: pricing restructured across a big share of the industry

Pricing moved more than any other business dimension this year. In our re-audit, we found at least 39 platforms had added a genuine free tier that didn't exist during our original testing, and at least 28 platforms had restructured a single flat price into multiple tiers. That's a substantial share of a 129-platform database changing its core pricing model within about a year.

The direction of that change is consistent: more entry points, more tiers, and more granular gating of features like voice, images, and memory behind specific price points rather than an all-or-nothing subscription. I break this specific pattern down in more depth in a dedicated piece on how AI girlfriend pricing shifted in 2026, but the short version is that competitive pressure from having 129-plus alternatives available is clearly pushing platforms toward lower-friction entry points rather than higher prices.

Trend 3: churn stayed high, and it's a defining feature of this market, not a footnote

The most consequential number from this year's re-audit, in my view, is churn. At least 23 of the 129 platforms in our database, about 18%, had gone dark, been sold and rebranded, or started silently redirecting to an unrelated product within roughly a year of our original testing. That's not a small failure rate. It's a genuine structural feature of this industry right now, and it means any snapshot of "the market" is already partially out of date within months.

This churn also connects directly to something I've written about separately: naming collisions. When a platform disappears or rebrands, its old name and domain often get repurposed, sometimes by an unrelated company, which is part of why the industry's naming collision problem keeps getting worse rather than settling down over time.

A gap that didn't close: customer support investment

One area our re-audit found essentially unchanged is customer support. 78% of the 129 platforms we track still have no clearly documented support channel, a figure that's barely moved even as platforms poured visible investment into video generation and pricing restructuring. That's worth calling out specifically because it's the kind of gap that's easy to deprioritize precisely because it's invisible until a user actually needs it, unlike a new video feature that shows up directly in marketing screenshots.

What stayed remarkably stable

Not everything moved this year. A few things held steady across our re-audit:

• Chat quality stayed the strongest core feature, averaging 3.26 out of 5, roughly consistent with what we measured previously.
• The NSFW-to-SFW split held close to the same proportions, with 104 of 129 platforms allowing NSFW content and 25 remaining SFW-only, both groups still averaging identically at 2.5 out of 5.
• Customer support remained the industry's second-weakest category, with 78% of platforms still lacking a clearly documented support channel, a gap that new feature investment hasn't meaningfully closed.

That mix of movement and stability is itself informative. Platforms are willing to invest in visible, marketable features like video and pricing structure, but less willing to invest in less visible fundamentals like support infrastructure and voice quality, even though both remain significant gaps.

What I'll be watching in the next re-audit

Based on this year's pattern, three things I expect to keep tracking closely: whether video generation adoption keeps climbing from its current 22%, whether the churn rate stays around 18% or shifts as the market matures, and whether voice interaction finally starts closing its gap with chat quality given how far behind it remains. None of these are predictions in the market-research sense, just the specific things our own re-audit process is built to catch the next time we run it.

How we actually track year-over-year change

This report is built entirely from comparing our own testing data across time, not from industry forecasts or third-party market research. We maintain a live database of 129 platforms, re-audit it periodically to catch closures, rebrands, new entrants, and feature or pricing changes, and re-score each platform using the same five-category methodology every time so comparisons stay apples-to-apples. You can read the full testing methodology, and see the complete current snapshot of every number we track on our data hub.

A report like this one is only useful if it helps you make an actual decision, not just track abstract trends. If you're trying to figure out which platform in this crowded, fast-changing market is actually worth your time and money right now, rather than reading about the industry in the abstract, our best AI girlfriend rankings apply this exact same testing process to all 129 platforms individually, updated as our re-audits catch changes like the ones described in this report.

Further reading

• How AI Girlfriend Pricing Shifted in 2026
• We Found 23 AI Girlfriend Platforms That Went Dark, Got Sold, or Rebranded
• The State of the AI Girlfriend Industry in Numbers