1/4
Every fight on the next cards, one tap away
22 fights across 10 weight classes: pick any bout on the next UFC cards, from the main event to the first prelim, by card or by division.
The world's first AI fight analyst.
+42K satisfied users
Champion · Featherweight title · UFC 333
1,574 hours
of UFC fights since 1993: all 8,943 of them
4,196 fighters
every athlete on UFC.com, read in
Introducing
The CageFrame model. One rating per fighter, seven factors, one win probability.
Topuria vs Holloway · UFC 308
Selected exchanges against Emmett, Volkanovski and Holloway show Topuria lowering his right hand as he throws his left jab. Holloway’s response illustrates a possible counter window. Curated from fight footage and the existing analysis, with original fight audio. Pattern detected: left jab, right hand drops, counter window. Explore possible counterattacks.
From the card to the cage door, here's how the fight desk turns the numbers into a read on every fight.
1/4
22 fights across 10 weight classes: pick any bout on the next UFC cards, from the main event to the first prelim, by card or by division.
2/4
Records, size, reach, age, the last five fights and every career average, lined up on one screen so the differences jump out. Official numbers only.
3/4
The model's win probability and the factors that move it, then a written breakdown: the key edges, how each fighter wins, what to watch in the cage.
4/4
Type a question, get a short answer from the fight's own numbers, with its sources.
Illustrative reviews.
Every plan includes the fighter stats and Prometheus's win probability. Pick how many full AI breakdowns you want.
Full AI breakdowns for the fights you follow most.
$12.99per month
Every fight on CageFrame, fully broken down.
$24.99per month
Every published fight analysis, with more room to ask.
$99per month
| Starter, $12.99 a month | Pro, $24.99 a month | Ultra, $99 a month | |
|---|---|---|---|
| Fighter stats and comparisons | |||
| Fighter stats and comparisons | Included | Included | Included |
| Model win probability | |||
| Model win probability | Included | Included | Included |
| Full AI analysesOne fight each, open until after its event | |||
| Full AI analysesOne fight each, open until after its event | 5 a month | Unlimited | Unlimited |
| Patterns and weaknesses | |||
| Patterns and weaknesses | Included | Included | Included |
| Victory scenarios | |||
| Victory scenarios | Not included | Included | Included |
| Sourced fight-week news | |||
| Sourced fight-week news | Not included | Included | Included |
| Ask CageFrame questions | |||
| Ask CageFrame questions | 10 a month | 75 a month | 200 a month |
A clear answer to all your questions about CageFrame.
No. CageFrame is fight analysis for fans: no wagers, no deposits, no betting accounts, no odds.
Each "Model win probability" is set by a statistical model. The AI only writes the matchup text around it and never sets a number. The model rates every fighter from all UFC results on Wikipedia since 1993 (a Glicko-2 rating), then weighs that rating with age, reach, time since the last fight, UFC experience, current streak and the strength of recent opponents. It was fitted on 2005–2014 fights, its settings were chosen on 2015–2020 fights, and it was then tested on 2,968 UFC fights with a winner since 2021, none of which it was trained or tuned on. That test period has been scored three times, so its pass is weaker evidence than a single look (the credits page lists each scoring); the real check is the ledger of numbers frozen before each fight. It learned from Wikipedia and Wikidata only. For a fighter on the next cards whose date of birth or reach is missing there, or disagrees with our hand-checked card data, it uses the card value, and the fight page says so (with no value anywhere, it uses the division average and flags that too). Each percentage is the chance of winning if the fight has a winner. It is frozen before the fight in an append-only ledger: a correction is a new entry, and the fight page still lists the number it replaced, under "Sources and versions". When the whole percentages read 50/50, neither fighter is picked. It doesn't use striking or grappling stats (there's no licensed fight-by-fight history of them) or betting odds. When two fighters both make their UFC debut there isn't enough history, so there's no number.
Records and career averages come from UFC.com athlete profiles and UFC's event data, cross-checked against Wikipedia, Sherdog and ESPN. Every fighter profile links to its exact sources.
Every bout on the next UFC cards: UFC Fight Night Buckley vs. Malott (Sat, Oct 17, 13 fights) and UFC 333 Volkanovski vs. Evloev (Sat, Oct 24, 9 fights).
All 4,196 athletes with a UFC.com profile, active or not, with their record, ranking, career stats and latest UFC fights as UFC.com publishes them. Fighters on the next cards get a deeper, cross-checked profile.
No. CageFrame is independent and is not affiliated with, endorsed by or sponsored by the UFC, Zuffa or TKO.
Every fight on the next cards, broken down before the first bell.
See analysis