UFC 333: Volkanovski vs. Evloev — AI crowd leans Evloev, Claude Sonnet doesn't

The AI crowd picks Movsar Evloev to beat Alexander Volkanovski in Saturday's featherweight title fight, but only narrowly. The average chance was 54%, and 73 of 100 forecasters took Evloev. The crowd is 100 forecasters played by five Claude and GPT models, all reading the same fact sheet, most from an assigned angle.

Last week, at UFC Fight Night: Buckley vs. Malott, the crowd picked 7 of 12 winners. Our model and the betting favourites each got 6 of 12. The crowd's two biggest misses were Erin Blanchfield and Melissa Croden, both picked at 68%. Jasmine Jasudavicius beat Blanchfield on a decision, and Chelsea Chandler submitted Croden in round 1.

Claude Sonnet sided with Volkanovski

The main event split the crowd by model more cleanly than any other fight. Claude Sonnet 5.5 picked Volkanovski with 21 of its 22 forecasters. Claude Opus 5.5 and Claude Fable 5.1 each had 20 of 21 on Evloev, and both GPT models had 16 of 18.

A Claude Sonnet 5.5 forecaster weighing how the judges see it gave Volkanovski 58%: "Champion with five-round striking edge and two Lopes wins, but Evloev's high-volume wrestling and unbeaten decision grinding make judges' scoring close." A Claude Fable 5.1 forecaster going by recent form gave Evloev the same 58%: "Evloev 20-0 and improving, took Murphy down 9 times; Volkanovski is 38 with two recent KO losses before beating Lopes twice."

The forecasters weighing the stand-up were the only angle group to favour Volkanovski, 8 of 10. The betting market has the fight dead even at 50%. Our model leans Evloev at 57%, a little further than the crowd.

Three fights where the crowd went against the odds and our model

In three fights, the crowd picked a fighter whom both the betting market and our model rate as the underdog:

• Jacob Malkoun: 97 of 100 forecasters took him, at 58%. The odds have Abubakar Vagaev at 58%, our model at 57%.
• Noah Gugnon: 88 of 100 took him, at 56%. The odds have Kody Steele at 59%, our model at 57%.
• Cam Rowston: 78 of 100 took him, at 54%. The odds and our model both have Abus Magomedov at 53%.

On Malkoun, a Claude Opus 5.5 forecaster weighing age, time away and the scale wrote: "Vagaev is moving up from welterweight on one month's notice to face Malkoun, a natural middleweight wrestler in his third fight of 2026."

Rowston is also where Claude and GPT parted ways. Claude forecasters backed him 63 of 64, while GPT forecasters took Magomedov 21 of 36. A GPT Astra forecaster going by recent form put the case for Magomedov: "Magomedov has won four of five against stronger opposition; Rowston's unbeaten UFC run has yet to include a ranked opponent."

Where the crowd was surest

Four fights were unanimous: Abdul Rakhman Yakhyaev (72%), Asaf Chopurov (70%), Alexander Volkov (66%) and Aaron Pico (64%). The odds rate Yakhyaev and Chopurov even higher, at 79% and 82%. Volkov is the outlier. The crowd's 66% compares with 53% in the betting market and 51% from our model.

Iasmin Lucindo also sits at 72%, though the two families see her differently. GPT forecasters average 80%, Claude forecasters 67%. Eight of 22 Claude Sonnet 5.5 forecasters picked Nicolle Caliari, the only votes she got.

The closest call was Grant Dawson over Nurullo Aliev, at 53%. Claude Fable 5.1 went the other way, with 15 of 21 on Aliev, and our model leans Aliev at 52%.

Our model and the betting market as of Oct 22, 5:11 PM ET.

How the AI crowd works

The crowd is 100 AI forecasters, played by Claude Opus 5.5, Claude Sonnet 5.5 and Claude Fable 5.1 from Anthropic, and GPT Astra and GPT-6.1 Sol from OpenAI. Most were told to weigh the fights from one of nine angles: the stand-up, the grappling, the numbers, durability and cardio, age and time away, recent form, the matchup, the sceptic's case and the judging. The rest got no angle at all, only the fact sheet and the question.

All of them read the same fact sheet: each fighter's record, size, career stats and recent UFC fights from our database, plus this week's news, gathered once for each fight. None of them saw the betting odds, our model or each other's picks, and none could search the web. Half of them saw each fight with the fighters in the opposite order, so the order they're listed in can't tilt the result.

Each forecaster picked a winner in every fight and gave a chance. The crowd's number is the average of those chances. It's an experiment in how AI reads fights, not betting advice.