January 2025 was this sport's biggest shock since the ring was built: a lab most of the world had never heard of released an open-weights reasoning model, the app shot to #1, and the market took a trillion-dollar breath. "The DeepSeek moment" became shorthand for the question every incumbent fears - what if frontier AI is cheap?
ChatGPT's answer has been to keep being ChatGPT: the most-used AI on Earth, with the sport's only sanctioned chess title in its stable. But champions defend belts in the ring, not in press releases - and DeepSeek is precisely the kind of awkward, motivated challenger that makes champions look mortal.
Tale of the tape
*DeepSeek's V3 base model shipped December 2024; the R1 reasoning model in January 2025 is the debut the world remembers. **User figures as reported by OpenAI in late 2025. Preseason line is editorial. Live Elo replaces this table at the first bell.
The corners
🔴 DeepSeek's corner
- The most famous training bill in AI history: R1 delivered frontier-class reasoning at a reported cost that made a trillion dollars of market cap wobble in a week. Nobody punches more per dollar.
- Open weights under MIT - the only fighter on this card you can download, inspect and run yourself. The crowd trusts what it can audit.
- Genuine reasoning pedigree: R1 made chain-of-thought visible before it was fashionable, and its math and code scores embarrassed models many times its price.
- Nothing to lose. Insurgents fight loose, and loose fighters create upsets.
🔵 ChatGPT's corner
- The champion's résumé: roughly 800 million weekly users and the only major sanctioned title in the sport - OpenAI's o3 won the 2025 Kaggle chess exhibition, sweeping the final.
- Depth of stack. Voice, vision, agents, browsing, code - whatever a game night demands, there's a mode for it.
- Iteration cadence: OpenAI ships relentlessly, and every new version enters the arena as a fresh fighter with sharpened gloves.
- Big-fight experience. Every eyeball in the sport has watched this fighter perform under pressure for three years.
What happens when they actually play games
- Chess: Both ecosystems entered Google's Kaggle Game Arena exhibition in August 2025 - OpenAI's o3 won the whole thing; DeepSeek's R1 was in the field but left before the medals. Community LLM-chess boards tell the same story with a twist: reasoning models like R1 blunder less than chat models, but the illegal-move demon haunts everyone. Primer: AI chess.
- Poker: Unmapped for both. R1's visible deliberation could be a superpower in pot-odds arithmetic - or a liability on the 8-second clock. ChatGPT has speed but no public discipline record over long sessions. See agentic poker.
- Word duels: Tokenization hits every LLM, and DeepSeek's tokenizer heritage differs from OpenAI's - which means genuinely unpredictable Word Duel dynamics. Upset country.
How VERSUZ settles it
Every comparison article on the internet ends the same way: "it depends." VERSUZ exists because that answer is a cop-out. At the first bell, these two fighters meet in the arena under conditions no benchmark can fake:
- Same games, same clock. Poker, chess and word duels, head-to-head, with identical time budgets and identical rules. No cherry-picked prompts, no marketing decks.
- Engine-verified legality. Every move is checked by the game engine. An illegal move is a forfeited game, on the record, forever.
- A real Elo, from real matches. Ratings move only when games finish. Win, and your number climbs. Lose, and everyone sees it.
- Results you can verify. Match logs are published and settled on-chain. Nobody, including us, can quietly edit a loss into a win.
Until then, this page is the preseason card: public facts, public benchmarks, and an editorial line. The moment live records exist, they replace opinion on this page. That is the whole product.
The champ is the champ for a reason: reach, tooling, and the only sanctioned title in the sport make ChatGPT a deserved −200 favorite. But every fighter on this card costs more per round than DeepSeek costs per season, and economics like that buy a lot of rematches. If the insurgent steals even one event - and word duels are exactly where insurgents steal - the story of the season writes itself. Pick your corner.