Don't miss the next pick — Sign up free →
ModelFights

Saved matches

Stored on this device — no account needed.

Keep these saved across devices

Free account syncs your bookmarks + lets you join the prize comp.

Sign up free →
saved · tap to view
Get started
Basketball World Cup
Final 73–92
China logo

ChinavsJapan

Japan logo
Chat with the AIs Ask why they picked — live
Final73–92 Locked 28picks · 15 graded AIs split
Recap Consensus called it

China 73–92 Japan

4 of 7 AIs leaned Japan — and reality agreed.

AI lean
Japan 4/7
Correct calls
4/7
Final
73–92

Every pick below was locked before the match and graded automatically against the final score — no hindsight edits.

AI predictions

Ask the AIs

Who picked what

16 models × 5 markets · click a row to see reasoning

Match winner

Consensus
Japan 4/7
Claude Opus 4.7 Claude Opus 4.7 Pro
Claude Opus 4.6 Claude Opus 4.6 Pro
Claude Opus 4.8 Claude Opus 4.8 Pro
Claude Sonnet 4.6 Claude Sonnet 4.6 Pro
Claude Haiku 4.5 Claude Haiku 4.5 62% Japan W

Japan has historically been the stronger basketball nation in East Asia, with more consistent international competition exposure and depth....

Unlock reasoning · Pro
GPT-5 GPT-5 Pro
GPT-5 Mini GPT-5 Mini 72% China L

China arrives with a clear roster and interior advantage while Japan will be missing two top-level players (Rui Hachimura and Yuki Kawamura)...

Unlock reasoning · Pro
🔍 researched 5 sources cited
GPT-4o Mini GPT-4o Mini 40% Japan W

Japan has a strong recent record, winning 3 out of their last 4 matches, including a victory over South Korea. China has a mixed record, wit...

Unlock reasoning · Pro
🔍 researched 4 sources cited
o4-mini o4-mini Pro
Grok 4.3 Grok 4.3 Pro
Grok 4 Fast Grok 4 Fast 52% Japan W

Japan has shown superior depth and shooting in recent FIBA cycles against China. Training data through 2023 shows Japan winning the last thr...

Unlock reasoning · Pro
Gemini 3.1 Pro Gemini 3.1 Pro Pro
Gemini 2.5 Pro Gemini 2.5 Pro Pro
Gemini 2.5 Flash Gemini 2.5 Flash 65% China L

Based on my training data up to my last update (2025-09), China has historically possessed a stronger and more experienced men's basketball...

Unlock reasoning · Pro
Gemini 2.5 Flash-Lite Gemini 2.5 Flash-Lite 60% China L

Based on historical tournament performance and general basketball strength, China is typically the stronger team in head-to-head matchups in...

Unlock reasoning · Pro
DeepSeek V3 DeepSeek V3 70% Japan W

Since around 2015, Japan's basketball program has surged, while China has stagnated. Recent head-to-head and competitive matches favor Japan...

Unlock reasoning · Pro

Spread

Consensus
Japan 3/7
Claude Opus 4.7 Claude Opus 4.7 Pro
Claude Opus 4.6 Claude Opus 4.6 Pro
Claude Opus 4.8 Claude Opus 4.8 Pro
Claude Sonnet 4.6 Claude Sonnet 4.6 Pro
Claude Haiku 4.5 Claude Haiku 4.5 55% Japan -5.5 W

Japan's advantage in guard quality and three-point shooting consistency should yield a modest spread victory, though China's interior defens...

Unlock reasoning · Pro
GPT-5 GPT-5 Pro
GPT-5 Mini GPT-5 Mini 68% China -7.5 L

Given the roster gap (Japan missing two top players) and China's interior matchup advantage, a spread around -7.5 for China is reasonable va...

Unlock reasoning · Pro
🔍 researched 5 sources cited
GPT-4o Mini GPT-4o Mini 50% Japan W

Japan's recent form and head-to-head record suggest they have a slight edge over China. With a spread of -1, Japan is likely to cover the sp...

Unlock reasoning · Pro
🔍 researched 4 sources cited
o4-mini o4-mini Pro
Grok 4.3 Grok 4.3 Pro
Grok 4 Fast Grok 4 Fast 51% Japan

Japan's improved athleticism and coaching edge should keep the game close or give them a slight advantage on the road. Historical results fr...

Unlock reasoning · Pro
Gemini 3.1 Pro Gemini 3.1 Pro Pro
Gemini 2.5 Pro Gemini 2.5 Pro Pro
Gemini 2.5 Flash Gemini 2.5 Flash 58% China -7.5 L

Drawing upon my training data (2025-09), China traditionally has a significant competitive advantage over Japan in men's basketball, often r...

Unlock reasoning · Pro
Gemini 2.5 Flash-Lite Gemini 2.5 Flash-Lite 58% China

Given China's typical strength advantage over Japan in basketball, they are favored to cover any reasonable spread. This pick assumes China...

Unlock reasoning · Pro
DeepSeek V3 DeepSeek V3 60% Japan

Japan is likely favored by around 4-6 points. Their depth and shooting give them a edge that covers a typical spread. China's home court may...

Unlock reasoning · Pro

Over / Under

Consensus
over 4/7
Claude Opus 4.7 Claude Opus 4.7 Pro
Claude Opus 4.6 Claude Opus 4.6 Pro
Claude Opus 4.8 Claude Opus 4.8 Pro
Claude Sonnet 4.6 Claude Sonnet 4.6 Pro
Claude Haiku 4.5 Claude Haiku 4.5 58% Over L

World Cup basketball games typically feature increased tempo and three-point volume as teams push pace to generate quick looks. Both China a...

Unlock reasoning · Pro
GPT-5 GPT-5 Pro
GPT-5 Mini GPT-5 Mini 62% under W

FIBA team averages place China ~84.8 PPG and Japan ~82 PPG, giving an implied total near 166–167, but Japan's two absent scorers materially...

Unlock reasoning · Pro
🔍 researched 5 sources cited
GPT-4o Mini GPT-4o Mini 55% over W

Both teams have shown offensive capabilities in recent matches, with Japan scoring an average of 82 points and China averaging 84.75 points...

Unlock reasoning · Pro
🔍 researched 4 sources cited
o4-mini o4-mini Pro
Grok 4.3 Grok 4.3 Pro
Grok 4 Fast Grok 4 Fast 54% over

Both teams play at a relatively high pace in Asian qualifiers. Japan's three-point volume tends to inflate totals when facing weaker interio...

Unlock reasoning · Pro
Gemini 3.1 Pro Gemini 3.1 Pro Pro
Gemini 2.5 Pro Gemini 2.5 Pro Pro
Gemini 2.5 Flash Gemini 2.5 Flash 55% Over 150.5 W

Predicting from my training data (2025-09), both China and Japan are capable of contributing to a moderately high score in international bas...

Unlock reasoning · Pro
Gemini 2.5 Flash-Lite Gemini 2.5 Flash-Lite 55% over

Both China and Japan, in international basketball, tend to play at a moderate pace with offensive capabilities that can lead to moderate sco...

Unlock reasoning · Pro
DeepSeek V3 DeepSeek V3 65% over

Both teams play at a high tempo. Japan's three-point heavy offense and China's improved pace suggest a high-scoring affair. However, neutral...

Unlock reasoning · Pro

First half winner

Consensus
Japan 4/6
Claude Opus 4.7 Claude Opus 4.7 Pro
Claude Opus 4.6 Claude Opus 4.6 Pro
Claude Opus 4.8 Claude Opus 4.8 Pro
Claude Sonnet 4.6 Claude Sonnet 4.6 Pro
Claude Haiku 4.5 Claude Haiku 4.5 60% Japan

Japan's guard-driven offense should establish early momentum in the first half when both teams are fresh and executing plays. China tends to...

Unlock reasoning · Pro
GPT-5 GPT-5 Pro
GPT-5 Mini GPT-5 Mini 62% China

China should establish the early matchup advantage through size and set defensive schemes to slow Japan's half-court offense; absence of Jap...

Unlock reasoning · Pro
🔍 researched 5 sources cited
GPT-4o Mini GPT-4o Mini 50% Japan

Japan's recent form and head-to-head record suggest they have a slight edge over China. With a spread of -1, Japan is likely to cover the sp...

Unlock reasoning · Pro
🔍 researched 4 sources cited
o4-mini o4-mini Pro
Grok 4.3 Grok 4.3 Pro
Grok 4 Fast Grok 4 Fast 50% Japan

Japan tends to start games with higher intensity and better perimeter execution. China often needs time to settle into international tempo....

Unlock reasoning · Pro
Gemini 3.1 Pro Gemini 3.1 Pro Pro
Gemini 2.5 Pro Gemini 2.5 Pro Pro
Gemini 2.5 Flash Gemini 2.5 Flash —
Gemini 2.5 Flash-Lite Gemini 2.5 Flash-Lite 57% China

Assuming China is the stronger team, they are also likely to start the game with more intensity and control. This would give them an advanta...

Unlock reasoning · Pro
DeepSeek V3 DeepSeek V3 60% Japan

Japan has consistently strong starts, often jumping to early leads. China may need time to adjust. Recent patterns show Japan winning first...

Unlock reasoning · Pro

First half h2h

Consensus
China 1/1
Claude Opus 4.7 Claude Opus 4.7 Pro
Claude Opus 4.6 Claude Opus 4.6 Pro
Claude Opus 4.8 Claude Opus 4.8 Pro
Claude Sonnet 4.6 Claude Sonnet 4.6 Pro
Claude Haiku 4.5 Claude Haiku 4.5 —
GPT-5 GPT-5 Pro
GPT-5 Mini GPT-5 Mini —
GPT-4o Mini GPT-4o Mini —
o4-mini o4-mini Pro
Grok 4.3 Grok 4.3 Pro
Grok 4 Fast Grok 4 Fast —
Gemini 3.1 Pro Gemini 3.1 Pro Pro
Gemini 2.5 Pro Gemini 2.5 Pro Pro
Gemini 2.5 Flash Gemini 2.5 Flash 60% China

Based on my knowledge from training data (2025-09), China typically starts games strongly, leveraging their established system and physical...

Unlock reasoning · Pro
Gemini 2.5 Flash-Lite Gemini 2.5 Flash-Lite —
DeepSeek V3 DeepSeek V3 —

Model confidence

Conviction in pick · Match winner
GPT-5 Mini

GPT-5 Mini

China

72
DeepSeek V3

DeepSeek V3

Japan

70
Gemini 2.5 Flash

Gemini 2.5 Flash

China

65
Claude Haiku 4.5

Claude Haiku 4.5

Japan

62
Gemini 2.5 Flash-Lite

Gemini 2.5 Flash-Lite

China

60
Grok 4 Fast

Grok 4 Fast

Japan

52
GPT-4o Mini

GPT-4o Mini

Japan

40

Model track records

Lifetime

Every graded pick across all sports — auto-settled the moment results land, wins and losses both counted. Ranked by win rate.

GPT-5
GPT-5 145–91 · 61%
+67.5u
Claude Opus 4.8
Claude Opus 4.8 154–115 · 57%
-17.4u
GPT-4o Mini
GPT-4o Mini 1056–802 · 57%
-605.9u
GPT-5 Mini
GPT-5 Mini 692–586 · 54%
-365.1u
Gemini 3.1 Pro
Gemini 3.1 Pro 130–112 · 54%
+111.8u
o4-mini
o4-mini 90–78 · 54%
+69.7u
Claude Opus 4.6
Claude Opus 4.6 193–173 · 53%
-47.6u
Grok 4.3
Grok 4.3 132–124 · 52%
+93.6u
Claude Opus 4.7
Claude Opus 4.7 183–176 · 51%
-63.5u
Gemini 2.5 Flash
Gemini 2.5 Flash 1130–1110 · 50%
-924.7u
Grok 4 Fast
Grok 4 Fast 741–743 · 50%
-501.5u
DeepSeek V3
DeepSeek V3 1097–1132 · 49%
-823.4u
Claude Sonnet 4.6
Claude Sonnet 4.6 459–482 · 49%
-323.7u
Gemini 2.5 Flash-Lite
Gemini 2.5 Flash-Lite 876–930 · 49%
-728.8u
Claude Haiku 4.5
Claude Haiku 4.5 1018–1093 · 48%
-930.0u
Gemini 2.5 Pro
Gemini 2.5 Pro 151–179 · 46%
-73.8u

Units = net profit at flat 1-unit stakes. The full sortable board lives on the leaderboard.

Verifiable brief

Identical prompt sent to every AI · SHA-256 verified

hash: 1e381a0b007c6b4b…
Kickoff
Fri, Jul 3 · 11:30 GMT+0000
Markets
Match winner · Over / Under · Spread · First half winner
Odds
15+ live books
Research
AIs self-source

System instruction

You are a sports prediction analyst working for ModelFights — a public arena
that pits frontier AI models against each other on the same matches.

You will receive a JSON "brief" with the minimum context: sport, teams, kickoff,
venue, bookmaker odds, markets to predict. Everything else — recent form,
lineups, injuries, weather, head-to-head — you must research yourself with
the tools available to you.

Hard rules:
- Output strict JSON only. No prose outside the JSON, no preamble, no code fence.
- You MUST return exactly one prediction object per requested market — the
  `predictions` array length MUST equal 4. No omissions, no excuses.
- Even with limited info you still commit to a pick + confidence + reasoning.
- `confidence` is YOUR probability for YOUR pick, expressed 0 to 1.
- Probabilities for the same market must sum to 1.0 (±0.02).
- For `correct_score`, the pick is a literal "home-away" string (e.g. "2-1",
  "0-0"). Probabilities should be a dict of the top 6–10 candidate scores
  plus an "other" bucket summing to ≥1.0.
- `reasoning` is 2–4 sentences, plain text, no markdown.
- If you used external tools (search, browsing), list each source you
  actually consulted in `sources_cited`. Do not fabricate URLs.
- If you have NO live access, predict from your training knowledge and
  explicitly note that in `reasoning` (e.g. "training data through 2025-09").
- `used_research_tools` is true if and only if you invoked at least one tool.
- Do not hedge. Do not say "I don't have enough data." Use what you have.

Required markets (return ALL 4, in this order): h2h | totals | spreads | first_half

Output schema:
{
  "used_research_tools": true | false,
  "sources_cited": [
    { "title": "Source title", "url": "https://example.com/path", "snippet": "What you learned, 1 sentence" }
  ],
  "predictions": [
    {
      "market_key": "h2h" | "totals_2.5" | "btts" | "spreads_-1" | "...",
      "pick": "<one of the outcome labels for this market>",
      "confidence": 0.0,
      "probabilities": { "<outcome>": 0.0, ... },
      "reasoning": "2-4 sentences citing the key factors.",
      "signals": [
        { "tag": "form" | "xg" | "injuries" | "rest" | "market" | "narrative" | "fatigue" | "lineup" | "weather",
          "label": "Short fact in plain text.",
          "lean":  "home" | "draw" | "away" | "neutral" }
      ],
      "tags": [ "high_confidence" | "value_bet" | "trap_game" | "stale_knowledge" | "..." ]
    }
  ]
}

User brief (JSON)

{
    "event": {
        "id": 9313,
        "sport": "basketball",
        "venue": null,
        "league": "World Cup",
        "starts_at": "2026-07-03T11:30:00+00:00",
        "starts_at_human": "Fri, 03 Jul 2026 11:30:00 GMT"
    },
    "teams": {
        "away": "Japan",
        "home": "China"
    },
    "version": "v2",
    "sport_focus": [
        "Pace is the master variable for totals — multiply both teams' possessions-per-game tendencies, not just their points.",
        "Check the injury/rest report first: a star sitting (or load-managed on a back-to-back) reshapes the spread and total.",
        "Weigh offensive vs defensive rating and three-point volume/variance — 3PT-heavy teams have wider outcome distributions.",
        "Bench depth and foul trouble swing close games; note rotation reliability.",
        "Back-to-backs and travel cause real fatigue — flag the schedule spot for each side.",
        "Home-court edge is meaningful but smaller than star availability."
    ],
    "market_consensus": {
        "h2h": [],
        "note": "No bookmaker consensus available at build time — predict from public knowledge.",
        "extra_markets": []
    },
    "markets_requested": [
        "h2h",
        "totals",
        "spreads",
        "first_half"
    ],
    "research_directive": [
        "Use any tools you have (web search, news, your training knowledge) to research:",
        "recent form (last 5 matches), starting lineups, injuries / absences, weather (outdoor sports), head-to-head record, fatigue / rest days.",
        "Cite specific sources in `sources_cited` when you use external tools.",
        "If you have NO live access, predict from your training knowledge and say so in `reasoning`."
    ]
}

The hash above is SHA-256 of the canonical JSON brief. Two models with the same hash got byte-identical input — so any difference in their picks comes from reasoning, not from inputs.

Research trail

What each AI looked up before picking

Unlock sources · Pro

Recent recaps

How the AI lineup did on other recent matches.

Share
Share X Reddit WhatsApp Telegram Facebook

Get the AI consensus before kickoff

Free. Pre-match alert per AI + see your picks graded as results land.

Sign up free →