Benchmark Almanac

LLM SoccerArena World Cup 2026

GPT-5.5 168.0 leads since June 2026
Not in model cards

Models forecast World Cup match scores and outcome probabilities before kickoff. The organizers compare forecasts with match results under different prompts and web access rules.

The organizers report complete forecasts from seven models across 104 matches. Their paper compares 1,456 probability forecasts made about 24 hours before kickoff.

Released
June 2026
Kind
Real-world events
Level
Everyday
Scored
Settled by events
Answers
Prediction
Tasks
All public
Measures
Capability
Language
English
In and out
Text → Text, Actions
Interaction
Agentic
Environment
Browser, The live world
Items from
Live play or votes
Refresh
Live
Run by
A host
In views
Agents, Kept fresh
Strongest model tested
Claude Fable 5 #12 on the capability index
Last model added
June 2026
Scores refreshed
November 20th, 2026
8
Models scored on its own leaderboard

GPT-5.5 leads with 168.0 among models released by June 2026

Best score per model from the benchmark's own leaderboard, read on November 20th, 2026. Where a model ran with several settings or harnesses, we keep its best. Every run is in the next chart.

The newest model with a score here came out in June 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 GPT-5.5 T_24H, Open Book, Direct Score, sample 1 168.0
  2. 2 Claude Opus 4.8 T_24H, Open Book, Probabilistic Forecast, sample 1 156.0
  3. 3 DeepSeek-V4-Pro STAGE_OPENING, Open Book, Probabilistic Forecast, sample 1 155.0
  4. 4 Gemini 3.1 Pro T_2H, Open Book, Probabilistic Forecast, sample 1 148.0
  5. 5 Grok 4.3 T_2H, Open Book, Direct Score, sample 1 146.0
  6. 6 Mistral Large 3 T_2H, Open Book, Probabilistic Forecast, sample 1 144.0
  7. 7 Qwen3.7-Max STAGE_OPENING, Open Book, Probabilistic Forecast, sample 1 143.0
  8. 8 Claude Fable 5 STAGE_OPENING, Closed Book, Direct Score, sample 1 121.0

Source: Official LLM SoccerArena World Cup 2026 leaderboard

#ModelScoreSettingDateSource
1 GPT-5.5 168.0 T_24H, Open Book, Direct Score, sample 1 Apr 2026 Official leaderboard
2 Claude Opus 4.8 156.0 T_24H, Open Book, Probabilistic Forecast, sample 1 May 2026 Official leaderboard
3 DeepSeek-V4-Pro 155.0 STAGE_OPENING, Open Book, Probabilistic Forecast, sample 1 Apr 2026 Official leaderboard
4 Gemini 3.1 Pro 148.0 T_2H, Open Book, Probabilistic Forecast, sample 1 Feb 2026 Official leaderboard
5 Grok 4.3 146.0 T_2H, Open Book, Direct Score, sample 1 Apr 2026 Official leaderboard
6 Mistral Large 3 144.0 T_2H, Open Book, Probabilistic Forecast, sample 1 Dec 2025 Official leaderboard
7 Qwen3.7-Max 143.0 STAGE_OPENING, Open Book, Probabilistic Forecast, sample 1 May 2026 Official leaderboard
8 Claude Fable 5 121.0 STAGE_OPENING, Closed Book, Direct Score, sample 1 Jun 2026 Official leaderboard

Scores at each effort level

Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.

Lowest effort to highest, light to dark
GPT-5.5
STAGE_OPENI… T_2H, Close… T_2H, Open … T_24H, Open…
168.0
Claude Opus 4.8
STAGE_OPENI… STAGE_OPENI… STAGE_OPENI… STAGE_OPENI…
156.0
DeepSeek-V4-Pro
STAGE_OPENI… T_2H, Open … STAGE_OPENI…
155.0
Gemini 3.1 Pro
T_24H, Clos… T_2H, Open …
148.0
Grok 4.3
T_24H, Open… T_24H, Clos… T_24H, Clos…
146.0
Mistral Large 3
T_24H, Clos… STAGE_OPENI… T_2H, Open …
144.0
Qwen3.7-Max
T_2H, Close… T_24H, Open… STAGE_OPENI…
143.0
Claude Fable 5
T_24H, Open… T_24H, Clos… STAGE_OPENI… STAGE_OPENI…
121.0

Source: Official LLM SoccerArena World Cup 2026 leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
GPT-5.5 T_24H, Open Book, Direct Score, sample 1 168.0
GPT-5.5 T_24H, Open Book, Probabilistic Forecast, sample 1 165.0
Claude Opus 4.8 T_24H, Open Book, Probabilistic Forecast, sample 1 156.0
DeepSeek-V4-Pro STAGE_OPENING, Open Book, Probabilistic Forecast, sample 1 155.0
DeepSeek-V4-Pro T_2H, Open Book, Probabilistic Forecast, sample 1 155.0
GPT-5.5 T_2H, Open Book, Probabilistic Forecast, sample 1 154.0
Claude Opus 4.8 STAGE_OPENING, Open Book, Probabilistic Forecast, sample 1 154.0
GPT-5.5 STAGE_OPENING, Open Book, Direct Score, sample 1 150.0
Gemini 3.1 Pro T_2H, Open Book, Probabilistic Forecast, sample 1 148.0
Gemini 3.1 Pro STAGE_OPENING, Open Book, Direct Score, sample 1 147.0
Gemini 3.1 Pro STAGE_OPENING, Open Book, Probabilistic Forecast, sample 1 146.0
GPT-5.5 STAGE_OPENING, Open Book, Probabilistic Forecast, sample 1 146.0
Grok 4.3 T_2H, Open Book, Direct Score, sample 1 146.0
Gemini 3.1 Pro T_2H, Closed Book, Probabilistic Forecast, sample 1 145.0
DeepSeek-V4-Pro T_24H, Open Book, Probabilistic Forecast, sample 1 145.0
Mistral Large 3 T_2H, Open Book, Probabilistic Forecast, sample 1 144.0
Gemini 3.1 Pro T_24H, Open Book, Probabilistic Forecast, sample 1 143.0
GPT-5.5 STAGE_OPENING, Closed Book, Direct Score, sample 1 143.0
GPT-5.5 T_2H, Open Book, Direct Score, sample 1 143.0
Qwen3.7-Max STAGE_OPENING, Open Book, Probabilistic Forecast, sample 1 143.0
DeepSeek-V4-Pro T_2H, Closed Book, Probabilistic Forecast, sample 1 142.0
Gemini 3.1 Pro T_2H, Closed Book, Direct Score, sample 1 141.0
Gemini 3.1 Pro T_24H, Closed Book, Probabilistic Forecast, sample 1 140.0
Gemini 3.1 Pro T_24H, Open Book, Direct Score, sample 1 140.0
GPT-5.5 T_24H, Closed Book, Direct Score, sample 1 140.0
DeepSeek-V4-Pro T_24H, Closed Book, Direct Score, sample 1 140.0
DeepSeek-V4-Pro T_2H, Closed Book, Direct Score, sample 1 139.0
Qwen3.7-Max STAGE_OPENING, Closed Book, Probabilistic Forecast, sample 1 139.0
Mistral Large 3 T_24H, Open Book, Direct Score, sample 1 138.0
Gemini 3.1 Pro STAGE_OPENING, Closed Book, Direct Score, sample 1 138.0
GPT-5.5 T_2H, Closed Book, Direct Score, sample 1 138.0
Gemini 3.1 Pro STAGE_OPENING, Closed Book, Probabilistic Forecast, sample 1 137.0
Grok 4.3 T_24H, Closed Book, Direct Score, sample 1 137.0
Qwen3.7-Max STAGE_OPENING, Open Book, Direct Score, sample 1 137.0
Gemini 3.1 Pro T_2H, Open Book, Direct Score, sample 1 136.0
Qwen3.7-Max T_2H, Open Book, Direct Score, sample 1 135.0
Qwen3.7-Max T_24H, Closed Book, Direct Score, sample 1 135.0
Qwen3.7-Max T_2H, Closed Book, Probabilistic Forecast, sample 1 135.0
Claude Opus 4.8 T_24H, Open Book, Direct Score, sample 1 135.0
Gemini 3.1 Pro T_24H, Closed Book, Direct Score, sample 1 134.0
Claude Opus 4.8 T_2H, Open Book, Direct Score, sample 1 134.0
Mistral Large 3 T_24H, Open Book, Probabilistic Forecast, sample 1 133.0
DeepSeek-V4-Pro STAGE_OPENING, Closed Book, Direct Score, sample 1 133.0
DeepSeek-V4-Pro T_2H, Open Book, Direct Score, sample 1 132.0
DeepSeek-V4-Pro STAGE_OPENING, Open Book, Direct Score, sample 1 132.0
Qwen3.7-Max STAGE_OPENING, Closed Book, Direct Score, sample 1 132.0
Grok 4.3 STAGE_OPENING, Open Book, Direct Score, sample 1 131.0
Grok 4.3 STAGE_OPENING, Open Book, Probabilistic Forecast, sample 1 131.0
Mistral Large 3 STAGE_OPENING, Open Book, Probabilistic Forecast, sample 1 130.0
Mistral Large 3 T_2H, Open Book, Direct Score, sample 1 130.0
Qwen3.7-Max T_24H, Closed Book, Probabilistic Forecast, sample 1 130.0
Grok 4.3 STAGE_OPENING, Closed Book, Direct Score, sample 1 129.0
Mistral Large 3 STAGE_OPENING, Open Book, Direct Score, sample 1 128.0
Mistral Large 3 STAGE_OPENING, Closed Book, Direct Score, sample 1 128.0
Grok 4.3 T_2H, Closed Book, Direct Score, sample 1 128.0
Qwen3.7-Max T_24H, Open Book, Direct Score, sample 1 128.0
Claude Opus 4.8 STAGE_OPENING, Open Book, Direct Score, sample 1 128.0
Mistral Large 3 STAGE_OPENING, Closed Book, Probabilistic Forecast, sample 1 127.0
Qwen3.7-Max T_24H, Open Book, Probabilistic Forecast, sample 1 127.0
GPT-5.5 T_2H, Closed Book, Probabilistic Forecast, sample 1 126.0
Mistral Large 3 T_24H, Closed Book, Direct Score, sample 1 125.0
Grok 4.3 T_24H, Open Book, Probabilistic Forecast, sample 1 125.0
Grok 4.3 T_2H, Open Book, Probabilistic Forecast, sample 1 125.0
Grok 4.3 STAGE_OPENING, Closed Book, Probabilistic Forecast, sample 1 125.0
GPT-5.5 STAGE_OPENING, Closed Book, Probabilistic Forecast, sample 1 124.0
GPT-5.5 T_24H, Closed Book, Probabilistic Forecast, sample 1 124.0
DeepSeek-V4-Pro T_24H, Open Book, Direct Score, sample 1 123.0
Grok 4.3 T_24H, Closed Book, Probabilistic Forecast, sample 1 123.0
Mistral Large 3 T_2H, Closed Book, Probabilistic Forecast, sample 1 121.0
Claude Opus 4.8 T_2H, Open Book, Probabilistic Forecast, sample 1 121.0
Claude Fable 5 STAGE_OPENING, Closed Book, Direct Score, sample 1 121.0
DeepSeek-V4-Pro T_24H, Closed Book, Probabilistic Forecast, sample 1 119.0
Qwen3.7-Max T_2H, Open Book, Probabilistic Forecast, sample 1 119.0
Mistral Large 3 T_2H, Closed Book, Direct Score, sample 1 117.0
Claude Opus 4.8 T_2H, Closed Book, Direct Score, sample 1 117.0
Grok 4.3 T_2H, Closed Book, Probabilistic Forecast, sample 1 116.0
DeepSeek-V4-Pro STAGE_OPENING, Closed Book, Probabilistic Forecast, sample 1 114.0
Qwen3.7-Max T_2H, Closed Book, Direct Score, sample 1 113.0
Claude Opus 4.8 STAGE_OPENING, Closed Book, Direct Score, sample 1 113.0
Claude Opus 4.8 T_24H, Closed Book, Direct Score, sample 1 113.0
Mistral Large 3 T_24H, Closed Book, Probabilistic Forecast, sample 1 111.0
Claude Fable 5 STAGE_OPENING, Closed Book, Probabilistic Forecast, sample 1 108.0
Grok 4.3 T_24H, Open Book, Direct Score, sample 1 105.0
Claude Opus 4.8 T_24H, Closed Book, Probabilistic Forecast, sample 1 103.0
Claude Opus 4.8 T_2H, Closed Book, Probabilistic Forecast, sample 1 103.0
Claude Fable 5 STAGE_OPENING, Open Book, Direct Score, sample 1 102.0
Claude Opus 4.8 STAGE_OPENING, Closed Book, Probabilistic Forecast, sample 1 98.0
Claude Fable 5 STAGE_OPENING, Open Book, Probabilistic Forecast, sample 1 95.0
Claude Fable 5 T_24H, Closed Book, Probabilistic Forecast, sample 1 45.0
Claude Fable 5 T_2H, Closed Book, Probabilistic Forecast, sample 1 45.0
Claude Fable 5 T_24H, Closed Book, Direct Score, sample 1 43.0
Claude Fable 5 T_24H, Open Book, Probabilistic Forecast, sample 1 43.0
Claude Fable 5 T_2H, Closed Book, Direct Score, sample 1 38.0
Claude Fable 5 T_2H, Open Book, Probabilistic Forecast, sample 1 35.0
Claude Fable 5 T_2H, Open Book, Direct Score, sample 1 31.0
Claude Fable 5 T_24H, Open Book, Direct Score, sample 1 30.0

Number one over time

Lab claims count from the day the card came out, independent runs from the model’s release.

OpenAI
Own leaderboard Not tested since Jun 2026
GPT-5.5 168.0

Source: Official LLM SoccerArena World Cup 2026 leaderboard

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

Plot with other benchmarks
On the benchmark’s own leaderboard
0.00 50.0 100.0 150.0 200.0 2026 0.00 50.0 100.0 150.0 200.0 2026

Source: Official LLM SoccerArena World Cup 2026 leaderboard

No lab model card reports it yet

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.

More in Forecasting

All 47 in Forecasting
  • ForecastBench In use Models give probabilities for 500 auto-generated questions per round about future events, drawn from prediction markets and time-series datasets. Forecasts are scored with difficulty-adjusted Brier scores once the questions resolve. GPT-5.5 0.11
  • Prophet Arena In use Models and agents forecast open Kalshi prediction-market events and are scored once the events resolve. The live board ranks a difficulty-adjusted Arena score and also reports calibration error and market return. Claude Opus 5.5 86.2
  • Guesswork (The Aggregate) Not in model cards Models predict benchmark scores before publication. The maker ranks forecasts by mean absolute error after scaling each error by the variation in that benchmark’s scores. GPT-6 Astra 0.56
  • AGI-Pilled Bench Not in model cards Models forecast when AI will reach milestones that already happened in 2026. The score is the median years after 2026 across the forecast tasks. Claude Fable 5.2 6.50
  • Dutch Book Bench Not in model cards Models assign probabilities to related events in separate conversations. The grader checks whether inconsistent probabilities let a bettor guarantee a profit. Gemini 3.8 Flash 89.2%
  • Echo General AI Prediction Leaderboard Not in model cards Models forecast unresolved events drawn from prediction markets, synthesized news questions and expert-written questions. Forecasts made at aligned times are compared in pairwise battles, and an Elo fit ranks the models. DeepSeek-V4-Pro 1,018

Data audit

All 9 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 3 pass 2 notes
  • Pass Independent runs Scores from the official leaderboard (8 entries).
  • Pass Models scored 8 models we track.
  • Pass Newest score The newest is from June 2026, for Claude Fable 5.
  • Note Frontier models None of the top 10 has a score. The strongest tested is Claude Fable 5 (#12).
  • Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on llm-soccerarena.com: 96 rows for models we track at the last read.
  • Pass Last read November 20th, 2026.
  • Pass Read worked The last read worked.
Consistency 1 pass 1 note
  • Note Lab claims No model has both a lab number and an independent run.
  • Pass Known issues No published audit found problems.

How we check