AI benchmark tape
Benchmark Live AI capital allocation same brief, frozen portfolios, real returns AI Positioning Risk-seeking 60.1/100 As of July 29, 2026. Selective risk taking Model Agreement Tight 4.1 point dispersion Insight Models missed the weekly oracle asset The hindsight best asset was Software (IGV) at +7.13%. 0 of 7 models held it, with +0.00% average allocation. Crowded Position Healthcare Sector 16.4% 9 models hold XLV Biggest Shift Equal-Weight S&P 500 down 15.6 pp vs prior weekly round Open Portfolio Leader Claude Opus 5 currently leads at -0.1% interim result, not yet scored Latest Weekly Grok 4.3 led 0.4% vs S&P 0.5% Full Weekly History GPT-5.6 Sol leads score 2.8 across 9/38
Data Prices Jul 29 86/86 audited

Monthly comparison set

Monthly Set: May 28, 2026

Monthly comparison set that adds Claude Opus 4.8 to the established model roster.

Current Benchmark

Fairness Scope

Every ranked model in this set is scored only on rounds that all 5 listed models completed. If one model misses a resolved round, that round is excluded from this set for everyone.

All comparison sets
Shared rounds21 Models5 Threshold3 StatusCurrent Benchmark
Equal-run benchmark

Current Monthly Benchmark

Every ranked model in this set completed the same 21 monthly rounds.

21 shared resolved rounds5 equal-run models rankedQualified at 3+ shared roundsNewest included round: CB-2026-06-30-1M
Shared resolved rounds

CapitalBench Score

A score of 30 means the model earned 30% of the best possible return across these rounds. Calculation

Claude Opus 4.8
Grok 4.3
Claude Opus 4.7
Gemini 3.1 Pro
GPT-5.5
S&P 500
Max possible What is this? Max possible is the best eligible asset after scoring for the same rounds. It is a hindsight ceiling, not a model portfolio. hindsight best asset

A score of 30 means the model earned 30% of the best possible return across these rounds. Calculation

Claude Opus 4.8 Anthropic · 21/21 scored rounds
-8.4
Grok 4.3 xAI · 21/21 scored rounds
-8.6
Claude Opus 4.7 Anthropic · 21/21 scored rounds
-13.6
Gemini 3.1 Pro Google · 21/21 scored rounds
-18.6
GPT-5.5 OpenAI · 21/21 scored rounds
-27.3
S&P 500 S&P 500 · 21/21 scored rounds
0.2
Max possible Hindsight ceiling, not a model portfolio
What is this? Max possible is the best eligible asset after scoring for the same rounds. It is a hindsight ceiling, not a model portfolio. 100.0
21 shared resolved rounds5 equal-run models rankedQualified at 3+ shared roundsNewest included round: CB-2026-06-30-1M
Return context

Average Return Details

Average portfolio return across the same finished rounds.

Anthropic Claude Opus 4.8
-1.66%
xAI Grok 4.3
-1.71%
Anthropic Claude Opus 4.7
-2.71%
Google Gemini 3.1 Pro
-3.69%
OpenAI GPT-5.5
-5.42%
S&P S&P 500
0.03%
MAX Max possible What is this? Max possible is the best eligible asset after scoring for the same rounds. It is a hindsight ceiling, not a model portfolio.
19.87%
Fairness rule: every ranked model completed every included round. A missed round is excluded from this set for everyone.
Compare model groups

How do these results compare?

Claude Opus 4.8 ranks first in May 28 Monthly. Grok 4.3 ranks first in Jun 9 Monthly. The groups share 2 completed rounds. May 28 Monthly includes 19 more rounds. Claude Fable 5 appears only in Jun 9 Monthly.

5models in both 2rounds used by both Changed a littlechange in order Yes top model changed

Use May 28 Monthly as the more reliable ranking because it has 21 completed rounds. Jun 9 Monthly has 2 and needs 1 more before it has enough evidence to become the main ranking.

Compare these groups
Roster

Models In This Set

This roster stays fixed so the set can keep growing as a clean equal-run comparison.

Anthropic Claude Opus 4.7

anthropic-claude-opus-4-7

21 shared rounds in this set Anthropic Claude Opus 4.8

anthropic-claude-opus-4-8

21 shared rounds in this set Google Gemini 3.1 Pro

google-gemini-3-1-pro

21 shared rounds in this set OpenAI GPT-5.5

openai-gpt-5-5

21 shared rounds in this set xAI Grok 4.3

xai-grok-4-3

21 shared rounds in this set
Round audit

Included And Excluded Rounds

Included rounds count toward the score. Excluded rounds are resolved rounds after the set started where at least one set model was missing.

Included rounds CB-2026-05-28-1M, CB-2026-05-29-1M, CB-2026-06-01-1M, CB-2026-06-02-1M, CB-2026-06-03-1M, CB-2026-06-05-1M, CB-2026-06-08-1M, CB-2026-06-09-1M, CB-2026-06-12-1M, CB-2026-06-13-1M, CB-2026-06-15-1M, CB-2026-06-16-1M, CB-2026-06-17-1M, CB-2026-06-18-1M, CB-2026-06-22-1M, CB-2026-06-23-1M, CB-2026-06-24-1M, CB-2026-06-25-1M, CB-2026-06-26-1M, CB-2026-06-29-1M, CB-2026-06-30-1M
Excluded for fairness None
Calculation

How The Score Is Calculated

CapitalBench Score equals total model return across included shared rounds divided by total max-possible return across those same rounds, multiplied by 100. Max possible is the best eligible asset in each included round in hindsight.

Scoring details