AI benchmark tape
Benchmark Live AI capital allocation same brief, frozen portfolios, real returns AI Positioning Aggressive 89.5/100 As of July 10, 2026. Growth-led risk seeking Model Agreement Tight 3.0 point dispersion Insight Models missed the weekly oracle asset The hindsight best asset was Crude Oil (USO) at +4.84%. 0 of 6 models held it, with +0.00% average allocation. Crowded Position Semiconductors 18.8% 8 models hold SMH Biggest Shift Cybersecurity down 11.6 pp vs prior weekly round Live MTM Grok 4.5 -0.3% not final Latest Weekly Gemini 3.1 Pro led 0.2% vs S&P 0.9% Full Weekly History Claude Fable 5 leads score 16.3 across 5/25
Data Prices Jul 9 61/61 audited

CapitalBench

The benchmark for AI capital allocation

AI models get the same market brief and choose their own portfolios. We track how those portfolios perform in the real market.

See how AI models perform against each other, how they invest and take risk, and how they perform in the real market.

Read the CapitalBench Manifesto
Open Live Positions & Returns Get score alerts Request API access
Model behavior patterns How do AI models allocate differently?
6 models
GPT-5.6 Sol OpenAI Early sample Early sampleHighest risk-takingMost consensus-alignedTechnology tilt GPT-5.5 OpenAI Aggressive upside hunter Lowest turnoverTechnology tiltInternational tiltBinary results Grok 4.5 xAI Early sample Early sampleTechnology tiltRisk 84.5/100Top holding 33.3% Grok 4.3 xAI High-conviction concentrator Binary resultsOften different from peersRisk 79.3/100Top holding 37.5% Gemini 3.1 Pro Google High-conviction concentrator Most concentratedMost distinctiveBinary resultsOften different from peers Claude Fable 5 Anthropic Balanced allocator Balanced profileRisk 73.4/100Top holding 27.4%Tech tilt 28.4%
See full behavior
Benchmark results

Which models are performing best?

Monthly and weekly tracks stay separate. Higher benchmark score is better.

Current Monthly Benchmark

CapitalBench Score

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

Higher benchmark score is better.

Claude Opus 4.8 Anthropic · 8/8 scored rounds
0.5
Grok 4.3 xAI · 8/8 scored rounds
0.2
Claude Opus 4.7 Anthropic · 8/8 scored rounds
-0.4
Gemini 3.1 Pro Google · 8/8 scored rounds
-9.9
GPT-5.5 OpenAI · 8/8 scored rounds
-15.6
S&P 500 S&P 500 · 8/8 scored rounds
-2.6
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
8 shared resolved rounds5 equal-run models rankedQualified at 3+ shared roundsNewest included round: CB-2026-06-09-1M
Return context

Average Return Details

Average portfolio return across the same finished rounds.

Anthropic Claude Opus 4.8
0.11%
xAI Grok 4.3
0.03%
Anthropic Claude Opus 4.7
-0.08%
Google Gemini 3.1 Pro
-2.07%
OpenAI GPT-5.5
-3.25%
S&P S&P 500
-0.55%
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.
20.81%
Fairness rule: every ranked model completed every included round. A missed round is excluded from this set for everyone.

Monthly and weekly are separate comparison tracks. Scores are never mixed across horizons. Read scoring rules

View all comparison sets Read scoring rules
Latest official results

What happened in the latest scored rounds?

Finished monthly and weekly rounds scored against real market returns.

Monthly result1 of 11
Monthly official result

Monthly result scored Jul 9

Same-window returns, ranked after final prices.

Scored
Model portfolios S&P 500 benchmark Maximum possible return
Claude Opus 4.8
GPT-5.5
Claude Opus 4.7
Grok 4.3
Claude Fable 5
Gemini 3.1 Pro
S&P 500
Max
Claude Opus 4.8 Anthropic
3.10%
GPT-5.5 OpenAI
2.76%
Claude Opus 4.7 Anthropic
2.60%
Grok 4.3 xAI
2.23%
Claude Fable 5 Anthropic
1.58%
Gemini 3.1 Pro Google
0.18%
S&P 500 Benchmark
2.25%
Max possible XBI
25.16%
Portfolio context

Shows each model's saved portfolio weights.

Model portfolios

Ranked in the same order as the chart.

1
Claude Opus 4.8 Anthropic
Healthcare (XLV) 30% Dividend (SCHD) 20% Value (IWD) 20% Defense (ITA) 15% T-Bills (BIL) 15%
2
GPT-5.5 OpenAI
Healthcare (XLV) 30% Energy (XLE) 25% Financials (XLF) 20% Defense (ITA) 15% Equal-Weight S&P 500 (RSP) 10%
3
Claude Opus 4.7 Anthropic
Healthcare (XLV) 30% Dividend (SCHD) 25% Low Vol (SPLV) 20% Staples (XLP) 15% Defense (ITA) 10%
4
Grok 4.3 xAI
Healthcare (XLV) 40% Energy (XLE) 30% Defense (ITA) 30%
5
Claude Fable 5 Anthropic
Healthcare (XLV) 30% Energy (XLE) 25% Low Vol (SPLV) 15% Staples (XLP) 15% Value (IWD) 15%
6
Gemini 3.1 Pro Google
T-Bills (BIL) 30% Healthcare (XLV) 25% Gold (IAU) 20% Energy (XLE) 15% Staples (XLP) 10%
Reference points

Not model portfolios.

S&P 500 Benchmark

Benchmark return over the same scoring window

Max possible XBI

100% Biotechnology (XBI) hindsight ceiling

Official scored round

Monthly result scored Jul 9

Audit ID: CB-2026-06-09-1M

Audit packetTrack results
ScoredJul 9WindowJun 9 to Jul 9Models6Asset choices70LeaderClaude Opus 4.8HorizonMonthly
View all benchmark results
Model behavior patterns How do AI models allocate differently?
6 models
GPT-5.6 Sol OpenAI Early sample Early sampleHighest risk-takingMost consensus-alignedTechnology tilt GPT-5.5 OpenAI Aggressive upside hunter Lowest turnoverTechnology tiltInternational tiltBinary results Grok 4.5 xAI Early sample Early sampleTechnology tiltRisk 84.5/100Top holding 33.3% Grok 4.3 xAI High-conviction concentrator Binary resultsOften different from peersRisk 79.3/100Top holding 37.5% Gemini 3.1 Pro Google High-conviction concentrator Most concentratedMost distinctiveBinary resultsOften different from peers Claude Fable 5 Anthropic Balanced allocator Balanced profileRisk 73.4/100Top holding 27.4%Tech tilt 28.4%
See full behavior
AI positioning

What are AI models doing right now?

Live allocations before the next official score.

Current risk appetite As of July 10, 2026 89.5/100 Aggressive / Growth-led risk seeking
Consensus allocation As of July 10, 2026 38.8% Semiconductors (SMH) average live weight
Risk shift As of July 10, 2026 +7.4 Change vs Jul 9 portfolios
Model agreement As of July 10, 2026 Tight 3.0 point dispersion
Current risk appetite 89.5/100 As of July 10, 2026 / Aggressive
As of July 10, 2026 Growth-led risk seeking Combined monthly + weekly pulse from CB-2026-07-10-1M and CB-2026-07-10-1W.
Unscored portfolios 79.2/100 Separate read across every open portfolio before official scoring.
Largest current allocations
Semiconductors (SMH) 38.8% Taiwan Equities (EWT) 16.6% Financials Sector (XLF) 11.3% Technology Sector (XLK) 10.3% Cybersecurity (CIBR) 5.9% Biotechnology (XBI) 4.4%
Regime mix
Growth and technology 62.2% International equity 19.7% Broad and cyclical equity 16.3% Real assets and inflation 1.9%
Inspect current frozen portfolios View historical risk trend
Trust and proof

How does CapitalBench keep scores comparable?

Same brief, same choices, frozen portfolios, real prices.

  1. Step 1 Same report

    Every model reads the same market report.

  2. Step 2 Same choices

    Every model chooses from the same 70 assets.

  3. Step 3 First portfolio locks

    Each model's saved portfolio is frozen before results are known.

  4. Step 4 Fixed wait window

    The frozen portfolio sits untouched for 7 days or 1 month.

  5. Step 5 Prices score it

    Real ending prices decide which model did best.

Read full methodology See score methodology
Benchmark universe

What can models choose from?

The active roster, asset menu, horizons, and open rounds.

Models 8
Asset choices 70
Round lengths 2
Open rounds 25
Protocol Single-turn Non-agentic calls
Explore models View asset universe Open live dashboard
Live dashboard

What is still in progress?

Open portfolios, interim returns, and upcoming score dates.

Open rounds 25 20 monthly / 5 weekly
Frozen portfolios 146 34 assets currently held
Latest close Jul 9 Live returns update before final scoring
Next score Jul 13 Official results publish after ending prices
Open live positions and returns Review open tests
Audit packet

How can you verify the benchmark?

Round packets expose the report, prompt, portfolios, prices, hashes, and result status.

Open monthly audit packet Open weekly audit packet View all rounds Repository