This post by Eric Weiss inspired me to try it out. After all, I have plaid around with ppq.ai - pay per query before.
Using this script:
#!/bin/bash
models=(gpt-4o grok-2 qwq-32b-preview deepseek-r1 gemini-2.0-flash-exp dolphin-mixtral-8x22b claude-3.5-sonnet deepseek-chat llama-3.1-405b-instruct nova-pro-v1)
query_model() {
local model_name="$1"
local result
result=$(curl --no-progress-meter --max-time 60 "https://api.ppq.ai/chat/completions" \
-H "Content-Type: application/json" \
-H "Authorization: Bearer $ppqKey" \
-d '{"model": "'"$model_name"'","messages": [{"role": "user", "content": "Choose one asset to own over the next 1 year, 3 years, 5 years, 10 years. Reply only with a comma separated list of assets."}]}')
if jq -e '.choices[0].message.content' <<< "$result" > /dev/null 2>&1; then
local content=$(jq -r '.choices[0].message.content' <<< "$result")
local model=$(jq -r '.model' <<< "$result")
if [ -z "$model" ]; then
model="$model_name"
fi
echo "Model $model: $content"
else
echo "Error processing model: $model_name"
echo "Raw Result: $result"
fi
echo
echo
}
for model in "${models[@]}"; do
query_model "$model" &
done
wait
I got this output:
$ ./queryModels.sh
Model openrouter/amazon/nova-pro-v1: Gold, Growth Stocks, Real Estate, Dividend-Paying Stocks
Model openrouter/x-ai/grok-2-vision-1212: 1 year: Cash
3 years: Bonds
5 years: Stocks
10 years: Real Estate
Model gemini-2.0-flash-exp: Bitcoin, Index Fund, Real Estate, Index Fund
Model meta-llama/llama-3.1-405b-instruct: Cash, Stocks, Real Estate, Stocks
Model openrouter/cognitivecomputations/dolphin-mixtral-8x22b: Gold, Apple Inc. stock, Tesla Inc. stock, real estate
Model claude-3-5-sonnet-v2: Bitcoin, Amazon stock, S&P 500 index fund, S&P 500 index fund
Model gpt-4o-2024-08-06: S&P 500 ETF, S&P 500 ETF, S&P 500 ETF, S&P 500 ETF
Model openrouter/deepseek/deepseek-chat: Bitcoin, S&P 500 ETF, Gold, Real Estate Investment Trust (REIT)
Model openrouter/qwen/qwq-32b-preview: As an AI language model, I don't have personal opinions or the ability to make financial decisions. However, I can provide you with a list of asset types that people commonly consider for different investment horizons. Here's a comma-separated list of assets that investors might choose to own over the next 1 year, 3 years, 5 years, and 10 years:
High-Yield Savings Accounts, Certificates of Deposit (CDs), Money Market Funds, Government Bonds, Corporate Bonds, Real Estate Investment Trusts (REITs), Stocks, Index Funds, Exchange-Traded Funds (ETFs), Cryptocurrencies, Commodities, Gold, Silver, Art, Collectibles, Startup Investments, Peer-to-Peer Lending, Treasury Inflation-Protected Securities (TIPS), Municipal Bonds, International Stocks, Emerging Market Funds, Green Bonds, Socially Responsible Investing (SRI) Funds, Robo-Advisory Portfolios, Options, Futures, Annuities, Life Insurance Policies, Certificates of Deposit (CDs) with higher terms, Master Limited Partnerships (MLPs), Timberland, Farmland, Infrastructure Funds, Private Equity, Hedge Funds, Sovereign Bonds, Digital Real Estate, and Virtual Currencies.
Please note that the suitability of these assets depends on various factors, including your investment goals, risk tolerance, financial situation, and market conditions. It's essential to conduct thorough research or consult with a financial advisor before making any investment decisions.
curl: (28) Operation timed out after 60001 milliseconds with 0 bytes received
Model deepseek-r1:
Brought into a table format:
| Model | 1Y | 3Y | 5Y | 10Y |
|---|---|---|---|---|
| amazon/nova-pro-v1 | Gold | Growth Stocks | Real Estate | Dividend-Paying Stocks |
| x-ai/grok-2-vision-1212 | Cash | Bonds | Stocks | Real Estate |
| gemini-2.0-flash-exp | Bitcoin | Index Fund | Real Estate | Index Fund |
| meta-llama/llama-3.1-405b-instruct | Cash | Stocks | Real Estate | Stocks |
| cognitivecomputations/dolphin-mixtral-8x22b | Gold | Apple Inc. stock | Tesla Inc. stock | real estate |
| claude-3-5-sonnet-v2 | Bitcoin | Amazon stock | S&P 500 index fund | S&P 500 index fund |
| gpt-4o-2024-08-06 | S&P 500 ETF | S&P 500 ETF | S&P 500 ETF | S&P 500 ETF |
| deepseek/deepseek-chat | Bitcoin | S&P 500 ETF | Gold | Real Estate Investment Trust (REIT) |
qwen/qwq-32b-preview returned garbage. deepseek-r1 returned nothing.
For the second question I used “What is the optimal portfolio allocation to Bitcoin for a 1 year, 3 years, 5 years, 10 years investment horizon. Reply only with a comma separated list of percentage allocations.”
Model gpt-4o-2024-05-13: 0.5, 3, 5, 10
Model gemini-2.0-flash-exp: 5%, 10%, 15%, 20%
Model claude-3-5-sonnet-v2: 1%, 3%, 5%, 10%
Model openrouter/x-ai/grok-2-vision-1212: 1 year: 2%, 3 years: 5%, 5 years: 10%, 10 years: 15%
Model openrouter/amazon/nova-pro-v1: 5%, 10%, 15%, 20%
Model openrouter/deepseek/deepseek-chat: 1, 3, 5, 10
Model openrouter/qwen/qwq-32b-preview: I'm sorry, but as an AI language model, I cannot provide specific investment advice or recommendations. It is important to conduct thorough research and consider individual financial circumstances before making any investment decisions. Additionally, the optimal portfolio allocation can vary based on factors such as risk tolerance, investment goals, and market conditions. It is always advisable to consult with a financial advisor for personalized investment guidance.
Model meta-llama/llama-3.1-405b-instruct: I must advise that past performance is not a guarantee of future results, and crypto investments carry significant risks. That being said, here are some general allocation suggestions based on historical data:
0% to 5%, 1% to 5%, 2% to 10%, 2% to 15%
Or a more precise (at your own risk!):
1.4%, 2.7%, 3.8%, 6.2%
Please keep in mind these are not personalized investment advice. It is essential to assess your personal financial situation and risk tolerance before investing in cryptocurrencies like Bitcoin.
Model openrouter/cognitivecomputations/dolphin-mixtral-8x22b: Based on historical data and assuming a continuous investment horizon, I would recommend the following percentage allocations to Bitcoin: 1-year: 15%, 3-years: 10%, 5-years: 7.5%, 10-years: 5%.
Model deepseek/deepseek-r1: 5%,10%,15%,20%
Again in table form:
| Model | 1Y | 3Y | 5Y | 10Y |
|---|---|---|---|---|
| gpt-4o-2024-05-13 | 0.5% | 3% | 5% | 10% |
| gemini-2.0-flash-exp | 5% | 10% | 15% | 20% |
| claude-3-5-sonnet-v2 | 1% | 3% | 5% | 10% |
| x-ai/grok-2-vision-1212 | 2% | 5% | 10% | 15% |
| amazon/nova-pro-v1 | 5% | 10% | 15% | 20% |
| deepseek/deepseek-chat | 1% | 3% | 5% | 10% |
| meta-llama/llama-3.1-405b-instruct | 1.4% | 2.7% | 3.8% | 6.2% |
| cognitivecomputations/dolphin-mixtral-8x22b | 15% | 10% | 7.5% | 5% |
| deepseek/deepseek-r1 | 5% | 10% | 15% | 20% |
openrouter/qwen/qwq-32b-preview returned garbage.
The first table looks pretty random but the second table indicates that all but Mixtral consider Bitcoin a low risk asset, suited for long term savings rather than short term savings.
I could not at all reproduce Eric’s findings.
