Best Calibration
OpenAI 33%
Lowest MAPE among tracked providers
Model Insights Desk
View of model behavior across providers: calibration quality, directional bias, and relationship stability.
Directionality
94.5%
Average bullish call rate across providers
Average Premium
64.2%
Consensus tilt versus current prices
Coverage Depth
4
Providers with valid calibration records
Top Wins by Provider
Shows which providers' models most frequently rank in the top 3 closest to average asset prices.
75Top 3 wins
xAI2432%
Google2229.3%
Anthropic2229.3%
OpenAI79.3%
Provider Bullish Bias
Shows which providers' models are most likely to recommend bullish price targets.
xAI
107.7%
Google
61%
Anthropic
58.1%
OpenAI
30.1%
Model Price Prediction Correlation
Shows how models' price predictions correlate with each other.
Correlation Heatmap
| Model | claude-haiku-4.5 | claude-sonnet-4 | claude-sonnet-4.5 | gemini-2.5-flash | gemini-2.5-pro | gemini-3-flash-preview | gemini-3-pro-preview | gpt-5 | gpt-5-chat | gpt-5-mini | gpt-5-nano | grok-4 | grok-4-fast |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| claude-haiku-4.5 | 1.000 | 0.639 | 0.514 | 0.652 | 0.466 | 0.691 | 0.514 | 0.211 | 0.562 | 0.265 | 0.154 | 0.701 | 0.653 |
| claude-sonnet-4 | 0.639 | 1.000 | 0.877 | 0.664 | 0.758 | 0.742 | 0.747 | 0.445 | 0.747 | 0.682 | 0.114 | 0.673 | 0.784 |
| claude-sonnet-4.5 | 0.514 | 0.877 | 1.000 | 0.718 | 0.800 | 0.822 | 0.860 | 0.725 | 0.837 | 0.603 | 0.062 | 0.748 | 0.891 |
| gemini-2.5-flash | 0.652 | 0.664 | 0.718 | 1.000 | 0.410 | 0.920 | 0.850 | 0.660 | 0.874 | 0.386 | 0.107 | 0.977 | 0.811 |
| gemini-2.5-pro | 0.466 | 0.758 | 0.800 | 0.410 | 1.000 | 0.545 | 0.647 | 0.521 | 0.701 | 0.660 | 0.110 | 0.447 | 0.715 |
| gemini-3-flash-preview | 0.691 | 0.742 | 0.822 | 0.920 | 0.545 | 1.000 | 0.867 | 0.664 | 0.877 | 0.503 | 0.157 | 0.941 | 0.840 |
| gemini-3-pro-preview | 0.514 | 0.747 | 0.860 | 0.850 | 0.647 | 0.867 | 1.000 | 0.808 | 0.928 | 0.478 | 0.118 | 0.840 | 0.809 |
| gpt-5 | 0.211 | 0.445 | 0.725 | 0.660 | 0.521 | 0.664 | 0.808 | 1.000 | 0.755 | 0.324 | -0.037 | 0.664 | 0.654 |
| gpt-5-chat | 0.562 | 0.747 | 0.837 | 0.874 | 0.701 | 0.877 | 0.928 | 0.755 | 1.000 | 0.559 | 0.113 | 0.856 | 0.817 |
| gpt-5-mini | 0.265 | 0.682 | 0.603 | 0.386 | 0.660 | 0.503 | 0.478 | 0.324 | 0.559 | 1.000 | 0.170 | 0.388 | 0.527 |
| gpt-5-nano | 0.154 | 0.114 | 0.062 | 0.107 | 0.110 | 0.157 | 0.118 | -0.037 | 0.113 | 0.170 | 1.000 | 0.087 | 0.105 |
| grok-4 | 0.701 | 0.673 | 0.748 | 0.977 | 0.447 | 0.941 | 0.840 | 0.664 | 0.856 | 0.388 | 0.087 | 1.000 | 0.826 |
| grok-4-fast | 0.653 | 0.784 | 0.891 | 0.811 | 0.715 | 0.840 | 0.809 | 0.654 | 0.817 | 0.527 | 0.105 | 0.826 | 1.000 |
Model Similarity Network
Edges are drawn for correlation >= 0.820; thicker lines indicate stronger similarity.
googleopenaianthropicx-ai
Provider Performance Snapshot
Best calibrated by MAPE: OpenAI (33%). Most bullish profile: xAI (107.7% average premium). Mean premium across tracked providers: 64.2%.
| Rank | Provider | Bias Profile | MAPE | Bullish Calls | Median Premium | Premium Std Dev | Top-Rank Frequency |
|---|---|---|---|---|---|---|---|
| 1 | OpenAI | Most conservative | 33% | 89.9% (89.9%) | 26.2% | 24% | 1 (4%) |
| 2 | Anthropic | Balanced | 59% | 97.3% (97.3%) | 34.5% | 57.6% | 1 (4%) |
| 3 | Conservative tilt | 66.9% | 90.9% (90.9%) | 38.9% | 67.3% | 1 (4%) | |
| 4 | xAI | Most bullish | 107.7% | 100% (100%) | 55.1% | 113.8% | 22 (88%) |
Metric Definitions
- Bias Profile: Relative stance versus peers based on premium and bullish-call behavior.
- MAPE: Mean Absolute Percentage Error. Lower means better calibration.
- Bullish Calls: Percent of predictions above current price for this provider.
- Median Premium: Typical premium versus current price. Uses median to reduce outlier impact.
- Premium Std Dev: Volatility of premium percentages. Higher values imply less consistent bias.
- Top-Rank Frequency: How often this provider had the highest average prediction, shown as count and share.