OPHAELIS INDEX
Ophaelis Index evaluates real-world performance across quality, speed, cost efficiency, and reliability — so you can choose the right model for your constraints, not guess.
Continuously evaluated. Fully explainable. Built for real decisions.
Most AI comparisons tell you what scores highest.
They don't tell you:
Ophaelis Index was built to make those tradeoffs visible — so decisions aren't guesswork.
Best Overall
Best Overall identifies the model with the strongest weighted tradeoff across quality, speed, cost efficiency, and reliability under the active ranking profile. It does not mean the model is the most capable in every scenario.
Overall Score
Overall Score is a composite ranking score calculated from quality, speed, cost efficiency, and reliability using the active weighting profile. It is designed to compare real-world tradeoffs, not measure raw intelligence.
Estimated Cost
Estimated Cost is calculated using normalized token-based pricing across providers. This allows fair comparison regardless of provider-specific billing differences.
Reliability
Reliability measures repeated-run stability across evaluation scenarios. A model with lower variation and more consistent outcomes scores higher. It reflects stability under current evaluation conditions, not raw capability.
Current System Verdict (Balanced Profile)
Best Overall reflects the strongest tradeoff across quality, speed, cost efficiency, and reliability — not the highest raw capability.
OpenAI
90.54
Overall Score is a composite ranking score calculated from quality, speed, cost efficiency, and reliability using the Balanced weighting profile.
GPT 5.4 Mini wins 7 decision areas and sustains a typical 1.18-point margin over the runner-up in those lanes.
Decision Areas Evaluated
10
Limited Coverage Areas
6
Last Evaluator Run
Apr 2, 2026, 1:46 PM
Public quick views for different operating goals
Best Overall under Balanced
OpenAI
90.54
Overall Score is a composite ranking score calculated from quality, speed, cost efficiency, and reliability using the Balanced weighting profile.
Strongest overall tradeoff across quality, speed, cost efficiency, and reliability under balanced weighting.
Highest Output Quality
93.00
Overall Score is a composite ranking score calculated from quality, speed, cost efficiency, and reliability using the Quality First weighting profile.
Strongest accuracy and task completion performance across evaluation scenarios.
Fastest Response
90.95
Overall Score is a composite ranking score calculated from quality, speed, cost efficiency, and reliability using the Speed First weighting profile.
Lowest response latency while maintaining usable quality for time-sensitive workflows.
Most Cost Efficient
93.90
Overall Score is a composite ranking score calculated from quality, speed, cost efficiency, and reliability using the Cost First weighting profile.
Best affordability-to-performance tradeoff when spend efficiency is prioritized.
Most Consistent
93.99
Overall Score is a composite ranking score calculated from quality, speed, cost efficiency, and reliability using the Reliability First weighting profile.
Most stable repeated-run outcomes across changing evaluation conditions.
Track how model performance shifts over time as new versions and conditions change.
Last run: Apr 2, 2026, 1:46 PM
Method: 3-run averaged evaluation
Dimensions: Quality • Speed • Cost Efficiency • Reliability
Coverage: 10 evaluation areas
Continuously updated as models and providers change
Sortable free view for quick model comparison
Last updated: Apr 2, 2026, 1:46 PM
Ranking Profile
Table Sort
Overall Score is a composite ranking score calculated from quality, speed, cost efficiency, and reliability using the Balanced weighting profile.
Ranking Profile changes how Overall Score is calculated.
Table Sort changes how rows are ordered for inspection.
Active weighting: Quality 50% | Speed 20% | Cost Efficiency 15% | Reliability 15%
Number of times the model ranked #1 across individual evaluation dimensions.
A model can have fewer Top Finishes and still rank highly overall when it stays consistently strong across all weighted dimensions.
Example: if you care most about speed, use Speed First. Then use table sort to inspect which fast models also retain strong quality.
Provenance: leaderboard uses the active weighting profile; overall score is a weighted tradeoff score; live results are based on 3 evaluation runs per benchmark item; evaluated decision areas: 10 (4 live, 6 limited coverage).
Viewing: Balanced
Sorted by: Overall Score
Overall Score reflects: Balanced weighting across Quality, Speed, Cost Efficiency, and Reliability.
| Model | Provider | Overall Score | Quality | Speed | Cost Efficiency | Reliability | Top Finishes |
|---|---|---|---|---|---|---|---|
| 1. GPT 5.4 Mini | OpenAI | 90.54 | 90.5 | 86.9 | 85.4 | 97.8 | 7 |
| 2. Gemini 3.1 Pro Preview | 89.84 | 98.0 | 84.8 | 64.2 | 95.0 | 0 | |
| 3. Gemini 2.5 Flash Lite | 89.64 | 83.3 | 81.5 | 100.0 | 100.0 | 0 | |
| 4. Gemini 3 Flash | 89.64 | 88.5 | 95.8 | 83.2 | 91.7 | 0 | |
| 5. Claude Haiku 4 5 | Anthropic | 88.14 | 84.3 | 87.6 | 91.7 | 94.4 | 0 |
| 6. Mistral Medium | Other frontier | 87.96 | 84.1 | 73.8 | 97.8 | 100.0 | 2 |
| 7. Mistral Large | Other frontier | 87.94 | 89.5 | 88.8 | 79.2 | 90.3 | 0 |
| 8. GPT 5.4 Nano | OpenAI | 87.61 | 82.2 | 86.8 | 96.9 | 93.8 | 1 |
| 9. Mistral Fast | Other frontier | 87.29 | 82.5 | 96.8 | 91.2 | 86.7 | 0 |
| 10. GPT 5.4 | OpenAI | 86.72 | 93.8 | 73.6 | 67.1 | 99.4 | 0 |
Deeper Access
Open the deeper view for full tables, decision-area breakdowns, per-model details, and cost transparency diagnostics.
Explore Full Benchmark DataMethodology