Compare two models

Compare any two models head-to-head across a shared benchmark — no signup needed.

Putting two models side-by-side is one of the highest-leverage things Stratix does. Use it during model selection, when a new model drops, or when leadership asks "is X better than Y."

Steps

1. Open Compare models

Go to stratix.layerlens.ai/compare-models.

2. Pick the first model

Search and select a model in the left column. The page populates with that model's benchmark scores.

3. Pick the second model

Same in the right column. The page now shows both models' scores on every benchmark either has been evaluated against.

4. Read the comparison

The score table highlights:

• Where each model wins (green)
• Where they tie (gray)
• Where the gap is large vs small

5. Pick a benchmark to focus on

Click any benchmark row to drill into the comparison for that benchmark — including raw scores, sample inputs, and (where available) sample outputs from each model.

6. Save the comparison URL

The URL contains both model IDs. Bookmark it or share with your team.

Verify

You should be able to compare GPT-5.3 vs Claude Opus 4.6 on MMLU and see both models' scores on the same row.

Common patterns

• Frontier vs cost-optimized. Compare a frontier model against a smaller, cheaper one for tasks where you suspect the smaller model is good enough.
• Provider against provider. Compare an OpenAI model against an Anthropic model on the benchmark closest to your use case.
• Old vs new generation. When a new generation drops, compare it against the predecessor on benchmarks you care about.

Where to next

• Browse benchmarks
• Premium — Compare models with your dataset
• Stratix Public — Compare models reference
• Use case: Model evaluation