Claude Sonnet 5 vs Llama 4 Maverick

Claude Sonnet 5 and Llama 4 Maverick measured on preventive / public health / zoonoses, on the same veterinary examination items, run on VetEval's own infrastructure at fixed decoding settings and scored by a panel that never grades its own model.

Llama 4 Maverick scores 11.5 points above Claude Sonnet 5 on Preventive / Public Health / Zoonoses, but their confidence intervals overlap, on this evidence the two cannot be separated.

Claude Sonnet 5 and Llama 4 Maverick carry a confirmed safety cap, which limits the Index regardless of how the clinical categories scored.

Side-by-side comparison of Claude Sonnet 5, Llama 4 Maverick across the VetEval Index, each scoring category, the safety gate, cost and latency.
Measure	Claude Sonnet 5Anthropic · proprietary	Llama 4 MaverickMeta · open weights
VetEval Index	22.3, best in this row21.4–23.1	19.018.1–19.8
Rank	5–7shared rank	8–9shared rank
Clinical Diagnosis & Differential	73.268.1–78.2	79.5, best in this row75.2–83.8
Pharmacology & Dosing	68.2, best in this row62.6–73.8	60.054.2–65.7
Species-Specific Knowledge	62.0, best in this row54.6–69.4	60.253.0–67.5
Triage & Emergency	76.6, best in this row70.1–83.0	67.460.4–74.5
Preventive / Public Health / Zoonoses	57.748.4–67.0	69.1, best in this row61.1–77.2
Communication & Professionalism	80.1, best in this row71.5–88.6	72.562.5–82.5
Safety gate	Capped×0.40	Capped×0.40
Cost per 1,000 items	$2.48list price at run time	$1.29, best in this rowlist price at run time
Median time per question	1.6s, best in this rowexcluding retries	26.9sexcluding retries
Weights	ProprietaryAnthropic	OpenMeta

A shaded cell is the best value in its row, and the bar shows the score out of 100. The line under the Index and each category is the 95% confidence interval, where two intervals overlap, the ordering is not something this measurement establishes. Cost is list price at the time of the run and is not part of the Index.

What this comparison does not tell you

These scores describe performance on veterinary examination items. They are not a certification, and no result here licenses a model for clinical use.
The Index weights categories for a general caseload. Every per-category score is published above so you can weigh them for yours instead, or compare on a single category using the control above.
Cost and latency are measured at run time and move separately from the scores. A model that is cheaper today may not be next month.

Related comparisons

Claude Sonnet 5 vs Llama 4 Maverick by category

Scope. VetEval measures model performance on veterinary examination items. It is not clinical advice, and no result here licenses a model for clinical use.