MedicalBenchmark
OpenAI: GPT-5.3-Codex provider

GPT-5.3-Codex

24

#24 of 319 models in the general ranking

Cumulative performance across 3 MIR exams

Net score

575.00 pts

Accuracy

96.5%

Correct / Incorrect

579 / 12

Total Cost

$0.00

Overall Performance

(vs. average)
Accuracy

96.5%

avg: 80.6%

Net score

575.00 pts

avg: 453.30 pts

Correct

579

avg: 483

Incorrect

12

avg: 90

Total Cost

$0.00

avg: $9.58

Average response time

9.9s

avg: 17.9s

Output Tokens

312K

avg: 1.3M

Reasoning Tokens

126K

avg: 898K

Average confidence

99.8%

avg: 95.4%

Breakdown by Exam

MIR 2024
40
Correct
191
Incorrect
5
Accuracy
95.5%
Net score
189.33
MIR 2025
11
Correct
193
Incorrect
5
Accuracy
96.5%
Net score
191.33
MIR 2026
33
Correct
195
Incorrect
2
Accuracy
97.5%
Net score
194.33