MedicalBenchmark
OpenAI: gpt-oss-safeguard-20b provider

gpt-oss-safeguard-20b

159

#159 of 319 models in the general ranking

Cumulative performance across 3 MIR exams

Net score

502.33 pts

Accuracy

87.7%

Correct / Incorrect

526 / 71

Total Cost

$0.20

Overall Performance

(vs. average)
Accuracy

87.7%

avg: 80.6%

Net score

502.33 pts

avg: 453.30 pts

Correct

526

avg: 483

Incorrect

71

avg: 90

Total Cost

$0.20

avg: $9.58

Average response time

2.0s

avg: 17.9s

Output Tokens

608K

avg: 1.3M

Reasoning Tokens

356K

avg: 898K

Average confidence

99.3%

avg: 95.4%

Breakdown by Exam

MIR 2024
171
Correct
177
Incorrect
22
Accuracy
88.5%
Net score
169.66
MIR 2025
152
Correct
172
Incorrect
28
Accuracy
86.0%
Net score
162.66
MIR 2026
162
Correct
177
Incorrect
21
Accuracy
88.5%
Net score
170.00