Openai / gpt-oss-20b

medium

96.1% 196 / 204 fixtures · 1 run(s)
46,289 input / 53,293 total output / 47,692 reasoning within output tokens $0.01038587
83.3% 170 / 204 fixtures · 1 run(s)
47,414 input / 60,579 total output / 40,037 reasoning within output tokens $0.01116696
Loading reliability summary…
Pass Rate Delta
-12.7% Text: 96.1% → JSON: 83.3%
+3
Gained
JSON pass / text fail
−29
Lost
Text pass / JSON fail
167
Unchanged Pass
Both pass
5
Unchanged Fail
Both fail
Fixture Reliability Delta
Fixture Text JSON Delta
f010 0% (0/1) 100% (1/1) -100%
Benchmark Deltas
Benchmark Text JSON Delta
reflog 100% 50% -50%
cherry_pick 83.3% 41.7% -41.7%
branch_cleanup 100% 75% -25%
merge_conflicts 91.7% 66.7% -25%
rebase 91.7% 75% -16.7%
tag_management 100% 83.3% -16.7%
commit_messages 91.7% 100% + 8.3%
git_bisect 100% 91.7% -8.3%
git_clean 100% 91.7% -8.3%
git_log_format 100% 91.7% -8.3%
git_show 100% 91.7% -8.3%
stash_recovery 100% 91.7% -8.3%
worktree_usage 83.3% 75% -8.3%
blame_forensics 100% 100% + 0%
commit_squash 91.7% 91.7% + 0%
git_grep 100% 100% + 0%
submodule_usage 100% 100% + 0%
Changed Fixtures (32)