Run 55e3cd85

Name: gmod-bench run 55e3cd85-0c88-40a1-9364-114f6ce4bb36
Creator: gmod-bench
License: https://github.com/Iydah/gmod-bench/blob/main/LICENSE

Jul 10, 2026, 17:44

Codex · 31 models · repeat ×1

55e3cd85-0c88-40a1-9364-114f6ce4bb36

Fixtures

1,178

Attempts

Models

10m 10s

Duration

Codex×1 repeat

Downloads

Audit pack

Results

31 models

#	Adapter	Model	Think	Pass %	Cover %	Score	Outcomes	Time
1	Codex	gpt-5.6-sol	Extra High	91.9%	97.4%	—	34/2/1/1	16.4s
#1 gpt-5.6-sol Extra High Codex Pass rate91.9% 34 pass · 37 scored Coverage97.4% 37 / 38 evidence-bearing Verified score— 95% interval — · 1 run(s) pass@k91.9% 34/38 fixtures ever solved Outcome mix Pass34 Partial2 Wrong1 Invalid output1 Invalid output means the model returned content that broke the answer contract, so it scores zero. Harness failures returned no answer and are excluded from score. Attempt funnel Attempts38 Scored37 Passed34 Fixtures solved34/38 Run facts Avg time 16.4s Repeat Single-shot Evidence share 37/38 (97.4%) Verified runs 1 95% interval — Raw label codex/gpt-5.6-sol@xhigh Run id 55e3cd85-0c88-40a1-9364-114f6ce4bb36 Latest run fixtures 34/38 passed in linked run perf-string-table-concatperf-angle-zeroperf-localplayer-cacheperf-looking-atents-iteratorhook-addisvalidnet-receiversperf-config-varperf-darkrpvarperf-disttosqrperf-ents-iteratorperf-find-ents-nearperf-find-players-nearperf-for-vs-whileperf-hook-onceperf-hudpaint-cacheperf-local-colorperf-local-vs-globalperf-meta-vs-argperf-msg-vs-printperf-pairs-ipairs-forperf-plain-lua-vs-fnperf-player-cache-hooksperf-ply-table-indexperf-seq-tablesperf-setdrawcolor-splitperf-sqlite-vs-mysqlperf-steamid-cacheperf-surface-drawrectperf-surface-textperf-table-count-hashperf-table-emptyperf-table-hasvalueperf-table-literalperf-table-randomperf-x-times-xplayer-iterator
2	Codex	gpt-5.6-terra	Low	89.2%	97.4%	—	33/3/1/1	12.7s
#2 gpt-5.6-terra Low Codex Pass rate89.2% 33 pass · 37 scored Coverage97.4% 37 / 38 evidence-bearing Verified score— 95% interval — · 1 run(s) pass@k89.2% 33/38 fixtures ever solved Outcome mix Pass33 Partial3 Wrong1 Invalid output1 Invalid output means the model returned content that broke the answer contract, so it scores zero. Harness failures returned no answer and are excluded from score. Attempt funnel Attempts38 Scored37 Passed33 Fixtures solved33/38 Run facts Avg time 12.7s Repeat Single-shot Evidence share 37/38 (97.4%) Verified runs 1 95% interval — Raw label codex/gpt-5.6-terra@low Run id 55e3cd85-0c88-40a1-9364-114f6ce4bb36 Latest run fixtures 33/38 passed in linked run perf-string-table-concatperf-table-hasvalueperf-angle-zeroperf-find-ents-nearperf-looking-atents-iteratorhook-addisvalidnet-receiversperf-config-varperf-darkrpvarperf-disttosqrperf-ents-iteratorperf-find-players-nearperf-for-vs-whileperf-hook-onceperf-hudpaint-cacheperf-local-colorperf-local-vs-globalperf-localplayer-cacheperf-meta-vs-argperf-msg-vs-printperf-pairs-ipairs-forperf-plain-lua-vs-fnperf-player-cache-hooksperf-ply-table-indexperf-seq-tablesperf-setdrawcolor-splitperf-sqlite-vs-mysqlperf-steamid-cacheperf-surface-drawrectperf-surface-textperf-table-count-hashperf-table-emptyperf-table-literalperf-table-randomperf-x-times-xplayer-iterator
3	Codex	gpt-5.6-luna	Max	89.2%	97.4%	—	33/2/2/1	21.1s
#3 gpt-5.6-luna Max Codex Pass rate89.2% 33 pass · 37 scored Coverage97.4% 37 / 38 evidence-bearing Verified score— 95% interval — · 1 run(s) pass@k89.2% 33/38 fixtures ever solved Outcome mix Pass33 Partial2 Wrong2 Invalid output1 Invalid output means the model returned content that broke the answer contract, so it scores zero. Harness failures returned no answer and are excluded from score. Attempt funnel Attempts38 Scored37 Passed33 Fixtures solved33/38 Run facts Avg time 21.1s Repeat Single-shot Evidence share 37/38 (97.4%) Verified runs 1 95% interval — Raw label codex/gpt-5.6-luna@max Run id 55e3cd85-0c88-40a1-9364-114f6ce4bb36 Latest run fixtures 33/38 passed in linked run perf-string-table-concatperf-angle-zeroperf-seq-tablesperf-localplayer-cacheperf-looking-atents-iteratorhook-addisvalidnet-receiversperf-config-varperf-darkrpvarperf-disttosqrperf-ents-iteratorperf-find-ents-nearperf-find-players-nearperf-for-vs-whileperf-hook-onceperf-hudpaint-cacheperf-local-colorperf-local-vs-globalperf-meta-vs-argperf-msg-vs-printperf-pairs-ipairs-forperf-plain-lua-vs-fnperf-player-cache-hooksperf-ply-table-indexperf-setdrawcolor-splitperf-sqlite-vs-mysqlperf-steamid-cacheperf-surface-drawrectperf-surface-textperf-table-count-hashperf-table-emptyperf-table-hasvalueperf-table-literalperf-table-randomperf-x-times-xplayer-iterator
4	Codex	gpt-5.6-luna	High	86.5%	97.4%	—	32/4/1/1	14.7s
#4 gpt-5.6-luna High Codex Pass rate86.5% 32 pass · 37 scored Coverage97.4% 37 / 38 evidence-bearing Verified score— 95% interval — · 1 run(s) pass@k86.5% 32/38 fixtures ever solved Outcome mix Pass32 Partial4 Wrong1 Invalid output1 Invalid output means the model returned content that broke the answer contract, so it scores zero. Harness failures returned no answer and are excluded from score. Attempt funnel Attempts38 Scored37 Passed32 Fixtures solved32/38 Run facts Avg time 14.7s Repeat Single-shot Evidence share 37/38 (97.4%) Verified runs 1 95% interval — Raw label codex/gpt-5.6-luna@high Run id 55e3cd85-0c88-40a1-9364-114f6ce4bb36 Latest run fixtures 32/38 passed in linked run perf-string-table-concatperf-pairs-ipairs-forperf-angle-zeroperf-localplayer-cacheperf-looking-atplayer-iteratorents-iteratorhook-addisvalidnet-receiversperf-config-varperf-darkrpvarperf-disttosqrperf-ents-iteratorperf-find-ents-nearperf-find-players-nearperf-for-vs-whileperf-hook-onceperf-hudpaint-cacheperf-local-colorperf-local-vs-globalperf-meta-vs-argperf-msg-vs-printperf-plain-lua-vs-fnperf-player-cache-hooksperf-ply-table-indexperf-seq-tablesperf-setdrawcolor-splitperf-sqlite-vs-mysqlperf-steamid-cacheperf-surface-drawrectperf-surface-textperf-table-count-hashperf-table-emptyperf-table-hasvalueperf-table-literalperf-table-randomperf-x-times-x
5	Codex	gpt-5.6-sol	Low	86.5%	97.4%	—	32/4/1/1	9.8s
#5 gpt-5.6-sol Low Codex Pass rate86.5% 32 pass · 37 scored Coverage97.4% 37 / 38 evidence-bearing Verified score— 95% interval — · 1 run(s) pass@k86.5% 32/38 fixtures ever solved Outcome mix Pass32 Partial4 Wrong1 Invalid output1 Invalid output means the model returned content that broke the answer contract, so it scores zero. Harness failures returned no answer and are excluded from score. Attempt funnel Attempts38 Scored37 Passed32 Fixtures solved32/38 Run facts Avg time 9.8s Repeat Single-shot Evidence share 37/38 (97.4%) Verified runs 1 95% interval — Raw label codex/gpt-5.6-sol@low Run id 55e3cd85-0c88-40a1-9364-114f6ce4bb36 Latest run fixtures 32/38 passed in linked run perf-string-table-concatperf-angle-zeroperf-find-ents-nearperf-for-vs-whileperf-localplayer-cacheperf-looking-atents-iteratorhook-addisvalidnet-receiversperf-config-varperf-darkrpvarperf-disttosqrperf-ents-iteratorperf-find-players-nearperf-hook-onceperf-hudpaint-cacheperf-local-colorperf-local-vs-globalperf-meta-vs-argperf-msg-vs-printperf-pairs-ipairs-forperf-plain-lua-vs-fnperf-player-cache-hooksperf-ply-table-indexperf-seq-tablesperf-setdrawcolor-splitperf-sqlite-vs-mysqlperf-steamid-cacheperf-surface-drawrectperf-surface-textperf-table-count-hashperf-table-emptyperf-table-hasvalueperf-table-literalperf-table-randomperf-x-times-xplayer-iterator
6	Codex	gpt-5.6-luna	Medium	86.5%	97.4%	—	32/3/2/1	11.7s
#6 gpt-5.6-luna Medium Codex Pass rate86.5% 32 pass · 37 scored Coverage97.4% 37 / 38 evidence-bearing Verified score— 95% interval — · 1 run(s) pass@k86.5% 32/38 fixtures ever solved Outcome mix Pass32 Partial3 Wrong2 Invalid output1 Invalid output means the model returned content that broke the answer contract, so it scores zero. Harness failures returned no answer and are excluded from score. Attempt funnel Attempts38 Scored37 Passed32 Fixtures solved32/38 Run facts Avg time 11.7s Repeat Single-shot Evidence share 37/38 (97.4%) Verified runs 1 95% interval — Raw label codex/gpt-5.6-luna@medium Run id 55e3cd85-0c88-40a1-9364-114f6ce4bb36 Latest run fixtures 32/38 passed in linked run perf-string-table-concatperf-plain-lua-vs-fnperf-ply-table-indexperf-angle-zeroperf-looking-atplayer-iteratorents-iteratorhook-addisvalidnet-receiversperf-config-varperf-darkrpvarperf-disttosqrperf-ents-iteratorperf-find-ents-nearperf-find-players-nearperf-for-vs-whileperf-hook-onceperf-hudpaint-cacheperf-local-colorperf-local-vs-globalperf-localplayer-cacheperf-meta-vs-argperf-msg-vs-printperf-pairs-ipairs-forperf-player-cache-hooksperf-seq-tablesperf-setdrawcolor-splitperf-sqlite-vs-mysqlperf-steamid-cacheperf-surface-drawrectperf-surface-textperf-table-count-hashperf-table-emptyperf-table-hasvalueperf-table-literalperf-table-randomperf-x-times-x
7	Codex	gpt-5.6-terra	Extra High	86.5%	97.4%	—	32/3/2/1	15.2s
#7 gpt-5.6-terra Extra High Codex Pass rate86.5% 32 pass · 37 scored Coverage97.4% 37 / 38 evidence-bearing Verified score— 95% interval — · 1 run(s) pass@k86.5% 32/38 fixtures ever solved Outcome mix Pass32 Partial3 Wrong2 Invalid output1 Invalid output means the model returned content that broke the answer contract, so it scores zero. Harness failures returned no answer and are excluded from score. Attempt funnel Attempts38 Scored37 Passed32 Fixtures solved32/38 Run facts Avg time 15.2s Repeat Single-shot Evidence share 37/38 (97.4%) Verified runs 1 95% interval — Raw label codex/gpt-5.6-terra@xhigh Run id 55e3cd85-0c88-40a1-9364-114f6ce4bb36 Latest run fixtures 32/38 passed in linked run perf-string-table-concatperf-plain-lua-vs-fnperf-table-hasvalueperf-find-ents-nearperf-localplayer-cacheperf-looking-atents-iteratorhook-addisvalidnet-receiversperf-angle-zeroperf-config-varperf-darkrpvarperf-disttosqrperf-ents-iteratorperf-find-players-nearperf-for-vs-whileperf-hook-onceperf-hudpaint-cacheperf-local-colorperf-local-vs-globalperf-meta-vs-argperf-msg-vs-printperf-pairs-ipairs-forperf-player-cache-hooksperf-ply-table-indexperf-seq-tablesperf-setdrawcolor-splitperf-sqlite-vs-mysqlperf-steamid-cacheperf-surface-drawrectperf-surface-textperf-table-count-hashperf-table-emptyperf-table-literalperf-table-randomperf-x-times-xplayer-iterator
8	Codex	gpt-5.6-terra	Max	86.5%	97.4%	—	32/2/3/1	18.1s
#8 gpt-5.6-terra Max Codex Pass rate86.5% 32 pass · 37 scored Coverage97.4% 37 / 38 evidence-bearing Verified score— 95% interval — · 1 run(s) pass@k86.5% 32/38 fixtures ever solved Outcome mix Pass32 Partial2 Wrong3 Invalid output1 Invalid output means the model returned content that broke the answer contract, so it scores zero. Harness failures returned no answer and are excluded from score. Attempt funnel Attempts38 Scored37 Passed32 Fixtures solved32/38 Run facts Avg time 18.1s Repeat Single-shot Evidence share 37/38 (97.4%) Verified runs 1 95% interval — Raw label codex/gpt-5.6-terra@max Run id 55e3cd85-0c88-40a1-9364-114f6ce4bb36 Latest run fixtures 32/38 passed in linked run perf-string-table-concatperf-angle-zeroperf-pairs-ipairs-forperf-table-hasvalueperf-find-ents-nearperf-looking-atents-iteratorhook-addisvalidnet-receiversperf-config-varperf-darkrpvarperf-disttosqrperf-ents-iteratorperf-find-players-nearperf-for-vs-whileperf-hook-onceperf-hudpaint-cacheperf-local-colorperf-local-vs-globalperf-localplayer-cacheperf-meta-vs-argperf-msg-vs-printperf-plain-lua-vs-fnperf-player-cache-hooksperf-ply-table-indexperf-seq-tablesperf-setdrawcolor-splitperf-sqlite-vs-mysqlperf-steamid-cacheperf-surface-drawrectperf-surface-textperf-table-count-hashperf-table-emptyperf-table-literalperf-table-randomperf-x-times-xplayer-iterator
9	Codex	gpt-5.6-terra	Medium	86.5%	97.4%	—	32/2/3/1	7.8s
#9 gpt-5.6-terra Medium Codex Pass rate86.5% 32 pass · 37 scored Coverage97.4% 37 / 38 evidence-bearing Verified score— 95% interval — · 1 run(s) pass@k86.5% 32/38 fixtures ever solved Outcome mix Pass32 Partial2 Wrong3 Invalid output1 Invalid output means the model returned content that broke the answer contract, so it scores zero. Harness failures returned no answer and are excluded from score. Attempt funnel Attempts38 Scored37 Passed32 Fixtures solved32/38 Run facts Avg time 7.8s Repeat Single-shot Evidence share 37/38 (97.4%) Verified runs 1 95% interval — Raw label codex/gpt-5.6-terra@medium Run id 55e3cd85-0c88-40a1-9364-114f6ce4bb36 Latest run fixtures 32/38 passed in linked run perf-string-table-concatperf-angle-zeroperf-plain-lua-vs-fnperf-table-hasvalueperf-find-ents-nearperf-looking-atents-iteratorhook-addisvalidnet-receiversperf-config-varperf-darkrpvarperf-disttosqrperf-ents-iteratorperf-find-players-nearperf-for-vs-whileperf-hook-onceperf-hudpaint-cacheperf-local-colorperf-local-vs-globalperf-localplayer-cacheperf-meta-vs-argperf-msg-vs-printperf-pairs-ipairs-forperf-player-cache-hooksperf-ply-table-indexperf-seq-tablesperf-setdrawcolor-splitperf-sqlite-vs-mysqlperf-steamid-cacheperf-surface-drawrectperf-surface-textperf-table-count-hashperf-table-emptyperf-table-literalperf-table-randomperf-x-times-xplayer-iterator
10	Codex	gpt-5.4-mini	Extra High	83.8%	97.4%	—	31/4/2/1	23.2s
#10 gpt-5.4-mini Extra High Codex Pass rate83.8% 31 pass · 37 scored Coverage97.4% 37 / 38 evidence-bearing Verified score— 95% interval — · 1 run(s) pass@k83.8% 31/38 fixtures ever solved Outcome mix Pass31 Partial4 Wrong2 Invalid output1 Invalid output means the model returned content that broke the answer contract, so it scores zero. Harness failures returned no answer and are excluded from score. Attempt funnel Attempts38 Scored37 Passed31 Fixtures solved31/38 Run facts Avg time 23.2s Repeat Single-shot Evidence share 37/38 (97.4%) Verified runs 1 95% interval — Raw label codex/gpt-5.4-mini@xhigh Run id 55e3cd85-0c88-40a1-9364-114f6ce4bb36 Latest run fixtures 31/38 passed in linked run perf-string-table-concatperf-config-varperf-pairs-ipairs-forperf-angle-zeroperf-localplayer-cacheperf-looking-atplayer-iteratorents-iteratorhook-addisvalidnet-receiversperf-darkrpvarperf-disttosqrperf-ents-iteratorperf-find-ents-nearperf-find-players-nearperf-for-vs-whileperf-hook-onceperf-hudpaint-cacheperf-local-colorperf-local-vs-globalperf-meta-vs-argperf-msg-vs-printperf-plain-lua-vs-fnperf-player-cache-hooksperf-ply-table-indexperf-seq-tablesperf-setdrawcolor-splitperf-sqlite-vs-mysqlperf-steamid-cacheperf-surface-drawrectperf-surface-textperf-table-count-hashperf-table-emptyperf-table-hasvalueperf-table-literalperf-table-randomperf-x-times-x
11	Codex	gpt-5.6-sol	Medium	83.8%	97.4%	—	31/4/2/1	12.9s
#11 gpt-5.6-sol Medium Codex Pass rate83.8% 31 pass · 37 scored Coverage97.4% 37 / 38 evidence-bearing Verified score— 95% interval — · 1 run(s) pass@k83.8% 31/38 fixtures ever solved Outcome mix Pass31 Partial4 Wrong2 Invalid output1 Invalid output means the model returned content that broke the answer contract, so it scores zero. Harness failures returned no answer and are excluded from score. Attempt funnel Attempts38 Scored37 Passed31 Fixtures solved31/38 Run facts Avg time 12.9s Repeat Single-shot Evidence share 37/38 (97.4%) Verified runs 1 95% interval — Raw label codex/gpt-5.6-sol@medium Run id 55e3cd85-0c88-40a1-9364-114f6ce4bb36 Latest run fixtures 31/38 passed in linked run perf-string-table-concatperf-angle-zeroperf-table-hasvalueperf-find-ents-nearperf-for-vs-whileperf-localplayer-cacheperf-looking-atents-iteratorhook-addisvalidnet-receiversperf-config-varperf-darkrpvarperf-disttosqrperf-ents-iteratorperf-find-players-nearperf-hook-onceperf-hudpaint-cacheperf-local-colorperf-local-vs-globalperf-meta-vs-argperf-msg-vs-printperf-pairs-ipairs-forperf-plain-lua-vs-fnperf-player-cache-hooksperf-ply-table-indexperf-seq-tablesperf-setdrawcolor-splitperf-sqlite-vs-mysqlperf-steamid-cacheperf-surface-drawrectperf-surface-textperf-table-count-hashperf-table-emptyperf-table-literalperf-table-randomperf-x-times-xplayer-iterator
12	Codex	gpt-5.6-terra	High	81.1%	97.4%	—	30/3/4/1	7.2s
#12 gpt-5.6-terra High Codex Pass rate81.1% 30 pass · 37 scored Coverage97.4% 37 / 38 evidence-bearing Verified score— 95% interval — · 1 run(s) pass@k81.1% 30/38 fixtures ever solved Outcome mix Pass30 Partial3 Wrong4 Invalid output1 Invalid output means the model returned content that broke the answer contract, so it scores zero. Harness failures returned no answer and are excluded from score. Attempt funnel Attempts38 Scored37 Passed30 Fixtures solved30/38 Run facts Avg time 7.2s Repeat Single-shot Evidence share 37/38 (97.4%) Verified runs 1 95% interval — Raw label codex/gpt-5.6-terra@high Run id 55e3cd85-0c88-40a1-9364-114f6ce4bb36 Latest run fixtures 30/38 passed in linked run perf-string-table-concatperf-angle-zeroperf-plain-lua-vs-fnperf-surface-textperf-table-hasvalueperf-darkrpvarperf-find-ents-nearperf-looking-atents-iteratorhook-addisvalidnet-receiversperf-config-varperf-disttosqrperf-ents-iteratorperf-find-players-nearperf-for-vs-whileperf-hook-onceperf-hudpaint-cacheperf-local-colorperf-local-vs-globalperf-localplayer-cacheperf-meta-vs-argperf-msg-vs-printperf-pairs-ipairs-forperf-player-cache-hooksperf-ply-table-indexperf-seq-tablesperf-setdrawcolor-splitperf-sqlite-vs-mysqlperf-steamid-cacheperf-surface-drawrectperf-table-count-hashperf-table-emptyperf-table-literalperf-table-randomperf-x-times-xplayer-iterator
13	Codex	gpt-5.4-mini	Medium	91.7%	94.7%	—	33/2/1/2	14.9s
#13 gpt-5.4-mini Medium Codex Pass rate91.7% 33 pass · 36 scored Coverage94.7% 36 / 38 evidence-bearing Verified score— 95% interval — · 1 run(s) pass@k91.7% 33/38 fixtures ever solved Outcome mix Pass33 Partial2 Wrong1 Invalid output2 Invalid output means the model returned content that broke the answer contract, so it scores zero. Harness failures returned no answer and are excluded from score. Attempt funnel Attempts38 Scored36 Passed33 Fixtures solved33/38 Run facts Avg time 14.9s Repeat Single-shot Evidence share 36/38 (94.7%) Verified runs 1 95% interval — Raw label codex/gpt-5.4-mini@medium Run id 55e3cd85-0c88-40a1-9364-114f6ce4bb36 Latest run fixtures 33/38 passed in linked run perf-find-ents-nearperf-string-table-concatperf-angle-zeroperf-looking-atplayer-iteratorents-iteratorhook-addisvalidnet-receiversperf-config-varperf-darkrpvarperf-disttosqrperf-ents-iteratorperf-find-players-nearperf-for-vs-whileperf-hook-onceperf-hudpaint-cacheperf-local-colorperf-local-vs-globalperf-localplayer-cacheperf-meta-vs-argperf-msg-vs-printperf-pairs-ipairs-forperf-plain-lua-vs-fnperf-player-cache-hooksperf-ply-table-indexperf-seq-tablesperf-setdrawcolor-splitperf-sqlite-vs-mysqlperf-steamid-cacheperf-surface-drawrectperf-surface-textperf-table-count-hashperf-table-emptyperf-table-hasvalueperf-table-literalperf-table-randomperf-x-times-x
14	Codex	gpt-5.6-luna	Low	91.7%	94.7%	—	33/2/1/2	5.8s
#14 gpt-5.6-luna Low Codex Pass rate91.7% 33 pass · 36 scored Coverage94.7% 36 / 38 evidence-bearing Verified score— 95% interval — · 1 run(s) pass@k91.7% 33/38 fixtures ever solved Outcome mix Pass33 Partial2 Wrong1 Invalid output2 Invalid output means the model returned content that broke the answer contract, so it scores zero. Harness failures returned no answer and are excluded from score. Attempt funnel Attempts38 Scored36 Passed33 Fixtures solved33/38 Run facts Avg time 5.8s Repeat Single-shot Evidence share 36/38 (94.7%) Verified runs 1 95% interval — Raw label codex/gpt-5.6-luna@low Run id 55e3cd85-0c88-40a1-9364-114f6ce4bb36 Latest run fixtures 33/38 passed in linked run perf-find-ents-nearperf-string-table-concatperf-angle-zeroperf-looking-atplayer-iteratorents-iteratorhook-addisvalidnet-receiversperf-config-varperf-darkrpvarperf-disttosqrperf-ents-iteratorperf-find-players-nearperf-for-vs-whileperf-hook-onceperf-hudpaint-cacheperf-local-colorperf-local-vs-globalperf-localplayer-cacheperf-meta-vs-argperf-msg-vs-printperf-pairs-ipairs-forperf-plain-lua-vs-fnperf-player-cache-hooksperf-ply-table-indexperf-seq-tablesperf-setdrawcolor-splitperf-sqlite-vs-mysqlperf-steamid-cacheperf-surface-drawrectperf-surface-textperf-table-count-hashperf-table-emptyperf-table-hasvalueperf-table-literalperf-table-randomperf-x-times-x
15	Codex	gpt-5.5	Extra High	88.9%	94.7%	—	32/3/1/2	19.3s
#15 gpt-5.5 Extra High Codex Pass rate88.9% 32 pass · 36 scored Coverage94.7% 36 / 38 evidence-bearing Verified score— 95% interval — · 1 run(s) pass@k88.9% 32/38 fixtures ever solved Outcome mix Pass32 Partial3 Wrong1 Invalid output2 Invalid output means the model returned content that broke the answer contract, so it scores zero. Harness failures returned no answer and are excluded from score. Attempt funnel Attempts38 Scored36 Passed32 Fixtures solved32/38 Run facts Avg time 19.3s Repeat Single-shot Evidence share 36/38 (94.7%) Verified runs 1 95% interval — Raw label codex/gpt-5.5@xhigh Run id 55e3cd85-0c88-40a1-9364-114f6ce4bb36 Latest run fixtures 32/38 passed in linked run perf-ents-iteratorperf-string-table-concatperf-table-hasvalueperf-angle-zeroperf-localplayer-cacheperf-looking-atents-iteratorhook-addisvalidnet-receiversperf-config-varperf-darkrpvarperf-disttosqrperf-find-ents-nearperf-find-players-nearperf-for-vs-whileperf-hook-onceperf-hudpaint-cacheperf-local-colorperf-local-vs-globalperf-meta-vs-argperf-msg-vs-printperf-pairs-ipairs-forperf-plain-lua-vs-fnperf-player-cache-hooksperf-ply-table-indexperf-seq-tablesperf-setdrawcolor-splitperf-sqlite-vs-mysqlperf-steamid-cacheperf-surface-drawrectperf-surface-textperf-table-count-hashperf-table-emptyperf-table-literalperf-table-randomperf-x-times-xplayer-iterator
16	Codex	gpt-5.6-sol	High	88.9%	94.7%	—	32/3/1/2	15.0s
#16 gpt-5.6-sol High Codex Pass rate88.9% 32 pass · 36 scored Coverage94.7% 36 / 38 evidence-bearing Verified score— 95% interval — · 1 run(s) pass@k88.9% 32/38 fixtures ever solved Outcome mix Pass32 Partial3 Wrong1 Invalid output2 Invalid output means the model returned content that broke the answer contract, so it scores zero. Harness failures returned no answer and are excluded from score. Attempt funnel Attempts38 Scored36 Passed32 Fixtures solved32/38 Run facts Avg time 15.0s Repeat Single-shot Evidence share 36/38 (94.7%) Verified runs 1 95% interval — Raw label codex/gpt-5.6-sol@high Run id 55e3cd85-0c88-40a1-9364-114f6ce4bb36 Latest run fixtures 32/38 passed in linked run perf-hook-onceperf-string-table-concatperf-angle-zeroperf-find-ents-nearperf-localplayer-cacheperf-looking-atents-iteratorhook-addisvalidnet-receiversperf-config-varperf-darkrpvarperf-disttosqrperf-ents-iteratorperf-find-players-nearperf-for-vs-whileperf-hudpaint-cacheperf-local-colorperf-local-vs-globalperf-meta-vs-argperf-msg-vs-printperf-pairs-ipairs-forperf-plain-lua-vs-fnperf-player-cache-hooksperf-ply-table-indexperf-seq-tablesperf-setdrawcolor-splitperf-sqlite-vs-mysqlperf-steamid-cacheperf-surface-drawrectperf-surface-textperf-table-count-hashperf-table-emptyperf-table-hasvalueperf-table-literalperf-table-randomperf-x-times-xplayer-iterator
17	Codex	gpt-5.4	Medium	88.9%	94.7%	—	32/2/2/2	11.1s
#17 gpt-5.4 Medium Codex Pass rate88.9% 32 pass · 36 scored Coverage94.7% 36 / 38 evidence-bearing Verified score— 95% interval — · 1 run(s) pass@k88.9% 32/38 fixtures ever solved Outcome mix Pass32 Partial2 Wrong2 Invalid output2 Invalid output means the model returned content that broke the answer contract, so it scores zero. Harness failures returned no answer and are excluded from score. Attempt funnel Attempts38 Scored36 Passed32 Fixtures solved32/38 Run facts Avg time 11.1s Repeat Single-shot Evidence share 36/38 (94.7%) Verified runs 1 95% interval — Raw label codex/gpt-5.4@medium Run id 55e3cd85-0c88-40a1-9364-114f6ce4bb36 Latest run fixtures 32/38 passed in linked run perf-find-ents-nearperf-string-table-concatperf-plain-lua-vs-fnperf-ply-table-indexperf-angle-zeroperf-looking-atents-iteratorhook-addisvalidnet-receiversperf-config-varperf-darkrpvarperf-disttosqrperf-ents-iteratorperf-find-players-nearperf-for-vs-whileperf-hook-onceperf-hudpaint-cacheperf-local-colorperf-local-vs-globalperf-localplayer-cacheperf-meta-vs-argperf-msg-vs-printperf-pairs-ipairs-forperf-player-cache-hooksperf-seq-tablesperf-setdrawcolor-splitperf-sqlite-vs-mysqlperf-steamid-cacheperf-surface-drawrectperf-surface-textperf-table-count-hashperf-table-emptyperf-table-hasvalueperf-table-literalperf-table-randomperf-x-times-xplayer-iterator
18	Codex	gpt-5.6-sol	Max	88.9%	94.7%	—	32/2/2/2	12.4s
#18 gpt-5.6-sol Max Codex Pass rate88.9% 32 pass · 36 scored Coverage94.7% 36 / 38 evidence-bearing Verified score— 95% interval — · 1 run(s) pass@k88.9% 32/38 fixtures ever solved Outcome mix Pass32 Partial2 Wrong2 Invalid output2 Invalid output means the model returned content that broke the answer contract, so it scores zero. Harness failures returned no answer and are excluded from score. Attempt funnel Attempts38 Scored36 Passed32 Fixtures solved32/38 Run facts Avg time 12.4s Repeat Single-shot Evidence share 36/38 (94.7%) Verified runs 1 95% interval — Raw label codex/gpt-5.6-sol@max Run id 55e3cd85-0c88-40a1-9364-114f6ce4bb36 Latest run fixtures 32/38 passed in linked run perf-hook-onceperf-string-table-concatperf-angle-zeroperf-table-hasvalueperf-localplayer-cacheperf-looking-atents-iteratorhook-addisvalidnet-receiversperf-config-varperf-darkrpvarperf-disttosqrperf-ents-iteratorperf-find-ents-nearperf-find-players-nearperf-for-vs-whileperf-hudpaint-cacheperf-local-colorperf-local-vs-globalperf-meta-vs-argperf-msg-vs-printperf-pairs-ipairs-forperf-plain-lua-vs-fnperf-player-cache-hooksperf-ply-table-indexperf-seq-tablesperf-setdrawcolor-splitperf-sqlite-vs-mysqlperf-steamid-cacheperf-surface-drawrectperf-surface-textperf-table-count-hashperf-table-emptyperf-table-literalperf-table-randomperf-x-times-xplayer-iterator
19	Codex	gpt-5.4-mini	High	83.3%	94.7%	—	30/3/3/2	11.3s
#19 gpt-5.4-mini High Codex Pass rate83.3% 30 pass · 36 scored Coverage94.7% 36 / 38 evidence-bearing Verified score— 95% interval — · 1 run(s) pass@k83.3% 30/38 fixtures ever solved Outcome mix Pass30 Partial3 Wrong3 Invalid output2 Invalid output means the model returned content that broke the answer contract, so it scores zero. Harness failures returned no answer and are excluded from score. Attempt funnel Attempts38 Scored36 Passed30 Fixtures solved30/38 Run facts Avg time 11.3s Repeat Single-shot Evidence share 36/38 (94.7%) Verified runs 1 95% interval — Raw label codex/gpt-5.4-mini@high Run id 55e3cd85-0c88-40a1-9364-114f6ce4bb36 Latest run fixtures 30/38 passed in linked run perf-find-ents-nearperf-string-table-concatperf-angle-zeroperf-config-varperf-pairs-ipairs-forperf-localplayer-cacheperf-looking-atplayer-iteratorents-iteratorhook-addisvalidnet-receiversperf-darkrpvarperf-disttosqrperf-ents-iteratorperf-find-players-nearperf-for-vs-whileperf-hook-onceperf-hudpaint-cacheperf-local-colorperf-local-vs-globalperf-meta-vs-argperf-msg-vs-printperf-plain-lua-vs-fnperf-player-cache-hooksperf-ply-table-indexperf-seq-tablesperf-setdrawcolor-splitperf-sqlite-vs-mysqlperf-steamid-cacheperf-surface-drawrectperf-surface-textperf-table-count-hashperf-table-emptyperf-table-hasvalueperf-table-literalperf-table-randomperf-x-times-x
20	Codex	gpt-5.5	Low	83.3%	94.7%	—	30/3/3/2	12.0s
#20 gpt-5.5 Low Codex Pass rate83.3% 30 pass · 36 scored Coverage94.7% 36 / 38 evidence-bearing Verified score— 95% interval — · 1 run(s) pass@k83.3% 30/38 fixtures ever solved Outcome mix Pass30 Partial3 Wrong3 Invalid output2 Invalid output means the model returned content that broke the answer contract, so it scores zero. Harness failures returned no answer and are excluded from score. Attempt funnel Attempts38 Scored36 Passed30 Fixtures solved30/38 Run facts Avg time 12.0s Repeat Single-shot Evidence share 36/38 (94.7%) Verified runs 1 95% interval — Raw label codex/gpt-5.5@low Run id 55e3cd85-0c88-40a1-9364-114f6ce4bb36 Latest run fixtures 30/38 passed in linked run perf-find-ents-nearperf-string-table-concatperf-darkrpvarperf-msg-vs-printplayer-iteratorperf-angle-zeroperf-localplayer-cacheperf-looking-atents-iteratorhook-addisvalidnet-receiversperf-config-varperf-disttosqrperf-ents-iteratorperf-find-players-nearperf-for-vs-whileperf-hook-onceperf-hudpaint-cacheperf-local-colorperf-local-vs-globalperf-meta-vs-argperf-pairs-ipairs-forperf-plain-lua-vs-fnperf-player-cache-hooksperf-ply-table-indexperf-seq-tablesperf-setdrawcolor-splitperf-sqlite-vs-mysqlperf-steamid-cacheperf-surface-drawrectperf-surface-textperf-table-count-hashperf-table-emptyperf-table-hasvalueperf-table-literalperf-table-randomperf-x-times-x
21	Codex	gpt-5.6-luna	Extra High	94.3%	92.1%	—	33/2/0/3	9.8s
#21 gpt-5.6-luna Extra High Codex Pass rate94.3% 33 pass · 35 scored Coverage92.1% 35 / 38 evidence-bearing Verified score— 95% interval — · 1 run(s) pass@k94.3% 33/38 fixtures ever solved Outcome mix Pass33 Partial2 Wrong0 Invalid output3 Invalid output means the model returned content that broke the answer contract, so it scores zero. Harness failures returned no answer and are excluded from score. Attempt funnel Attempts38 Scored35 Passed33 Fixtures solved33/38 Run facts Avg time 9.8s Repeat Single-shot Evidence share 35/38 (92.1%) Verified runs 1 95% interval — Raw label codex/gpt-5.6-luna@xhigh Run id 55e3cd85-0c88-40a1-9364-114f6ce4bb36 Latest run fixtures 33/38 passed in linked run perf-angle-zeroperf-find-ents-nearperf-string-table-concatperf-localplayer-cacheperf-looking-atents-iteratorhook-addisvalidnet-receiversperf-config-varperf-darkrpvarperf-disttosqrperf-ents-iteratorperf-find-players-nearperf-for-vs-whileperf-hook-onceperf-hudpaint-cacheperf-local-colorperf-local-vs-globalperf-meta-vs-argperf-msg-vs-printperf-pairs-ipairs-forperf-plain-lua-vs-fnperf-player-cache-hooksperf-ply-table-indexperf-seq-tablesperf-setdrawcolor-splitperf-sqlite-vs-mysqlperf-steamid-cacheperf-surface-drawrectperf-surface-textperf-table-count-hashperf-table-emptyperf-table-hasvalueperf-table-literalperf-table-randomperf-x-times-xplayer-iterator
22	Codex	gpt-5.4	High	88.6%	92.1%	—	31/4/0/3	16.6s
#22 gpt-5.4 High Codex Pass rate88.6% 31 pass · 35 scored Coverage92.1% 35 / 38 evidence-bearing Verified score— 95% interval — · 1 run(s) pass@k88.6% 31/38 fixtures ever solved Outcome mix Pass31 Partial4 Wrong0 Invalid output3 Invalid output means the model returned content that broke the answer contract, so it scores zero. Harness failures returned no answer and are excluded from score. Attempt funnel Attempts38 Scored35 Passed31 Fixtures solved31/38 Run facts Avg time 16.6s Repeat Single-shot Evidence share 35/38 (92.1%) Verified runs 1 95% interval — Raw label codex/gpt-5.4@high Run id 55e3cd85-0c88-40a1-9364-114f6ce4bb36 Latest run fixtures 31/38 passed in linked run perf-find-ents-nearperf-hook-onceperf-string-table-concatperf-angle-zeroperf-darkrpvarperf-localplayer-cacheperf-looking-atents-iteratorhook-addisvalidnet-receiversperf-config-varperf-disttosqrperf-ents-iteratorperf-find-players-nearperf-for-vs-whileperf-hudpaint-cacheperf-local-colorperf-local-vs-globalperf-meta-vs-argperf-msg-vs-printperf-pairs-ipairs-forperf-plain-lua-vs-fnperf-player-cache-hooksperf-ply-table-indexperf-seq-tablesperf-setdrawcolor-splitperf-sqlite-vs-mysqlperf-steamid-cacheperf-surface-drawrectperf-surface-textperf-table-count-hashperf-table-emptyperf-table-hasvalueperf-table-literalperf-table-randomperf-x-times-xplayer-iterator
23	Codex	gpt-5.4	Low	85.7%	92.1%	—	30/4/1/3	5.1s
#23 gpt-5.4 Low Codex Pass rate85.7% 30 pass · 35 scored Coverage92.1% 35 / 38 evidence-bearing Verified score— 95% interval — · 1 run(s) pass@k85.7% 30/38 fixtures ever solved Outcome mix Pass30 Partial4 Wrong1 Invalid output3 Invalid output means the model returned content that broke the answer contract, so it scores zero. Harness failures returned no answer and are excluded from score. Attempt funnel Attempts38 Scored35 Passed30 Fixtures solved30/38 Run facts Avg time 5.1s Repeat Single-shot Evidence share 35/38 (92.1%) Verified runs 1 95% interval — Raw label codex/gpt-5.4@low Run id 55e3cd85-0c88-40a1-9364-114f6ce4bb36 Latest run fixtures 30/38 passed in linked run perf-find-ents-nearperf-hook-onceperf-string-table-concatperf-table-hasvalueperf-angle-zeroperf-darkrpvarperf-localplayer-cacheperf-looking-atents-iteratorhook-addisvalidnet-receiversperf-config-varperf-disttosqrperf-ents-iteratorperf-find-players-nearperf-for-vs-whileperf-hudpaint-cacheperf-local-colorperf-local-vs-globalperf-meta-vs-argperf-msg-vs-printperf-pairs-ipairs-forperf-plain-lua-vs-fnperf-player-cache-hooksperf-ply-table-indexperf-seq-tablesperf-setdrawcolor-splitperf-sqlite-vs-mysqlperf-steamid-cacheperf-surface-drawrectperf-surface-textperf-table-count-hashperf-table-emptyperf-table-literalperf-table-randomperf-x-times-xplayer-iterator
24	Codex	gpt-5.4	Extra High	80.0%	92.1%	—	28/4/3/3	16.6s
#24 gpt-5.4 Extra High Codex Pass rate80.0% 28 pass · 35 scored Coverage92.1% 35 / 38 evidence-bearing Verified score— 95% interval — · 1 run(s) pass@k80.0% 28/38 fixtures ever solved Outcome mix Pass28 Partial4 Wrong3 Invalid output3 Invalid output means the model returned content that broke the answer contract, so it scores zero. Harness failures returned no answer and are excluded from score. Attempt funnel Attempts38 Scored35 Passed28 Fixtures solved28/38 Run facts Avg time 16.6s Repeat Single-shot Evidence share 35/38 (92.1%) Verified runs 1 95% interval — Raw label codex/gpt-5.4@xhigh Run id 55e3cd85-0c88-40a1-9364-114f6ce4bb36 Latest run fixtures 28/38 passed in linked run perf-ents-iteratorperf-hook-onceperf-string-table-concatperf-config-varperf-table-hasvalueperf-table-randomperf-angle-zeroperf-find-ents-nearperf-localplayer-cacheperf-looking-atents-iteratorhook-addisvalidnet-receiversperf-darkrpvarperf-disttosqrperf-find-players-nearperf-for-vs-whileperf-hudpaint-cacheperf-local-colorperf-local-vs-globalperf-meta-vs-argperf-msg-vs-printperf-pairs-ipairs-forperf-plain-lua-vs-fnperf-player-cache-hooksperf-ply-table-indexperf-seq-tablesperf-setdrawcolor-splitperf-sqlite-vs-mysqlperf-steamid-cacheperf-surface-drawrectperf-surface-textperf-table-count-hashperf-table-emptyperf-table-literalperf-x-times-xplayer-iterator
25	Codex	gpt-5.5	High	88.2%	89.5%	—	30/3/1/4	15.7s
#25 gpt-5.5 High Codex Pass rate88.2% 30 pass · 34 scored Coverage89.5% 34 / 38 evidence-bearing Verified score— 95% interval — · 1 run(s) pass@k88.2% 30/38 fixtures ever solved Outcome mix Pass30 Partial3 Wrong1 Invalid output4 Invalid output means the model returned content that broke the answer contract, so it scores zero. Harness failures returned no answer and are excluded from score. Attempt funnel Attempts38 Scored34 Passed30 Fixtures solved30/38 Run facts Avg time 15.7s Repeat Single-shot Evidence share 34/38 (89.5%) Verified runs 1 95% interval — Raw label codex/gpt-5.5@high Run id 55e3cd85-0c88-40a1-9364-114f6ce4bb36 Latest run fixtures 30/38 passed in linked run perf-ents-iteratorperf-find-ents-nearperf-seq-tablesperf-string-table-concatperf-table-hasvalueperf-angle-zeroperf-localplayer-cacheperf-looking-atents-iteratorhook-addisvalidnet-receiversperf-config-varperf-darkrpvarperf-disttosqrperf-find-players-nearperf-for-vs-whileperf-hook-onceperf-hudpaint-cacheperf-local-colorperf-local-vs-globalperf-meta-vs-argperf-msg-vs-printperf-pairs-ipairs-forperf-plain-lua-vs-fnperf-player-cache-hooksperf-ply-table-indexperf-setdrawcolor-splitperf-sqlite-vs-mysqlperf-steamid-cacheperf-surface-drawrectperf-surface-textperf-table-count-hashperf-table-emptyperf-table-literalperf-table-randomperf-x-times-xplayer-iterator
26	Codex	gpt-5.4-mini	Low	85.3%	89.5%	—	29/3/2/4	7.9s
#26 gpt-5.4-mini Low Codex Pass rate85.3% 29 pass · 34 scored Coverage89.5% 34 / 38 evidence-bearing Verified score— 95% interval — · 1 run(s) pass@k85.3% 29/38 fixtures ever solved Outcome mix Pass29 Partial3 Wrong2 Invalid output4 Invalid output means the model returned content that broke the answer contract, so it scores zero. Harness failures returned no answer and are excluded from score. Attempt funnel Attempts38 Scored34 Passed29 Fixtures solved29/38 Run facts Avg time 7.9s Repeat Single-shot Evidence share 34/38 (89.5%) Verified runs 1 95% interval — Raw label codex/gpt-5.4-mini@low Run id 55e3cd85-0c88-40a1-9364-114f6ce4bb36 Latest run fixtures 29/38 passed in linked run perf-find-ents-nearperf-ply-table-indexperf-seq-tablesperf-string-table-concatperf-angle-zeroperf-table-hasvalueperf-localplayer-cacheperf-looking-atplayer-iteratorents-iteratorhook-addisvalidnet-receiversperf-config-varperf-darkrpvarperf-disttosqrperf-ents-iteratorperf-find-players-nearperf-for-vs-whileperf-hook-onceperf-hudpaint-cacheperf-local-colorperf-local-vs-globalperf-meta-vs-argperf-msg-vs-printperf-pairs-ipairs-forperf-plain-lua-vs-fnperf-player-cache-hooksperf-setdrawcolor-splitperf-sqlite-vs-mysqlperf-steamid-cacheperf-surface-drawrectperf-surface-textperf-table-count-hashperf-table-emptyperf-table-literalperf-table-randomperf-x-times-x
27	Codex	gpt-5.5	Medium	90.9%	86.8%	—	30/2/1/5	6.9s
#27 gpt-5.5 Medium Codex Pass rate90.9% 30 pass · 33 scored Coverage86.8% 33 / 38 evidence-bearing Verified score— 95% interval — · 1 run(s) pass@k90.9% 30/38 fixtures ever solved Outcome mix Pass30 Partial2 Wrong1 Invalid output5 Invalid output means the model returned content that broke the answer contract, so it scores zero. Harness failures returned no answer and are excluded from score. Attempt funnel Attempts38 Scored33 Passed30 Fixtures solved30/38 Run facts Avg time 6.9s Repeat Single-shot Evidence share 33/38 (86.8%) Verified runs 1 95% interval — Raw label codex/gpt-5.5@medium Run id 55e3cd85-0c88-40a1-9364-114f6ce4bb36 Latest run fixtures 30/38 passed in linked run perf-ents-iteratorperf-find-ents-nearperf-hook-onceperf-ply-table-indexperf-string-table-concatperf-plain-lua-vs-fnperf-localplayer-cacheperf-looking-atents-iteratorhook-addisvalidnet-receiversperf-angle-zeroperf-config-varperf-darkrpvarperf-disttosqrperf-find-players-nearperf-for-vs-whileperf-hudpaint-cacheperf-local-colorperf-local-vs-globalperf-meta-vs-argperf-msg-vs-printperf-pairs-ipairs-forperf-player-cache-hooksperf-seq-tablesperf-setdrawcolor-splitperf-sqlite-vs-mysqlperf-steamid-cacheperf-surface-drawrectperf-surface-textperf-table-count-hashperf-table-emptyperf-table-hasvalueperf-table-literalperf-table-randomperf-x-times-xplayer-iterator
28	Codex	gpt-5.3-codex-spark	High	75.0%	84.2%	—	24/3/5/6	10.9s
#28 gpt-5.3-codex-spark High Codex Pass rate75.0% 24 pass · 32 scored Coverage84.2% 32 / 38 evidence-bearing Verified score— 95% interval — · 1 run(s) pass@k75.0% 24/38 fixtures ever solved Outcome mix Pass24 Partial3 Wrong5 Invalid output6 Invalid output means the model returned content that broke the answer contract, so it scores zero. Harness failures returned no answer and are excluded from score. Attempt funnel Attempts38 Scored32 Passed24 Fixtures solved24/38 Run facts Avg time 10.9s Repeat Single-shot Evidence share 32/38 (84.2%) Verified runs 1 95% interval — Raw label codex/gpt-5.3-codex-spark@high Run id 55e3cd85-0c88-40a1-9364-114f6ce4bb36 Latest run fixtures 24/38 passed in linked run perf-angle-zeroperf-ents-iteratorperf-find-ents-nearperf-seq-tablesperf-setdrawcolor-splitperf-string-table-concatperf-config-varperf-pairs-ipairs-forperf-ply-table-indexperf-surface-textperf-table-randomperf-darkrpvarperf-looking-atplayer-iteratorents-iteratorhook-addisvalidnet-receiversperf-disttosqrperf-find-players-nearperf-for-vs-whileperf-hook-onceperf-hudpaint-cacheperf-local-colorperf-local-vs-globalperf-localplayer-cacheperf-meta-vs-argperf-msg-vs-printperf-plain-lua-vs-fnperf-player-cache-hooksperf-sqlite-vs-mysqlperf-steamid-cacheperf-surface-drawrectperf-table-count-hashperf-table-emptyperf-table-hasvalueperf-table-literalperf-x-times-x
29	Codex	gpt-5.3-codex-spark	Extra High	75.0%	84.2%	—	24/3/5/6	12.4s
#29 gpt-5.3-codex-spark Extra High Codex Pass rate75.0% 24 pass · 32 scored Coverage84.2% 32 / 38 evidence-bearing Verified score— 95% interval — · 1 run(s) pass@k75.0% 24/38 fixtures ever solved Outcome mix Pass24 Partial3 Wrong5 Invalid output6 Invalid output means the model returned content that broke the answer contract, so it scores zero. Harness failures returned no answer and are excluded from score. Attempt funnel Attempts38 Scored32 Passed24 Fixtures solved24/38 Run facts Avg time 12.4s Repeat Single-shot Evidence share 32/38 (84.2%) Verified runs 1 95% interval — Raw label codex/gpt-5.3-codex-spark@xhigh Run id 55e3cd85-0c88-40a1-9364-114f6ce4bb36 Latest run fixtures 24/38 passed in linked run perf-angle-zeroperf-find-ents-nearperf-ply-table-indexperf-string-table-concatperf-table-hasvalueperf-table-literalperf-config-varperf-meta-vs-argperf-msg-vs-printperf-pairs-ipairs-forperf-table-randomperf-darkrpvarperf-localplayer-cacheplayer-iteratorents-iteratorhook-addisvalidnet-receiversperf-disttosqrperf-ents-iteratorperf-find-players-nearperf-for-vs-whileperf-hook-onceperf-hudpaint-cacheperf-local-colorperf-local-vs-globalperf-looking-atperf-plain-lua-vs-fnperf-player-cache-hooksperf-seq-tablesperf-setdrawcolor-splitperf-sqlite-vs-mysqlperf-steamid-cacheperf-surface-drawrectperf-surface-textperf-table-count-hashperf-table-emptyperf-x-times-x
30	Codex	gpt-5.3-codex-spark	Low	74.2%	81.6%	—	23/4/4/7	9.8s
#30 gpt-5.3-codex-spark Low Codex Pass rate74.2% 23 pass · 31 scored Coverage81.6% 31 / 38 evidence-bearing Verified score— 95% interval — · 1 run(s) pass@k74.2% 23/38 fixtures ever solved Outcome mix Pass23 Partial4 Wrong4 Invalid output7 Invalid output means the model returned content that broke the answer contract, so it scores zero. Harness failures returned no answer and are excluded from score. Attempt funnel Attempts38 Scored31 Passed23 Fixtures solved23/38 Run facts Avg time 9.8s Repeat Single-shot Evidence share 31/38 (81.6%) Verified runs 1 95% interval — Raw label codex/gpt-5.3-codex-spark@low Run id 55e3cd85-0c88-40a1-9364-114f6ce4bb36 Latest run fixtures 23/38 passed in linked run perf-find-ents-nearperf-ply-table-indexperf-seq-tablesperf-setdrawcolor-splitperf-string-table-concatperf-surface-drawrectperf-table-hasvalueperf-config-varperf-msg-vs-printperf-pairs-ipairs-forperf-plain-lua-vs-fnperf-angle-zeroperf-localplayer-cacheperf-looking-atplayer-iteratorents-iteratorhook-addisvalidnet-receiversperf-darkrpvarperf-disttosqrperf-ents-iteratorperf-find-players-nearperf-for-vs-whileperf-hook-onceperf-hudpaint-cacheperf-local-colorperf-local-vs-globalperf-meta-vs-argperf-player-cache-hooksperf-sqlite-vs-mysqlperf-steamid-cacheperf-surface-textperf-table-count-hashperf-table-emptyperf-table-literalperf-table-randomperf-x-times-x
31	Codex	gpt-5.3-codex-spark	Medium	74.2%	81.6%	—	23/4/4/7	4.8s
#31 gpt-5.3-codex-spark Medium Codex Pass rate74.2% 23 pass · 31 scored Coverage81.6% 31 / 38 evidence-bearing Verified score— 95% interval — · 1 run(s) pass@k74.2% 23/38 fixtures ever solved Outcome mix Pass23 Partial4 Wrong4 Invalid output7 Invalid output means the model returned content that broke the answer contract, so it scores zero. Harness failures returned no answer and are excluded from score. Attempt funnel Attempts38 Scored31 Passed23 Fixtures solved23/38 Run facts Avg time 4.8s Repeat Single-shot Evidence share 31/38 (81.6%) Verified runs 1 95% interval — Raw label codex/gpt-5.3-codex-spark@medium Run id 55e3cd85-0c88-40a1-9364-114f6ce4bb36 Latest run fixtures 23/38 passed in linked run perf-ents-iteratorperf-find-ents-nearperf-ply-table-indexperf-seq-tablesperf-string-table-concatperf-surface-drawrectperf-table-hasvalueperf-config-varperf-msg-vs-printperf-pairs-ipairs-forperf-table-randomperf-angle-zeroperf-localplayer-cacheperf-looking-atplayer-iteratorents-iteratorhook-addisvalidnet-receiversperf-darkrpvarperf-disttosqrperf-find-players-nearperf-for-vs-whileperf-hook-onceperf-hudpaint-cacheperf-local-colorperf-local-vs-globalperf-meta-vs-argperf-plain-lua-vs-fnperf-player-cache-hooksperf-setdrawcolor-splitperf-sqlite-vs-mysqlperf-steamid-cacheperf-surface-textperf-table-count-hashperf-table-emptyperf-table-literalperf-x-times-x

Verified score aggregates compatible runs. Outcomes = pass / partial / wrong / invalid output; harness failures are excluded. Click a row for evidence and the linked run's fixtures.

Fixtures

gmod.ents-iterator.v1
gmod.hook-add.v1
gmod.isvalid.v1
gmod.net-receivers.v1
gmod.perf.angle-zero.v1
gmod.perf.config-var.v1
gmod.perf.darkrpvar.v1
gmod.perf.disttosqr.v1
gmod.perf.ents-iterator.v1
gmod.perf.find-ents-near.v1
gmod.perf.find-players-near.v1
gmod.perf.for-vs-while.v1
gmod.perf.hook-once.v1
gmod.perf.hudpaint-cache.v1
gmod.perf.local-color.v1
gmod.perf.local-vs-global.v1
gmod.perf.localplayer-cache.v1
gmod.perf.looking-at.v1
gmod.perf.meta-vs-arg.v1
gmod.perf.msg-vs-print.v1
gmod.perf.pairs-ipairs-for.v1
gmod.perf.plain-lua-vs-fn.v1
gmod.perf.player-cache-hooks.v1
gmod.perf.ply-table-index.v1
gmod.perf.seq-tables.v1
gmod.perf.setdrawcolor-split.v1
gmod.perf.sqlite-vs-mysql.v1
gmod.perf.steamid-cache.v1
gmod.perf.string-table-concat.v1
gmod.perf.surface-drawrect.v1
gmod.perf.surface-text.v1
gmod.perf.table-count-hash.v1
gmod.perf.table-empty.v1
gmod.perf.table-hasvalue.v1
gmod.perf.table-literal.v1
gmod.perf.table-random.v1
gmod.perf.x-times-x.v1
gmod.player-iterator.v1

Downloads

Results

Outcome mix

Attempt funnel

Run facts

Latest run fixtures

Outcome mix

Attempt funnel

Run facts

Latest run fixtures

Outcome mix

Attempt funnel

Run facts

Latest run fixtures

Outcome mix

Attempt funnel

Run facts

Latest run fixtures

Outcome mix

Attempt funnel

Run facts

Latest run fixtures

Outcome mix

Attempt funnel

Run facts

Latest run fixtures

Outcome mix

Attempt funnel

Run facts

Latest run fixtures

Outcome mix

Attempt funnel

Run facts

Latest run fixtures

Outcome mix

Attempt funnel

Run facts

Latest run fixtures

Outcome mix

Attempt funnel

Run facts

Latest run fixtures

Outcome mix

Attempt funnel

Run facts

Latest run fixtures

Outcome mix

Attempt funnel

Run facts

Latest run fixtures

Outcome mix

Attempt funnel

Run facts

Latest run fixtures

Outcome mix

Attempt funnel

Run facts

Latest run fixtures

Outcome mix

Attempt funnel

Run facts

Latest run fixtures

Outcome mix

Attempt funnel

Run facts

Latest run fixtures

Outcome mix

Attempt funnel

Run facts

Latest run fixtures

Outcome mix

Attempt funnel

Run facts

Latest run fixtures

Outcome mix

Attempt funnel

Run facts

Latest run fixtures

Outcome mix

Attempt funnel