Rerun SubQ with full method context

README.md

@@ -51,7 +51,7 @@ Failed requests count as incorrect decisions: a failed issue is a miss; a failed
 | --- | ---: | ---: | ---: | ---: |
 | [GPT-6 Luna Decisions](https://developers.openai.com/api/docs/guides/decisions) | 48.7% | 38.7% | **10.0%** | $1.102 |
 | [DiffusionGemma Jev](https://github.com/razorback16/openjev) | 33.4% | 27.1% | 6.3% | $0.239 |
-| [SubQ Snap Preview](https://docs.subq.ai/using-the-api/models/) | 15.3% | 9.9% | 5.3% | $0.568 |
+| [SubQ Snap Preview](https://docs.subq.ai/using-the-api/models/) | 15.2% | 9.5% | 5.7% | $0.600 |
 | [Microsoft-Decision-1](https://openrouter.ai/microsoft/microsoft-decision-1) | 24.0% | 18.8% | 5.1% | $0.481 |
 | [Jev](https://typesafe.ai/blog/introducing-system-one-models-and-jev) | 17.7% | 12.7% | 5.0% | $0.551 |
 | [SemiF](https://github.com/TheoLeeCJ/SemIf) | 43.0% | 39.7% | 3.3% | $0.239 |
@@ -71,7 +71,7 @@ Failed requests count as incorrect decisions: a failed issue is a miss; a failed
 | --- | ---: | ---: | ---: | ---: |
 | [GPT-6 Luna Decisions](https://developers.openai.com/api/docs/guides/decisions) | 52.4% | 41.3% | **11.1%** | $0.247 |
 | [DiffusionGemma Jev](https://github.com/razorback16/openjev) | 35.8% | 27.3% | 8.6% | $0.054 |
-| [SubQ Snap Preview](https://docs.subq.ai/using-the-api/models/) | 19.3% | 11.2% | 8.2% | $0.133 |
+| [SubQ Snap Preview](https://docs.subq.ai/using-the-api/models/) | 19.3% | 11.2% | 8.2% | $0.134 |
 | [TypeLLM](https://typellm.ai/docs) | 11.3% | 5.8% | 5.6% | $0.129 |
 | [Microsoft-Decision-1](https://openrouter.ai/microsoft/microsoft-decision-1) | 25.5% | 20.6% | 4.8% | $0.108 |
 | [Jev](https://typesafe.ai/blog/introducing-system-one-models-and-jev) | 20.0% | 15.6% | 4.5% | $0.123 |
@@ -97,7 +97,7 @@ Failed requests count as incorrect decisions: a failed issue is a miss; a failed
 | [TypeLLM](https://typellm.ai/docs) | 27.8% | 17.4% | 10.5% | $0.566 |
 | [GPT-6 Luna Decisions](https://developers.openai.com/api/docs/guides/decisions) | 54.1% | 44.5% | 9.7% | $1.469 |
 | [Mercury Decide](https://openrouter.ai/inception/mercury-decide) | 27.1% | 17.8% | 9.3% | **$0.131** |
-| [SubQ Snap Preview](https://docs.subq.ai/using-the-api/models/) | 24.7% | 15.9% | 8.8% | $0.817 |
+| [SubQ Snap Preview](https://docs.subq.ai/using-the-api/models/) | 24.1% | 15.5% | 8.6% | $0.956 |
 | [DiffusionGemma Jev](https://github.com/razorback16/openjev) | 33.1% | 25.8% | 7.3% | $0.208 |
 | [Microsoft-Decision-1](https://openrouter.ai/microsoft/microsoft-decision-1) | 43.9% | 38.2% | 5.6% | $0.455 |
 | [Span-01](https://www.respan.ai/docs/documentation/span-01/concept) | 21.0% | 17.2% | 3.9% | $0.251 |
@@ -116,9 +116,9 @@ Failed requests count as incorrect decisions: a failed issue is a miss; a failed
 | [Liquid d1](https://openrouter.ai/liquid/d1) | 34.7% | 29.4% | 5.3% | $2.362 |
 | [Jev](https://typesafe.ai/blog/introducing-system-one-models-and-jev) | 20.6% | 16.7% | 3.9% | $0.182 |
 | [DiffusionGemma Jev](https://github.com/razorback16/openjev) | 21.9% | 18.5% | 3.3% | $0.074 |
+| [SubQ Snap Preview](https://docs.subq.ai/using-the-api/models/) | 11.1% | 8.3% | 2.7% | $0.344 |
 | [TypeLLM](https://typellm.ai/docs) | 8.8% | 6.2% | 2.5% | $0.203 |
 | [Mercury Decide](https://openrouter.ai/inception/mercury-decide) | 11.1% | 9.0% | 2.1% | **$0.047** |
-| [SubQ Snap Preview](https://docs.subq.ai/using-the-api/models/) | 10.3% | 8.3% | 2.0% | $0.278 |
 | [Microsoft-Decision-1](https://openrouter.ai/microsoft/microsoft-decision-1) | 18.1% | 16.7% | 1.4% | $0.163 |
 | [GLiNER-2.5-Decide](https://huggingface.co/fastino/GLiNER2.5-Decide) | 2.0% | **1.5%** | 0.6% | $0.142 |
 | [CLM-v0.1-8B](https://huggingface.co/Contrastive-LM/CLM-v0.1-8B) | **100.0%** | 100.0% | 0.0% | ~$0.528 |
@@ -137,7 +137,7 @@ Failed requests count as incorrect decisions: a failed issue is a miss; a failed
 | --- | ---: | ---: | ---: | ---: |
 | [Mercury Decide](https://openrouter.ai/inception/mercury-decide) | **81.8%** | 6.5% | **75.2%** | $0.020 |
 | [GPT-6 Luna Decisions](https://developers.openai.com/api/docs/guides/decisions) | 76.6% | 4.7% | 72.0% | $0.093 |
-| [SubQ Snap Preview](https://docs.subq.ai/using-the-api/models/) | 71.5% | 2.8% | 68.7% | $0.045 |
+| [SubQ Snap Preview](https://docs.subq.ai/using-the-api/models/) | 71.0% | 2.8% | 68.2% | $0.053 |
 | [TypeLLM](https://typellm.ai/docs) | 61.7% | 2.8% | 58.9% | $0.050 |
 | [Jev](https://typesafe.ai/blog/introducing-system-one-models-and-jev) | 50.0% | 0.5% | 49.5% | $0.048 |
 | [DiffusionGemma Jev](https://github.com/razorback16/openjev) | 52.8% | 5.1% | 47.7% | $0.018 |
@@ -154,7 +154,7 @@ Failed requests count as incorrect decisions: a failed issue is a miss; a failed
 | --- | ---: | ---: | ---: | ---: |
 | [Mercury Decide](https://openrouter.ai/inception/mercury-decide) | **87.3%** | 8.5% | **78.9%** | $0.006 |
 | [GPT-6 Luna Decisions](https://developers.openai.com/api/docs/guides/decisions) | 84.5% | 7.0% | 77.5% | $0.030 |
-| [SubQ Snap Preview](https://docs.subq.ai/using-the-api/models/) | 71.8% | 4.2% | 67.6% | $0.016 |
+| [SubQ Snap Preview](https://docs.subq.ai/using-the-api/models/) | 71.8% | 5.6% | 66.2% | $0.017 |
 | [TypeLLM](https://typellm.ai/docs) | 59.2% | 5.6% | 53.5% | $0.016 |
 | [Jev](https://typesafe.ai/blog/introducing-system-one-models-and-jev) | 53.5% | 1.4% | 52.1% | $0.016 |
 | [DiffusionGemma Jev](https://github.com/razorback16/openjev) | 53.5% | 8.5% | 45.1% | $0.006 |
@@ -173,7 +173,7 @@ Span-01 uses its [behavior-scoring API](https://www.respan.ai/docs/documentation
 
 Liquid d1 received the same Perch method state and all applicable questions in one API call per method. Semantic lint used the same compiled `ensure` rule as Jev.
 
-SubQ Snap Preview uses its native noul probabilities with an 8,192-token limit reported by the server. Perch sizes the method context and batches its questions; large methods use overlapping source readings that cover every target-source byte. Each question’s highest issue score supplies the method decision. Input-token usage returned by SubQ is priced at its [published $0.05 per million rate](https://docs.subq.ai/using-the-api/models/).
+SubQ Snap Preview scores the full saved Perch context through native noul questions, in batches of at most 16. Its [documented context window](https://docs.subq.ai/endpoints/decisions-api/) is 393,216 tokens. Perch retains the method source, imports, callers, callees and graph edges within that budget. Input-token usage returned by SubQ is priced at its [published $0.05 per million rate](https://docs.subq.ai/using-the-api/models/).
 
 GLiNER-2.5-Decide scores the answer criteria as native classification labels. Perch sizes each method's context and batches its applicable questions to fit the endpoint. Large methods are read in overlapping source chunks; each question's highest issue score supplies the method decision. Returned label confidences supply the scores. Timeouts and requests that still cannot fit are recorded as failures. Costs use API-reported prompt tokens at the live catalog rate of $0.03 per million input tokens.
 

predictions/current/security/subq-snap.jsonl

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a61de1126c4ec1bf4a53c1549643bde0253144bc436dab29b4c8bfb618d47994
-size 17403214
+oid sha256:5a94f2fa0725516f73dea7c05b485a757489fd448353f697461426969c6a6a24
+size 17004709

results-current.jsonl

@@ -64,9 +64,9 @@
 {"alert_rule": "any issue score > max(0.5, question min)", "answered_methods": 996, "average_precision": 0.4758656734035225, "average_precision_answered_methods": 0.4758656734035225, "benchmark_methods": 996, "benchmark_pairs": 498, "benchmark_sha256": "3df4ade65fedd1128d30bc17c9c9869c8ab9bd526e666156d0e16685e78f8a73", "complete": true, "confusion": {"fn": 396, "fp": 24, "tn": 548, "tp": 28}, "context_limit_tokens": 65536, "domain": "bug", "f1": 0.11764705882352941, "failed_methods": 0, "failure_counts": {}, "failure_counts_per_question": {}, "failure_policy": "failed positive counts as FN; failed negative counts as FP; failed pair counts as incorrect ordering", "false_positive_rate": 0.04195804195804196, "inference_cost_basis": "OpenRouter API-reported usage.cost", "inference_cost_usd": 0.19667368, "inference_endpoint": "https://openrouter.ai/api/v1/systemone", "inference_host": "OpenRouter / Liquid", "input_tokens": 4916842, "latency_basis": "Per-method elapsed time, including client rate-limit waits and 429 backoff", "latency_median_seconds": 1.7929241659585387, "latency_p95_seconds": 2.340836458024569, "model_id": "liquid-d1", "model_revision": "liquid/d1-20260930", "pair_order": 0.5062344139650873, "pair_order_correct": 203, "pair_order_eligible": 401, "precision": 0.5384615384615384, "protocol_id": "liquid-d1-openrouter-perch-systemone-v1", "question_pack_sha256": "afe116cc8461552aaca996f8f0b639cdf5e42e9d8291c59995482309b5c330dc", "recall": 0.0660377358490566, "refused_methods": 0, "requested_model": "liquid/d1", "run_id": "liquid-d1-openrouter-20261012", "threshold": 0.5, "track": "private", "vulnerable_or_buggy_before_above_fixed_after": {"correct": 203, "eligible": 401, "ties": 5}}
 {"alert_rule": "any issue score > max(0.5, question min)", "answered_methods": 2622, "average_precision": 0.49708136950251325, "average_precision_answered_methods": 0.49708136950251325, "benchmark_methods": 2622, "benchmark_pairs": 1311, "benchmark_sha256": "2407059caa3486371749884138b997d89043eec646e548ee8f44ba3ab08514b8", "complete": true, "confusion": {"fn": 545, "fp": 603, "tn": 877, "tp": 597}, "context_limit_tokens": 65536, "domain": "security", "f1": 0.5098206660973527, "failed_methods": 0, "failure_counts": {}, "failure_counts_per_question": {}, "failure_policy": "failed positive counts as FN; failed negative counts as FP; failed pair counts as incorrect ordering", "false_positive_rate": 0.40743243243243243, "inference_cost_basis": "OpenRouter API-reported usage.cost", "inference_cost_usd": 6.13173568, "inference_endpoint": "https://openrouter.ai/api/v1/systemone", "inference_host": "OpenRouter / Liquid", "input_tokens": 153293392, "latency_basis": "Per-method elapsed time, including client rate-limit waits and 429 backoff", "latency_median_seconds": 2.0774508339818567, "latency_p95_seconds": 2.837762249982916, "model_id": "liquid-d1", "model_revision": "liquid/d1-20260930", "pair_order": 0.5284697508896797, "pair_order_correct": 594, "pair_order_eligible": 1124, "precision": 0.4975, "protocol_id": "liquid-d1-openrouter-perch-systemone-v1", "question_pack_sha256": "afe116cc8461552aaca996f8f0b639cdf5e42e9d8291c59995482309b5c330dc", "recall": 0.5227670753064798, "refused_methods": 0, "requested_model": "liquid/d1", "run_id": "liquid-d1-openrouter-20261012", "threshold": 0.5, "track": "public", "vulnerable_or_buggy_before_above_fixed_after": {"correct": 594, "eligible": 1124, "ties": 15}}
 {"alert_rule": "any issue score > max(0.5, question min)", "answered_methods": 878, "average_precision": 0.4793341864639895, "average_precision_answered_methods": 0.4793341864639895, "benchmark_methods": 878, "benchmark_pairs": 439, "benchmark_sha256": "a4b81728e9e520cf1c6f342974bec369d72f5b7d8aabc1c9cf2aceba966f8051", "complete": true, "confusion": {"fn": 260, "fp": 141, "tn": 339, "tp": 138}, "context_limit_tokens": 65536, "domain": "security", "f1": 0.4076809453471196, "failed_methods": 0, "failure_counts": {}, "failure_counts_per_question": {}, "failure_policy": "failed positive counts as FN; failed negative counts as FP; failed pair counts as incorrect ordering", "false_positive_rate": 0.29375, "inference_cost_basis": "OpenRouter API-reported usage.cost", "inference_cost_usd": 2.36226772, "inference_endpoint": "https://openrouter.ai/api/v1/systemone", "inference_host": "OpenRouter / Liquid", "input_tokens": 59056693, "latency_basis": "Per-method elapsed time, including client rate-limit waits and 429 backoff", "latency_median_seconds": 2.1423684580950066, "latency_p95_seconds": 2.9545628749765456, "model_id": "liquid-d1", "model_revision": "liquid/d1-20260930", "pair_order": 0.4547738693467337, "pair_order_correct": 181, "pair_order_eligible": 398, "precision": 0.4946236559139785, "protocol_id": "liquid-d1-openrouter-perch-systemone-v1", "question_pack_sha256": "afe116cc8461552aaca996f8f0b639cdf5e42e9d8291c59995482309b5c330dc", "recall": 0.34673366834170855, "refused_methods": 0, "requested_model": "liquid/d1", "run_id": "liquid-d1-openrouter-20261012", "threshold": 0.5, "track": "private", "vulnerable_or_buggy_before_above_fixed_after": {"correct": 181, "eligible": 398, "ties": 4}}
-{"alert_rule": "rule violation probability > 0.7", "answered_methods": 428, "average_precision": 0.954599215445173, "average_precision_answered_methods": 0.954599215445173, "benchmark_methods": 428, "benchmark_pairs": 214, "benchmark_sha256": "534f6c2abddc507c40f4597607e97e07168c544011d617b83aa39e08abb35d75", "complete": true, "confusion": {"fn": 61, "fp": 6, "tn": 208, "tp": 153}, "context_limit_basis": "SubQ server context-length validation: maximum model length 8192", "context_limit_tokens": 8192, "domain": "lint", "f1": 0.8203753351206434, "failed_methods": 0, "failure_counts": {}, "failure_counts_per_question": {}, "failure_policy": "failed positive counts as FN; failed negative counts as FP; failed pair counts as incorrect ordering", "false_positive_rate": 0.028037383177570093, "inference_cost_basis": "API-reported input tokens at SubQ published $0.05 per million input tokens", "inference_cost_usd": 0.045379050000000004, "inference_endpoint": "https://api.subq.ai/v1/systemone", "inference_host": "SubQ", "input_tokens": 907581, "latency_basis": "Per-method elapsed time, including client rate-limit waits and 429 backoff", "latency_median_seconds": 1.8166153749916703, "latency_p95_seconds": 2.551551292068325, "model_id": "subq-snap", "model_revision": "subq-snap-preview", "pair_order": 0.9672897196261683, "pair_order_correct": 207, "pair_order_eligible": 214, "perch_revision": "4af67af46d76ac99398d61c7ad0a45441760d4dc", "perch_sizing_bundle_sha256": "17592731f69d88e2d1dda3f93b52eab2e3d2fc440d6b3bf0af6583e718cbd51a", "precision": 0.9622641509433962, "protocol_id": "subq-snap-perch-sized-systemone-v1", "question_pack_sha256": null, "recall": 0.7149532710280374, "refused_methods": 0, "requested_model": "subq-snap-preview", "run_id": "subq-snap-20261012", "sizing": {"adapter_sha256": "cef0475d20258b14a8820ce734cfcd422f397131b18ecf3fba5b40c5ec9d7b87", "request_budget": 6500, "state_budget": 4500}, "threshold": 0.7, "track": "public", "vulnerable_or_buggy_before_above_fixed_after": {"correct": 207, "eligible": 214, "ties": 0}}
-{"alert_rule": "rule violation probability > 0.7", "answered_methods": 142, "average_precision": 0.9686294456382798, "average_precision_answered_methods": 0.9686294456382798, "benchmark_methods": 142, "benchmark_pairs": 71, "benchmark_sha256": "a6e9f2e4ee8ae77bdaf23af46920d97d1ba47375dc9ce03b950cdbe95c2777f9", "complete": true, "confusion": {"fn": 20, "fp": 3, "tn": 68, "tp": 51}, "context_limit_basis": "SubQ server context-length validation: maximum model length 8192", "context_limit_tokens": 8192, "domain": "lint", "f1": 0.816, "failed_methods": 0, "failure_counts": {}, "failure_counts_per_question": {}, "failure_policy": "failed positive counts as FN; failed negative counts as FP; failed pair counts as incorrect ordering", "false_positive_rate": 0.04225352112676056, "inference_cost_basis": "API-reported input tokens at SubQ published $0.05 per million input tokens", "inference_cost_usd": 0.015941200000000003, "inference_endpoint": "https://api.subq.ai/v1/systemone", "inference_host": "SubQ", "input_tokens": 318824, "latency_basis": "Per-method elapsed time, including client rate-limit waits and 429 backoff", "latency_median_seconds": 2.152747832937166, "latency_p95_seconds": 2.732683040900156, "model_id": "subq-snap", "model_revision": "subq-snap-preview", "pair_order": 0.9577464788732394, "pair_order_correct": 68, "pair_order_eligible": 71, "perch_revision": "4af67af46d76ac99398d61c7ad0a45441760d4dc", "perch_sizing_bundle_sha256": "17592731f69d88e2d1dda3f93b52eab2e3d2fc440d6b3bf0af6583e718cbd51a", "precision": 0.9444444444444444, "protocol_id": "subq-snap-perch-sized-systemone-v1", "question_pack_sha256": null, "recall": 0.7183098591549296, "refused_methods": 0, "requested_model": "subq-snap-preview", "run_id": "subq-snap-20261012", "sizing": {"adapter_sha256": "cef0475d20258b14a8820ce734cfcd422f397131b18ecf3fba5b40c5ec9d7b87", "request_budget": 6500, "state_budget": 4500}, "threshold": 0.7, "track": "private", "vulnerable_or_buggy_before_above_fixed_after": {"correct": 68, "eligible": 71, "ties": 0}}
-{"alert_rule": "any issue score > max(0.5, question min)", "answered_methods": 4426, "average_precision": 0.48487100893482865, "average_precision_answered_methods": 0.48487100893482865, "benchmark_methods": 4440, "benchmark_pairs": 2220, "benchmark_sha256": "32bc68bc85d5535ad28e103adbfa0c09c73e09d0f53a53f7ca35ec219a96e6e2", "complete": true, "confusion": {"fn": 1577, "fp": 256, "tn": 2323, "tp": 284}, "context_limit_basis": "SubQ server context-length validation: maximum model length 8192", "context_limit_tokens": 8192, "domain": "bug", "f1": 0.23656809662640566, "failed_methods": 14, "failure_counts": {"Context sizing failed": 14}, "failure_counts_per_question": {}, "failure_policy": "failed positive counts as FN; failed negative counts as FP; failed pair counts as incorrect ordering", "false_positive_rate": 0.09926328034121752, "inference_cost_basis": "API-reported input tokens at SubQ published $0.05 per million input tokens", "inference_cost_usd": 0.56760645, "inference_endpoint": "https://api.subq.ai/v1/systemone", "inference_host": "SubQ", "input_tokens": 11352129, "latency_basis": "Per-method elapsed time, including client rate-limit waits and 429 backoff", "latency_median_seconds": 1.812008082983084, "latency_p95_seconds": 2.3090446670539677, "model_id": "subq-snap", "model_revision": "subq-snap-preview", "pair_order": 0.5095398428731762, "pair_order_correct": 908, "pair_order_eligible": 1782, "perch_revision": "4af67af46d76ac99398d61c7ad0a45441760d4dc", "perch_sizing_bundle_sha256": "17592731f69d88e2d1dda3f93b52eab2e3d2fc440d6b3bf0af6583e718cbd51a", "precision": 0.5259259259259259, "protocol_id": "subq-snap-perch-sized-systemone-v1", "question_pack_sha256": "afe116cc8461552aaca996f8f0b639cdf5e42e9d8291c59995482309b5c330dc", "recall": 0.15260612573885007, "refused_methods": 0, "requested_model": "subq-snap-preview", "run_id": "subq-snap-20261012", "sizing": {"adapter_sha256": "cef0475d20258b14a8820ce734cfcd422f397131b18ecf3fba5b40c5ec9d7b87", "request_budget": 6500, "state_budget": 4500}, "threshold": 0.5, "track": "public", "vulnerable_or_buggy_before_above_fixed_after": {"correct": 908, "eligible": 1782, "ties": 2}}
-{"alert_rule": "any issue score > max(0.5, question min)", "answered_methods": 992, "average_precision": 0.5253123825884426, "average_precision_answered_methods": 0.5253123825884426, "benchmark_methods": 996, "benchmark_pairs": 498, "benchmark_sha256": "3df4ade65fedd1128d30bc17c9c9869c8ab9bd526e666156d0e16685e78f8a73", "complete": true, "confusion": {"fn": 342, "fp": 64, "tn": 508, "tp": 82}, "context_limit_basis": "SubQ server context-length validation: maximum model length 8192", "context_limit_tokens": 8192, "domain": "bug", "f1": 0.28771929824561404, "failed_methods": 4, "failure_counts": {"Context sizing failed": 4}, "failure_counts_per_question": {}, "failure_policy": "failed positive counts as FN; failed negative counts as FP; failed pair counts as incorrect ordering", "false_positive_rate": 0.11188811188811189, "inference_cost_basis": "API-reported input tokens at SubQ published $0.05 per million input tokens", "inference_cost_usd": 0.13267365, "inference_endpoint": "https://api.subq.ai/v1/systemone", "inference_host": "SubQ", "input_tokens": 2653473, "latency_basis": "Per-method elapsed time, including client rate-limit waits and 429 backoff", "latency_median_seconds": 1.776716374908574, "latency_p95_seconds": 2.5099464589729905, "model_id": "subq-snap", "model_revision": "subq-snap-preview", "pair_order": 0.49875311720698257, "pair_order_correct": 200, "pair_order_eligible": 401, "perch_revision": "4af67af46d76ac99398d61c7ad0a45441760d4dc", "perch_sizing_bundle_sha256": "17592731f69d88e2d1dda3f93b52eab2e3d2fc440d6b3bf0af6583e718cbd51a", "precision": 0.5616438356164384, "protocol_id": "subq-snap-perch-sized-systemone-v1", "question_pack_sha256": "afe116cc8461552aaca996f8f0b639cdf5e42e9d8291c59995482309b5c330dc", "recall": 0.19339622641509435, "refused_methods": 0, "requested_model": "subq-snap-preview", "run_id": "subq-snap-20261012", "sizing": {"adapter_sha256": "cef0475d20258b14a8820ce734cfcd422f397131b18ecf3fba5b40c5ec9d7b87", "request_budget": 6500, "state_budget": 4500}, "threshold": 0.5, "track": "private", "vulnerable_or_buggy_before_above_fixed_after": {"correct": 200, "eligible": 401, "ties": 1}}
-{"alert_rule": "any issue score > max(0.5, question min)", "answered_methods": 2621, "average_precision": 0.5145437196389815, "average_precision_answered_methods": 0.5145437196389815, "benchmark_methods": 2622, "benchmark_pairs": 1311, "benchmark_sha256": "2407059caa3486371749884138b997d89043eec646e548ee8f44ba3ab08514b8", "complete": true, "confusion": {"fn": 860, "fp": 235, "tn": 1245, "tp": 282}, "context_limit_basis": "SubQ server context-length validation: maximum model length 8192", "context_limit_tokens": 8192, "domain": "security", "f1": 0.3399638336347197, "failed_methods": 1, "failure_counts": {"Context sizing failed": 1}, "failure_counts_per_question": {}, "failure_policy": "failed positive counts as FN; failed negative counts as FP; failed pair counts as incorrect ordering", "false_positive_rate": 0.15878378378378377, "inference_cost_basis": "API-reported input tokens at SubQ published $0.05 per million input tokens", "inference_cost_usd": 0.81717625, "inference_endpoint": "https://api.subq.ai/v1/systemone", "inference_host": "SubQ", "input_tokens": 16343525, "latency_basis": "Per-method elapsed time, including client rate-limit waits and 429 backoff", "latency_median_seconds": 2.3714272920042276, "latency_p95_seconds": 4.737156999995932, "model_id": "subq-snap", "model_revision": "subq-snap-preview", "pair_order": 0.5355871886120996, "pair_order_correct": 602, "pair_order_eligible": 1124, "perch_revision": "4af67af46d76ac99398d61c7ad0a45441760d4dc", "perch_sizing_bundle_sha256": "17592731f69d88e2d1dda3f93b52eab2e3d2fc440d6b3bf0af6583e718cbd51a", "precision": 0.5454545454545454, "protocol_id": "subq-snap-perch-sized-systemone-v1", "question_pack_sha256": "afe116cc8461552aaca996f8f0b639cdf5e42e9d8291c59995482309b5c330dc", "recall": 0.2469352014010508, "refused_methods": 0, "requested_model": "subq-snap-preview", "run_id": "subq-snap-20261012", "sizing": {"adapter_sha256": "cef0475d20258b14a8820ce734cfcd422f397131b18ecf3fba5b40c5ec9d7b87", "request_budget": 6500, "state_budget": 4500}, "threshold": 0.5, "track": "public", "vulnerable_or_buggy_before_above_fixed_after": {"correct": 602, "eligible": 1124, "ties": 1}}
-{"alert_rule": "any issue score > max(0.5, question min)", "answered_methods": 878, "average_precision": 0.5004935106105339, "average_precision_answered_methods": 0.5004935106105339, "benchmark_methods": 878, "benchmark_pairs": 439, "benchmark_sha256": "a4b81728e9e520cf1c6f342974bec369d72f5b7d8aabc1c9cf2aceba966f8051", "complete": true, "confusion": {"fn": 357, "fp": 40, "tn": 440, "tp": 41}, "context_limit_basis": "SubQ server context-length validation: maximum model length 8192", "context_limit_tokens": 8192, "domain": "security", "f1": 0.17118997912317327, "failed_methods": 0, "failure_counts": {}, "failure_counts_per_question": {}, "failure_policy": "failed positive counts as FN; failed negative counts as FP; failed pair counts as incorrect ordering", "false_positive_rate": 0.08333333333333333, "inference_cost_basis": "API-reported input tokens at SubQ published $0.05 per million input tokens", "inference_cost_usd": 0.2778463, "inference_endpoint": "https://api.subq.ai/v1/systemone", "inference_host": "SubQ", "input_tokens": 5556926, "latency_basis": "Per-method elapsed time, including client rate-limit waits and 429 backoff", "latency_median_seconds": 2.590956083033234, "latency_p95_seconds": 4.45469124999363, "model_id": "subq-snap", "model_revision": "subq-snap-preview", "pair_order": 0.4648241206030151, "pair_order_correct": 185, "pair_order_eligible": 398, "perch_revision": "4af67af46d76ac99398d61c7ad0a45441760d4dc", "perch_sizing_bundle_sha256": "17592731f69d88e2d1dda3f93b52eab2e3d2fc440d6b3bf0af6583e718cbd51a", "precision": 0.5061728395061729, "protocol_id": "subq-snap-perch-sized-systemone-v1", "question_pack_sha256": "afe116cc8461552aaca996f8f0b639cdf5e42e9d8291c59995482309b5c330dc", "recall": 0.10301507537688442, "refused_methods": 0, "requested_model": "subq-snap-preview", "run_id": "subq-snap-20261012", "sizing": {"adapter_sha256": "cef0475d20258b14a8820ce734cfcd422f397131b18ecf3fba5b40c5ec9d7b87", "request_budget": 6500, "state_budget": 4500}, "threshold": 0.5, "track": "private", "vulnerable_or_buggy_before_above_fixed_after": {"correct": 185, "eligible": 398, "ties": 0}}
+{"alert_rule": "rule violation probability > 0.7", "answered_methods": 428, "average_precision": 0.9559749264228476, "average_precision_answered_methods": 0.9559749264228476, "benchmark_methods": 428, "benchmark_pairs": 214, "benchmark_sha256": "534f6c2abddc507c40f4597607e97e07168c544011d617b83aa39e08abb35d75", "complete": true, "confusion": {"fn": 62, "fp": 6, "tn": 208, "tp": 152}, "context_limit_basis": "SubQ Decisions API documentation: 393216 tokens; prior 8547-token rejection now succeeds", "context_limit_tokens": 393216, "domain": "lint", "f1": 0.8172043010752689, "failed_methods": 0, "failure_counts": {}, "failure_counts_per_question": {}, "failure_policy": "failed positive counts as FN; failed negative counts as FP; failed pair counts as incorrect ordering", "false_positive_rate": 0.028037383177570093, "inference_cost_basis": "API-reported input tokens at SubQ published $0.05 per million input tokens", "inference_cost_usd": 0.052627200000000006, "inference_endpoint": "https://api.subq.ai/v1/systemone", "inference_host": "SubQ", "input_tokens": 1052544, "latency_basis": "Per-method elapsed time, including client rate-limit waits and 429 backoff", "latency_median_seconds": 1.776795000070706, "latency_p95_seconds": 1.9558179170126095, "model_id": "subq-snap", "model_revision": "subq-snap-preview", "pair_order": 0.9626168224299065, "pair_order_correct": 206, "pair_order_eligible": 214, "perch_revision": "4af67af46d76ac99398d61c7ad0a45441760d4dc", "perch_sizing_bundle_sha256": "17592731f69d88e2d1dda3f93b52eab2e3d2fc440d6b3bf0af6583e718cbd51a", "precision": 0.9620253164556962, "protocol_id": "subq-snap-perch-full-context-systemone-v1", "question_pack_sha256": null, "recall": 0.7102803738317757, "refused_methods": 0, "requested_model": "subq-snap-preview", "run_id": "subq-snap-20261013", "sizing": {"adapter_sha256": "cef0475d20258b14a8820ce734cfcd422f397131b18ecf3fba5b40c5ec9d7b87", "max_questions": 16, "request_budget": 380000, "state_budget": 360000}, "threshold": 0.7, "track": "public", "vulnerable_or_buggy_before_above_fixed_after": {"correct": 206, "eligible": 214, "ties": 0}}
+{"alert_rule": "rule violation probability > 0.7", "answered_methods": 142, "average_precision": 0.9568189095822353, "average_precision_answered_methods": 0.9568189095822353, "benchmark_methods": 142, "benchmark_pairs": 71, "benchmark_sha256": "a6e9f2e4ee8ae77bdaf23af46920d97d1ba47375dc9ce03b950cdbe95c2777f9", "complete": true, "confusion": {"fn": 20, "fp": 4, "tn": 67, "tp": 51}, "context_limit_basis": "SubQ Decisions API documentation: 393216 tokens; prior 8547-token rejection now succeeds", "context_limit_tokens": 393216, "domain": "lint", "f1": 0.8095238095238095, "failed_methods": 0, "failure_counts": {}, "failure_counts_per_question": {}, "failure_policy": "failed positive counts as FN; failed negative counts as FP; failed pair counts as incorrect ordering", "false_positive_rate": 0.056338028169014086, "inference_cost_basis": "API-reported input tokens at SubQ published $0.05 per million input tokens", "inference_cost_usd": 0.0169827, "inference_endpoint": "https://api.subq.ai/v1/systemone", "inference_host": "SubQ", "input_tokens": 339654, "latency_basis": "Per-method elapsed time, including client rate-limit waits and 429 backoff", "latency_median_seconds": 1.8107664169510826, "latency_p95_seconds": 1.9817961669759825, "model_id": "subq-snap", "model_revision": "subq-snap-preview", "pair_order": 0.9436619718309859, "pair_order_correct": 67, "pair_order_eligible": 71, "perch_revision": "4af67af46d76ac99398d61c7ad0a45441760d4dc", "perch_sizing_bundle_sha256": "17592731f69d88e2d1dda3f93b52eab2e3d2fc440d6b3bf0af6583e718cbd51a", "precision": 0.9272727272727272, "protocol_id": "subq-snap-perch-full-context-systemone-v1", "question_pack_sha256": null, "recall": 0.7183098591549296, "refused_methods": 0, "requested_model": "subq-snap-preview", "run_id": "subq-snap-20261013", "sizing": {"adapter_sha256": "cef0475d20258b14a8820ce734cfcd422f397131b18ecf3fba5b40c5ec9d7b87", "max_questions": 16, "request_budget": 380000, "state_budget": 360000}, "threshold": 0.7, "track": "private", "vulnerable_or_buggy_before_above_fixed_after": {"correct": 67, "eligible": 71, "ties": 0}}
+{"alert_rule": "any issue score > max(0.5, question min)", "answered_methods": 4440, "average_precision": 0.4857875881931312, "average_precision_answered_methods": 0.4857875881931312, "benchmark_methods": 4440, "benchmark_pairs": 2220, "benchmark_sha256": "32bc68bc85d5535ad28e103adbfa0c09c73e09d0f53a53f7ca35ec219a96e6e2", "complete": true, "confusion": {"fn": 1579, "fp": 244, "tn": 2335, "tp": 282}, "context_limit_basis": "SubQ Decisions API documentation: 393216 tokens; prior 8547-token rejection now succeeds", "context_limit_tokens": 393216, "domain": "bug", "f1": 0.23627984918307499, "failed_methods": 0, "failure_counts": {}, "failure_counts_per_question": {}, "failure_policy": "failed positive counts as FN; failed negative counts as FP; failed pair counts as incorrect ordering", "false_positive_rate": 0.09461031407522295, "inference_cost_basis": "API-reported input tokens at SubQ published $0.05 per million input tokens", "inference_cost_usd": 0.59961915, "inference_endpoint": "https://api.subq.ai/v1/systemone", "inference_host": "SubQ", "input_tokens": 11992383, "latency_basis": "Per-method elapsed time, including client rate-limit waits and 429 backoff", "latency_median_seconds": 1.9240903329337016, "latency_p95_seconds": 2.0959387910552323, "model_id": "subq-snap", "model_revision": "subq-snap-preview", "pair_order": 0.5129068462401796, "pair_order_correct": 914, "pair_order_eligible": 1782, "perch_revision": "4af67af46d76ac99398d61c7ad0a45441760d4dc", "perch_sizing_bundle_sha256": "17592731f69d88e2d1dda3f93b52eab2e3d2fc440d6b3bf0af6583e718cbd51a", "precision": 0.5361216730038023, "protocol_id": "subq-snap-perch-full-context-systemone-v1", "question_pack_sha256": "afe116cc8461552aaca996f8f0b639cdf5e42e9d8291c59995482309b5c330dc", "recall": 0.15153143471252015, "refused_methods": 0, "requested_model": "subq-snap-preview", "run_id": "subq-snap-20261013", "sizing": {"adapter_sha256": "cef0475d20258b14a8820ce734cfcd422f397131b18ecf3fba5b40c5ec9d7b87", "max_questions": 16, "request_budget": 380000, "state_budget": 360000}, "threshold": 0.5, "track": "public", "vulnerable_or_buggy_before_above_fixed_after": {"correct": 914, "eligible": 1782, "ties": 0}}
+{"alert_rule": "any issue score > max(0.5, question min)", "answered_methods": 996, "average_precision": 0.5227841128395672, "average_precision_answered_methods": 0.5227841128395672, "benchmark_methods": 996, "benchmark_pairs": 498, "benchmark_sha256": "3df4ade65fedd1128d30bc17c9c9869c8ab9bd526e666156d0e16685e78f8a73", "complete": true, "confusion": {"fn": 342, "fp": 64, "tn": 508, "tp": 82}, "context_limit_basis": "SubQ Decisions API documentation: 393216 tokens; prior 8547-token rejection now succeeds", "context_limit_tokens": 393216, "domain": "bug", "f1": 0.28771929824561404, "failed_methods": 0, "failure_counts": {}, "failure_counts_per_question": {}, "failure_policy": "failed positive counts as FN; failed negative counts as FP; failed pair counts as incorrect ordering", "false_positive_rate": 0.11188811188811189, "inference_cost_basis": "API-reported input tokens at SubQ published $0.05 per million input tokens", "inference_cost_usd": 0.13404015, "inference_endpoint": "https://api.subq.ai/v1/systemone", "inference_host": "SubQ", "input_tokens": 2680803, "latency_basis": "Per-method elapsed time, including client rate-limit waits and 429 backoff", "latency_median_seconds": 1.8881599999731407, "latency_p95_seconds": 2.034594291006215, "model_id": "subq-snap", "model_revision": "subq-snap-preview", "pair_order": 0.516209476309227, "pair_order_correct": 207, "pair_order_eligible": 401, "perch_revision": "4af67af46d76ac99398d61c7ad0a45441760d4dc", "perch_sizing_bundle_sha256": "17592731f69d88e2d1dda3f93b52eab2e3d2fc440d6b3bf0af6583e718cbd51a", "precision": 0.5616438356164384, "protocol_id": "subq-snap-perch-full-context-systemone-v1", "question_pack_sha256": "afe116cc8461552aaca996f8f0b639cdf5e42e9d8291c59995482309b5c330dc", "recall": 0.19339622641509435, "refused_methods": 0, "requested_model": "subq-snap-preview", "run_id": "subq-snap-20261013", "sizing": {"adapter_sha256": "cef0475d20258b14a8820ce734cfcd422f397131b18ecf3fba5b40c5ec9d7b87", "max_questions": 16, "request_budget": 380000, "state_budget": 360000}, "threshold": 0.5, "track": "private", "vulnerable_or_buggy_before_above_fixed_after": {"correct": 207, "eligible": 401, "ties": 0}}
+{"alert_rule": "any issue score > max(0.5, question min)", "answered_methods": 2622, "average_precision": 0.5132836117796884, "average_precision_answered_methods": 0.5132836117796884, "benchmark_methods": 2622, "benchmark_pairs": 1311, "benchmark_sha256": "2407059caa3486371749884138b997d89043eec646e548ee8f44ba3ab08514b8", "complete": true, "confusion": {"fn": 867, "fp": 229, "tn": 1251, "tp": 275}, "context_limit_basis": "SubQ Decisions API documentation: 393216 tokens; prior 8547-token rejection now succeeds", "context_limit_tokens": 393216, "domain": "security", "f1": 0.33414337788578374, "failed_methods": 0, "failure_counts": {}, "failure_counts_per_question": {}, "failure_policy": "failed positive counts as FN; failed negative counts as FP; failed pair counts as incorrect ordering", "false_positive_rate": 0.15472972972972973, "inference_cost_basis": "API-reported input tokens at SubQ published $0.05 per million input tokens", "inference_cost_usd": 0.9561698000000001, "inference_endpoint": "https://api.subq.ai/v1/systemone", "inference_host": "SubQ", "input_tokens": 19123396, "latency_basis": "Per-method elapsed time, including client rate-limit waits and 429 backoff", "latency_median_seconds": 4.092253290931694, "latency_p95_seconds": 4.715962875052355, "model_id": "subq-snap", "model_revision": "subq-snap-preview", "pair_order": 0.5400355871886121, "pair_order_correct": 607, "pair_order_eligible": 1124, "perch_revision": "4af67af46d76ac99398d61c7ad0a45441760d4dc", "perch_sizing_bundle_sha256": "17592731f69d88e2d1dda3f93b52eab2e3d2fc440d6b3bf0af6583e718cbd51a", "precision": 0.5456349206349206, "protocol_id": "subq-snap-perch-full-context-systemone-v1", "question_pack_sha256": "afe116cc8461552aaca996f8f0b639cdf5e42e9d8291c59995482309b5c330dc", "recall": 0.24080560420315236, "refused_methods": 0, "requested_model": "subq-snap-preview", "run_id": "subq-snap-20261013", "sizing": {"adapter_sha256": "cef0475d20258b14a8820ce734cfcd422f397131b18ecf3fba5b40c5ec9d7b87", "max_questions": 16, "request_budget": 380000, "state_budget": 360000}, "threshold": 0.5, "track": "public", "vulnerable_or_buggy_before_above_fixed_after": {"correct": 607, "eligible": 1124, "ties": 0}}
+{"alert_rule": "any issue score > max(0.5, question min)", "answered_methods": 878, "average_precision": 0.502038239201598, "average_precision_answered_methods": 0.502038239201598, "benchmark_methods": 878, "benchmark_pairs": 439, "benchmark_sha256": "a4b81728e9e520cf1c6f342974bec369d72f5b7d8aabc1c9cf2aceba966f8051", "complete": true, "confusion": {"fn": 354, "fp": 40, "tn": 440, "tp": 44}, "context_limit_basis": "SubQ Decisions API documentation: 393216 tokens; prior 8547-token rejection now succeeds", "context_limit_tokens": 393216, "domain": "security", "f1": 0.1825726141078838, "failed_methods": 0, "failure_counts": {}, "failure_counts_per_question": {}, "failure_policy": "failed positive counts as FN; failed negative counts as FP; failed pair counts as incorrect ordering", "false_positive_rate": 0.08333333333333333, "inference_cost_basis": "API-reported input tokens at SubQ published $0.05 per million input tokens", "inference_cost_usd": 0.344366, "inference_endpoint": "https://api.subq.ai/v1/systemone", "inference_host": "SubQ", "input_tokens": 6887320, "latency_basis": "Per-method elapsed time, including client rate-limit waits and 429 backoff", "latency_median_seconds": 4.16351912508253, "latency_p95_seconds": 4.830607582931407, "model_id": "subq-snap", "model_revision": "subq-snap-preview", "pair_order": 0.4824120603015075, "pair_order_correct": 192, "pair_order_eligible": 398, "perch_revision": "4af67af46d76ac99398d61c7ad0a45441760d4dc", "perch_sizing_bundle_sha256": "17592731f69d88e2d1dda3f93b52eab2e3d2fc440d6b3bf0af6583e718cbd51a", "precision": 0.5238095238095238, "protocol_id": "subq-snap-perch-full-context-systemone-v1", "question_pack_sha256": "afe116cc8461552aaca996f8f0b639cdf5e42e9d8291c59995482309b5c330dc", "recall": 0.11055276381909548, "refused_methods": 0, "requested_model": "subq-snap-preview", "run_id": "subq-snap-20261013", "sizing": {"adapter_sha256": "cef0475d20258b14a8820ce734cfcd422f397131b18ecf3fba5b40c5ec9d7b87", "max_questions": 16, "request_budget": 380000, "state_budget": 360000}, "threshold": 0.5, "track": "private", "vulnerable_or_buggy_before_above_fixed_after": {"correct": 192, "eligible": 398, "ties": 0}}

results.jsonl

@@ -122,3 +122,9 @@
 {"alert_rule": "any issue score > max(0.5, question min)", "answered_methods": 992, "average_precision": 0.5253123825884426, "average_precision_answered_methods": 0.5253123825884426, "benchmark_methods": 996, "benchmark_pairs": 498, "benchmark_sha256": "3df4ade65fedd1128d30bc17c9c9869c8ab9bd526e666156d0e16685e78f8a73", "complete": true, "confusion": {"fn": 342, "fp": 64, "tn": 508, "tp": 82}, "context_limit_basis": "SubQ server context-length validation: maximum model length 8192", "context_limit_tokens": 8192, "domain": "bug", "f1": 0.28771929824561404, "failed_methods": 4, "failure_counts": {"Context sizing failed": 4}, "failure_counts_per_question": {}, "failure_policy": "failed positive counts as FN; failed negative counts as FP; failed pair counts as incorrect ordering", "false_positive_rate": 0.11188811188811189, "inference_cost_basis": "API-reported input tokens at SubQ published $0.05 per million input tokens", "inference_cost_usd": 0.13267365, "inference_endpoint": "https://api.subq.ai/v1/systemone", "inference_host": "SubQ", "input_tokens": 2653473, "latency_basis": "Per-method elapsed time, including client rate-limit waits and 429 backoff", "latency_median_seconds": 1.776716374908574, "latency_p95_seconds": 2.5099464589729905, "model_id": "subq-snap", "model_revision": "subq-snap-preview", "pair_order": 0.49875311720698257, "pair_order_correct": 200, "pair_order_eligible": 401, "perch_revision": "4af67af46d76ac99398d61c7ad0a45441760d4dc", "perch_sizing_bundle_sha256": "17592731f69d88e2d1dda3f93b52eab2e3d2fc440d6b3bf0af6583e718cbd51a", "precision": 0.5616438356164384, "protocol_id": "subq-snap-perch-sized-systemone-v1", "question_pack_sha256": "afe116cc8461552aaca996f8f0b639cdf5e42e9d8291c59995482309b5c330dc", "recall": 0.19339622641509435, "refused_methods": 0, "requested_model": "subq-snap-preview", "run_id": "subq-snap-20261012", "sizing": {"adapter_sha256": "cef0475d20258b14a8820ce734cfcd422f397131b18ecf3fba5b40c5ec9d7b87", "request_budget": 6500, "state_budget": 4500}, "threshold": 0.5, "track": "private", "vulnerable_or_buggy_before_above_fixed_after": {"correct": 200, "eligible": 401, "ties": 1}}
 {"alert_rule": "any issue score > max(0.5, question min)", "answered_methods": 2621, "average_precision": 0.5145437196389815, "average_precision_answered_methods": 0.5145437196389815, "benchmark_methods": 2622, "benchmark_pairs": 1311, "benchmark_sha256": "2407059caa3486371749884138b997d89043eec646e548ee8f44ba3ab08514b8", "complete": true, "confusion": {"fn": 860, "fp": 235, "tn": 1245, "tp": 282}, "context_limit_basis": "SubQ server context-length validation: maximum model length 8192", "context_limit_tokens": 8192, "domain": "security", "f1": 0.3399638336347197, "failed_methods": 1, "failure_counts": {"Context sizing failed": 1}, "failure_counts_per_question": {}, "failure_policy": "failed positive counts as FN; failed negative counts as FP; failed pair counts as incorrect ordering", "false_positive_rate": 0.15878378378378377, "inference_cost_basis": "API-reported input tokens at SubQ published $0.05 per million input tokens", "inference_cost_usd": 0.81717625, "inference_endpoint": "https://api.subq.ai/v1/systemone", "inference_host": "SubQ", "input_tokens": 16343525, "latency_basis": "Per-method elapsed time, including client rate-limit waits and 429 backoff", "latency_median_seconds": 2.3714272920042276, "latency_p95_seconds": 4.737156999995932, "model_id": "subq-snap", "model_revision": "subq-snap-preview", "pair_order": 0.5355871886120996, "pair_order_correct": 602, "pair_order_eligible": 1124, "perch_revision": "4af67af46d76ac99398d61c7ad0a45441760d4dc", "perch_sizing_bundle_sha256": "17592731f69d88e2d1dda3f93b52eab2e3d2fc440d6b3bf0af6583e718cbd51a", "precision": 0.5454545454545454, "protocol_id": "subq-snap-perch-sized-systemone-v1", "question_pack_sha256": "afe116cc8461552aaca996f8f0b639cdf5e42e9d8291c59995482309b5c330dc", "recall": 0.2469352014010508, "refused_methods": 0, "requested_model": "subq-snap-preview", "run_id": "subq-snap-20261012", "sizing": {"adapter_sha256": "cef0475d20258b14a8820ce734cfcd422f397131b18ecf3fba5b40c5ec9d7b87", "request_budget": 6500, "state_budget": 4500}, "threshold": 0.5, "track": "public", "vulnerable_or_buggy_before_above_fixed_after": {"correct": 602, "eligible": 1124, "ties": 1}}
 {"alert_rule": "any issue score > max(0.5, question min)", "answered_methods": 878, "average_precision": 0.5004935106105339, "average_precision_answered_methods": 0.5004935106105339, "benchmark_methods": 878, "benchmark_pairs": 439, "benchmark_sha256": "a4b81728e9e520cf1c6f342974bec369d72f5b7d8aabc1c9cf2aceba966f8051", "complete": true, "confusion": {"fn": 357, "fp": 40, "tn": 440, "tp": 41}, "context_limit_basis": "SubQ server context-length validation: maximum model length 8192", "context_limit_tokens": 8192, "domain": "security", "f1": 0.17118997912317327, "failed_methods": 0, "failure_counts": {}, "failure_counts_per_question": {}, "failure_policy": "failed positive counts as FN; failed negative counts as FP; failed pair counts as incorrect ordering", "false_positive_rate": 0.08333333333333333, "inference_cost_basis": "API-reported input tokens at SubQ published $0.05 per million input tokens", "inference_cost_usd": 0.2778463, "inference_endpoint": "https://api.subq.ai/v1/systemone", "inference_host": "SubQ", "input_tokens": 5556926, "latency_basis": "Per-method elapsed time, including client rate-limit waits and 429 backoff", "latency_median_seconds": 2.590956083033234, "latency_p95_seconds": 4.45469124999363, "model_id": "subq-snap", "model_revision": "subq-snap-preview", "pair_order": 0.4648241206030151, "pair_order_correct": 185, "pair_order_eligible": 398, "perch_revision": "4af67af46d76ac99398d61c7ad0a45441760d4dc", "perch_sizing_bundle_sha256": "17592731f69d88e2d1dda3f93b52eab2e3d2fc440d6b3bf0af6583e718cbd51a", "precision": 0.5061728395061729, "protocol_id": "subq-snap-perch-sized-systemone-v1", "question_pack_sha256": "afe116cc8461552aaca996f8f0b639cdf5e42e9d8291c59995482309b5c330dc", "recall": 0.10301507537688442, "refused_methods": 0, "requested_model": "subq-snap-preview", "run_id": "subq-snap-20261012", "sizing": {"adapter_sha256": "cef0475d20258b14a8820ce734cfcd422f397131b18ecf3fba5b40c5ec9d7b87", "request_budget": 6500, "state_budget": 4500}, "threshold": 0.5, "track": "private", "vulnerable_or_buggy_before_above_fixed_after": {"correct": 185, "eligible": 398, "ties": 0}}
+{"alert_rule": "rule violation probability > 0.7", "answered_methods": 428, "average_precision": 0.9559749264228476, "average_precision_answered_methods": 0.9559749264228476, "benchmark_methods": 428, "benchmark_pairs": 214, "benchmark_sha256": "534f6c2abddc507c40f4597607e97e07168c544011d617b83aa39e08abb35d75", "complete": true, "confusion": {"fn": 62, "fp": 6, "tn": 208, "tp": 152}, "context_limit_basis": "SubQ Decisions API documentation: 393216 tokens; prior 8547-token rejection now succeeds", "context_limit_tokens": 393216, "domain": "lint", "f1": 0.8172043010752689, "failed_methods": 0, "failure_counts": {}, "failure_counts_per_question": {}, "failure_policy": "failed positive counts as FN; failed negative counts as FP; failed pair counts as incorrect ordering", "false_positive_rate": 0.028037383177570093, "inference_cost_basis": "API-reported input tokens at SubQ published $0.05 per million input tokens", "inference_cost_usd": 0.052627200000000006, "inference_endpoint": "https://api.subq.ai/v1/systemone", "inference_host": "SubQ", "input_tokens": 1052544, "latency_basis": "Per-method elapsed time, including client rate-limit waits and 429 backoff", "latency_median_seconds": 1.776795000070706, "latency_p95_seconds": 1.9558179170126095, "model_id": "subq-snap", "model_revision": "subq-snap-preview", "pair_order": 0.9626168224299065, "pair_order_correct": 206, "pair_order_eligible": 214, "perch_revision": "4af67af46d76ac99398d61c7ad0a45441760d4dc", "perch_sizing_bundle_sha256": "17592731f69d88e2d1dda3f93b52eab2e3d2fc440d6b3bf0af6583e718cbd51a", "precision": 0.9620253164556962, "protocol_id": "subq-snap-perch-full-context-systemone-v1", "question_pack_sha256": null, "recall": 0.7102803738317757, "refused_methods": 0, "requested_model": "subq-snap-preview", "run_id": "subq-snap-20261013", "sizing": {"adapter_sha256": "cef0475d20258b14a8820ce734cfcd422f397131b18ecf3fba5b40c5ec9d7b87", "max_questions": 16, "request_budget": 380000, "state_budget": 360000}, "threshold": 0.7, "track": "public", "vulnerable_or_buggy_before_above_fixed_after": {"correct": 206, "eligible": 214, "ties": 0}}
+{"alert_rule": "rule violation probability > 0.7", "answered_methods": 142, "average_precision": 0.9568189095822353, "average_precision_answered_methods": 0.9568189095822353, "benchmark_methods": 142, "benchmark_pairs": 71, "benchmark_sha256": "a6e9f2e4ee8ae77bdaf23af46920d97d1ba47375dc9ce03b950cdbe95c2777f9", "complete": true, "confusion": {"fn": 20, "fp": 4, "tn": 67, "tp": 51}, "context_limit_basis": "SubQ Decisions API documentation: 393216 tokens; prior 8547-token rejection now succeeds", "context_limit_tokens": 393216, "domain": "lint", "f1": 0.8095238095238095, "failed_methods": 0, "failure_counts": {}, "failure_counts_per_question": {}, "failure_policy": "failed positive counts as FN; failed negative counts as FP; failed pair counts as incorrect ordering", "false_positive_rate": 0.056338028169014086, "inference_cost_basis": "API-reported input tokens at SubQ published $0.05 per million input tokens", "inference_cost_usd": 0.0169827, "inference_endpoint": "https://api.subq.ai/v1/systemone", "inference_host": "SubQ", "input_tokens": 339654, "latency_basis": "Per-method elapsed time, including client rate-limit waits and 429 backoff", "latency_median_seconds": 1.8107664169510826, "latency_p95_seconds": 1.9817961669759825, "model_id": "subq-snap", "model_revision": "subq-snap-preview", "pair_order": 0.9436619718309859, "pair_order_correct": 67, "pair_order_eligible": 71, "perch_revision": "4af67af46d76ac99398d61c7ad0a45441760d4dc", "perch_sizing_bundle_sha256": "17592731f69d88e2d1dda3f93b52eab2e3d2fc440d6b3bf0af6583e718cbd51a", "precision": 0.9272727272727272, "protocol_id": "subq-snap-perch-full-context-systemone-v1", "question_pack_sha256": null, "recall": 0.7183098591549296, "refused_methods": 0, "requested_model": "subq-snap-preview", "run_id": "subq-snap-20261013", "sizing": {"adapter_sha256": "cef0475d20258b14a8820ce734cfcd422f397131b18ecf3fba5b40c5ec9d7b87", "max_questions": 16, "request_budget": 380000, "state_budget": 360000}, "threshold": 0.7, "track": "private", "vulnerable_or_buggy_before_above_fixed_after": {"correct": 67, "eligible": 71, "ties": 0}}
+{"alert_rule": "any issue score > max(0.5, question min)", "answered_methods": 4440, "average_precision": 0.4857875881931312, "average_precision_answered_methods": 0.4857875881931312, "benchmark_methods": 4440, "benchmark_pairs": 2220, "benchmark_sha256": "32bc68bc85d5535ad28e103adbfa0c09c73e09d0f53a53f7ca35ec219a96e6e2", "complete": true, "confusion": {"fn": 1579, "fp": 244, "tn": 2335, "tp": 282}, "context_limit_basis": "SubQ Decisions API documentation: 393216 tokens; prior 8547-token rejection now succeeds", "context_limit_tokens": 393216, "domain": "bug", "f1": 0.23627984918307499, "failed_methods": 0, "failure_counts": {}, "failure_counts_per_question": {}, "failure_policy": "failed positive counts as FN; failed negative counts as FP; failed pair counts as incorrect ordering", "false_positive_rate": 0.09461031407522295, "inference_cost_basis": "API-reported input tokens at SubQ published $0.05 per million input tokens", "inference_cost_usd": 0.59961915, "inference_endpoint": "https://api.subq.ai/v1/systemone", "inference_host": "SubQ", "input_tokens": 11992383, "latency_basis": "Per-method elapsed time, including client rate-limit waits and 429 backoff", "latency_median_seconds": 1.9240903329337016, "latency_p95_seconds": 2.0959387910552323, "model_id": "subq-snap", "model_revision": "subq-snap-preview", "pair_order": 0.5129068462401796, "pair_order_correct": 914, "pair_order_eligible": 1782, "perch_revision": "4af67af46d76ac99398d61c7ad0a45441760d4dc", "perch_sizing_bundle_sha256": "17592731f69d88e2d1dda3f93b52eab2e3d2fc440d6b3bf0af6583e718cbd51a", "precision": 0.5361216730038023, "protocol_id": "subq-snap-perch-full-context-systemone-v1", "question_pack_sha256": "afe116cc8461552aaca996f8f0b639cdf5e42e9d8291c59995482309b5c330dc", "recall": 0.15153143471252015, "refused_methods": 0, "requested_model": "subq-snap-preview", "run_id": "subq-snap-20261013", "sizing": {"adapter_sha256": "cef0475d20258b14a8820ce734cfcd422f397131b18ecf3fba5b40c5ec9d7b87", "max_questions": 16, "request_budget": 380000, "state_budget": 360000}, "threshold": 0.5, "track": "public", "vulnerable_or_buggy_before_above_fixed_after": {"correct": 914, "eligible": 1782, "ties": 0}}
+{"alert_rule": "any issue score > max(0.5, question min)", "answered_methods": 996, "average_precision": 0.5227841128395672, "average_precision_answered_methods": 0.5227841128395672, "benchmark_methods": 996, "benchmark_pairs": 498, "benchmark_sha256": "3df4ade65fedd1128d30bc17c9c9869c8ab9bd526e666156d0e16685e78f8a73", "complete": true, "confusion": {"fn": 342, "fp": 64, "tn": 508, "tp": 82}, "context_limit_basis": "SubQ Decisions API documentation: 393216 tokens; prior 8547-token rejection now succeeds", "context_limit_tokens": 393216, "domain": "bug", "f1": 0.28771929824561404, "failed_methods": 0, "failure_counts": {}, "failure_counts_per_question": {}, "failure_policy": "failed positive counts as FN; failed negative counts as FP; failed pair counts as incorrect ordering", "false_positive_rate": 0.11188811188811189, "inference_cost_basis": "API-reported input tokens at SubQ published $0.05 per million input tokens", "inference_cost_usd": 0.13404015, "inference_endpoint": "https://api.subq.ai/v1/systemone", "inference_host": "SubQ", "input_tokens": 2680803, "latency_basis": "Per-method elapsed time, including client rate-limit waits and 429 backoff", "latency_median_seconds": 1.8881599999731407, "latency_p95_seconds": 2.034594291006215, "model_id": "subq-snap", "model_revision": "subq-snap-preview", "pair_order": 0.516209476309227, "pair_order_correct": 207, "pair_order_eligible": 401, "perch_revision": "4af67af46d76ac99398d61c7ad0a45441760d4dc", "perch_sizing_bundle_sha256": "17592731f69d88e2d1dda3f93b52eab2e3d2fc440d6b3bf0af6583e718cbd51a", "precision": 0.5616438356164384, "protocol_id": "subq-snap-perch-full-context-systemone-v1", "question_pack_sha256": "afe116cc8461552aaca996f8f0b639cdf5e42e9d8291c59995482309b5c330dc", "recall": 0.19339622641509435, "refused_methods": 0, "requested_model": "subq-snap-preview", "run_id": "subq-snap-20261013", "sizing": {"adapter_sha256": "cef0475d20258b14a8820ce734cfcd422f397131b18ecf3fba5b40c5ec9d7b87", "max_questions": 16, "request_budget": 380000, "state_budget": 360000}, "threshold": 0.5, "track": "private", "vulnerable_or_buggy_before_above_fixed_after": {"correct": 207, "eligible": 401, "ties": 0}}
+{"alert_rule": "any issue score > max(0.5, question min)", "answered_methods": 2622, "average_precision": 0.5132836117796884, "average_precision_answered_methods": 0.5132836117796884, "benchmark_methods": 2622, "benchmark_pairs": 1311, "benchmark_sha256": "2407059caa3486371749884138b997d89043eec646e548ee8f44ba3ab08514b8", "complete": true, "confusion": {"fn": 867, "fp": 229, "tn": 1251, "tp": 275}, "context_limit_basis": "SubQ Decisions API documentation: 393216 tokens; prior 8547-token rejection now succeeds", "context_limit_tokens": 393216, "domain": "security", "f1": 0.33414337788578374, "failed_methods": 0, "failure_counts": {}, "failure_counts_per_question": {}, "failure_policy": "failed positive counts as FN; failed negative counts as FP; failed pair counts as incorrect ordering", "false_positive_rate": 0.15472972972972973, "inference_cost_basis": "API-reported input tokens at SubQ published $0.05 per million input tokens", "inference_cost_usd": 0.9561698000000001, "inference_endpoint": "https://api.subq.ai/v1/systemone", "inference_host": "SubQ", "input_tokens": 19123396, "latency_basis": "Per-method elapsed time, including client rate-limit waits and 429 backoff", "latency_median_seconds": 4.092253290931694, "latency_p95_seconds": 4.715962875052355, "model_id": "subq-snap", "model_revision": "subq-snap-preview", "pair_order": 0.5400355871886121, "pair_order_correct": 607, "pair_order_eligible": 1124, "perch_revision": "4af67af46d76ac99398d61c7ad0a45441760d4dc", "perch_sizing_bundle_sha256": "17592731f69d88e2d1dda3f93b52eab2e3d2fc440d6b3bf0af6583e718cbd51a", "precision": 0.5456349206349206, "protocol_id": "subq-snap-perch-full-context-systemone-v1", "question_pack_sha256": "afe116cc8461552aaca996f8f0b639cdf5e42e9d8291c59995482309b5c330dc", "recall": 0.24080560420315236, "refused_methods": 0, "requested_model": "subq-snap-preview", "run_id": "subq-snap-20261013", "sizing": {"adapter_sha256": "cef0475d20258b14a8820ce734cfcd422f397131b18ecf3fba5b40c5ec9d7b87", "max_questions": 16, "request_budget": 380000, "state_budget": 360000}, "threshold": 0.5, "track": "public", "vulnerable_or_buggy_before_above_fixed_after": {"correct": 607, "eligible": 1124, "ties": 0}}
+{"alert_rule": "any issue score > max(0.5, question min)", "answered_methods": 878, "average_precision": 0.502038239201598, "average_precision_answered_methods": 0.502038239201598, "benchmark_methods": 878, "benchmark_pairs": 439, "benchmark_sha256": "a4b81728e9e520cf1c6f342974bec369d72f5b7d8aabc1c9cf2aceba966f8051", "complete": true, "confusion": {"fn": 354, "fp": 40, "tn": 440, "tp": 44}, "context_limit_basis": "SubQ Decisions API documentation: 393216 tokens; prior 8547-token rejection now succeeds", "context_limit_tokens": 393216, "domain": "security", "f1": 0.1825726141078838, "failed_methods": 0, "failure_counts": {}, "failure_counts_per_question": {}, "failure_policy": "failed positive counts as FN; failed negative counts as FP; failed pair counts as incorrect ordering", "false_positive_rate": 0.08333333333333333, "inference_cost_basis": "API-reported input tokens at SubQ published $0.05 per million input tokens", "inference_cost_usd": 0.344366, "inference_endpoint": "https://api.subq.ai/v1/systemone", "inference_host": "SubQ", "input_tokens": 6887320, "latency_basis": "Per-method elapsed time, including client rate-limit waits and 429 backoff", "latency_median_seconds": 4.16351912508253, "latency_p95_seconds": 4.830607582931407, "model_id": "subq-snap", "model_revision": "subq-snap-preview", "pair_order": 0.4824120603015075, "pair_order_correct": 192, "pair_order_eligible": 398, "perch_revision": "4af67af46d76ac99398d61c7ad0a45441760d4dc", "perch_sizing_bundle_sha256": "17592731f69d88e2d1dda3f93b52eab2e3d2fc440d6b3bf0af6583e718cbd51a", "precision": 0.5238095238095238, "protocol_id": "subq-snap-perch-full-context-systemone-v1", "question_pack_sha256": "afe116cc8461552aaca996f8f0b639cdf5e42e9d8291c59995482309b5c330dc", "recall": 0.11055276381909548, "refused_methods": 0, "requested_model": "subq-snap-preview", "run_id": "subq-snap-20261013", "sizing": {"adapter_sha256": "cef0475d20258b14a8820ce734cfcd422f397131b18ecf3fba5b40c5ec9d7b87", "max_questions": 16, "request_budget": 380000, "state_budget": 360000}, "threshold": 0.5, "track": "private", "vulnerable_or_buggy_before_above_fixed_after": {"correct": 192, "eligible": 398, "ties": 0}}

subq.py

@@ -1,8 +1,8 @@
 #!/usr/bin/env python3
 """Benchmark SubQ Snap with native decisions and Perch context sizing.
 
-SubQ's server reports an 8,192-token maximum. Perch batches questions and reads
-large methods in overlapping chunks, preserving every target-source byte.
+SubQ documents a 393,216-token window and at most 16 questions per request.
+Perch retains full saved context where it fits and batches the questions.
 Cost uses API-reported input tokens at the published $0.05 per million rate.
 """
 from collections import Counter
@@ -20,8 +20,10 @@ import openrouter_decisions as runner
 from gliner_decide import PerchSizer, SizeError
 
 MODEL = 'subq-snap-preview'
-STATE_BUDGET = 4500
-REQUEST_BUDGET = 6500
+CONTEXT_LIMIT = 393216
+STATE_BUDGET = 360000
+REQUEST_BUDGET = 380000
+MAX_QUESTIONS = 16
 _native_prediction = runner.prediction
 _native_report = runner.report
 _sizer = None
@@ -105,7 +107,7 @@ def predict(key, pair, role, domain, pack, timeout, limiter, model):
         for attempt in range(9):
             probabilities, scores, readings = {}, {}, []
             try:
-                plans = sizer().plan(pair[role], checks, budget, REQUEST_BUDGET, len(checks))
+                plans = sizer().plan(pair[role], checks, budget, REQUEST_BUDGET, MAX_QUESTIONS)
                 def ask(plan, names):
                     nonlocal requests, tokens
                     requests += 1
@@ -133,7 +135,9 @@ def predict(key, pair, role, domain, pack, timeout, limiter, model):
                                    state_sha256=hashlib.sha256(json.dumps(plan['state'], sort_keys=True).encode()).hexdigest())
                     readings.append(reading)
                 for plan in plans:
-                    ask(plan, plan['question_names'])
+                    names = plan['question_names']
+                    for start in range(0, len(names), MAX_QUESTIONS):
+                        ask(plan, names[start:start+MAX_QUESTIONS])
                 break
             except SizeError:
                 discarded += len(readings)
@@ -147,7 +151,7 @@ def predict(key, pair, role, domain, pack, timeout, limiter, model):
         row = prediction(pair, role, domain, pack, body, time.monotonic()-started, model)
         row.update(readings=readings, api_requests=requests, discarded_readings=discarded,
                    final_state_budget=budget, perch_revision=_source_revision,
-                   sizing={'state_budget': STATE_BUDGET, 'request_budget': REQUEST_BUDGET,
+                   sizing={'state_budget': STATE_BUDGET, 'request_budget': REQUEST_BUDGET, 'max_questions': MAX_QUESTIONS,
                            'adapter_sha256': hashlib.sha256((runner.ROOT/'perch-sizing.mjs').read_bytes()).hexdigest()})
         return row
     except Exception as exc:
@@ -171,8 +175,8 @@ def report(group, pack, model):
             category = 'Invalid or incomplete decision'
         failures[category] += 1
     result['failure_counts'] = dict(failures)
-    result.update(context_limit_basis='SubQ server context-length validation: maximum model length 8192',
-                  sizing={'state_budget': STATE_BUDGET, 'request_budget': REQUEST_BUDGET},
+    result.update(context_limit_basis='SubQ Decisions API documentation: 393216 tokens; prior 8547-token rejection now succeeds',
+                  sizing={'state_budget': STATE_BUDGET, 'request_budget': REQUEST_BUDGET, 'max_questions': MAX_QUESTIONS},
                   perch_revision=next((row.get('perch_revision') for row in group['predictions'].values()
                                        if row.get('perch_revision')), _source_revision))
     return result
@@ -184,8 +188,8 @@ def configure():
     runner.KEY_ENV = 'SUBQ_API_KEY'
     runner.COST_BASIS = 'API-reported input tokens at SubQ published $0.05 per million input tokens'
     runner.HOST_PREFIX = ''
-    runner.CATALOG = {MODEL: ('subq-snap', MODEL, 'subq-snap-perch-sized-systemone-v1',
-                             'subq-snap-20261012', 'SubQ', 8192, .05)}
+    runner.CATALOG = {MODEL: ('subq-snap', MODEL, 'subq-snap-perch-full-context-systemone-v1',
+                             'subq-snap-20261013', 'SubQ', CONTEXT_LIMIT, .05)}
     runner.MODELS = (MODEL,)
     runner.prediction = prediction
     runner.predict = predict

test_subq.py

@@ -34,6 +34,20 @@ class SubQTests(unittest.TestCase):
         self.assertAlmostEqual(row['usage']['cost'], .000015)
         self.assertEqual(len(row['readings']), 2)
 
+    def test_security_pack_is_split_into_at_most_sixteen_questions(self):
+        checks = subq.runner.questions_for(self.pair, 'security', self.pack)
+        self.assertGreater(len(checks), subq.MAX_QUESTIONS)
+        sizer = Mock(); sizer.plan.return_value = [self.plan([c['name'] for c in checks])]
+        def answer(key, state, batch, timeout, limiter):
+            self.assertLessEqual(len(batch), 16)
+            return {'model': subq.MODEL, 'answers': {c['name']: {'type':'noul','noul':.9} for c in batch},
+                    'usage': {'input_tokens':100,'output_tokens':0}}
+        with patch.object(subq, 'sizer', return_value=sizer), patch.object(subq, 'request', side_effect=answer) as request:
+            row = subq.predict('key', self.pair, 'before', 'security', self.pack, 30, Mock(), subq.MODEL)
+        self.assertEqual(set(row['question_scores']), {c['name'] for c in checks})
+        self.assertGreater(request.call_count, 1)
+        self.assertTrue(all(len(r['question_names'])<=16 for r in row['readings']))
+
     def test_timeout_stays_failed_and_retains_prior_reading_cost(self):
         sizer = Mock(); sizer.plan.return_value = [self.plan(['semantic']), self.plan(['semantic'])]
         with patch.object(subq, 'sizer', return_value=sizer), patch.object(subq, 'request', side_effect=[self.answer(.9, 300), httpx.ReadTimeout('timeout')]) as request: