Latest results

17 configurations · 72 tasks each · Publication pending

Scoring & evidence

Loading interactive configuration matrix…

Ranking

Top configurations

Official calibrated estimates

Calibrated ability ↓

1
Sol mediumConditional 95% interval 65.0–86.6
77.6
2
Sol lowConditional 95% interval 63.5–85.7
76.4
3
Luna maxConditional 95% interval 63.4–85.7
76.3
4
Sol xhighConditional 95% interval 63.0–85.4
75.9
5
Luna xhighConditional 95% interval 62.8–85.3
75.7

Published evidence unavailable

Top configuration domain profile: The live public read could not be read. No synthetic measurements were substituted.

Exact run, configuration, scoring-version, and provenance identity could not be verified.

Evidence notesScoring, sensitivity, provenance, time, and cost

What this page claims

Official AIQ is a versioned 0–100 calibrated ability estimate. Its conditional 95% interval holds the estimated item bank fixed. Quality score, task-mix sensitivity, strict pass, time, and cost remain separate diagnostics. Synthetic previews expose quality only and are never ranked as Official.

Scoring version: Unavailable
Strict pass: Unavailable
Quality / task mix: Unavailable
Summed adapter time: Unavailable
API-equivalent cost: Unavailable
Evidence: Unavailable

Published evidence

Values come from RLS-protected public views. Inspect coverage, trust, scoring version, and run provenance before comparing entries.

Published evidence

Values come from RLS-protected public views. Inspect coverage, trust, scoring version, and run provenance before comparing entries.

Published evidence

Official efficiency: Values come from RLS-protected public reads.

Published evidence unavailable

Top estimate run context: The live public read could not be read. No synthetic measurements were substituted.

Exact run, configuration, scoring-version, and provenance identity could not be verified.

Time, token, and cost table

Official efficiency is unavailable.

17 of 17 expected Official efficiency rows are Unavailable. 17 returned rows were rejected. Exact run, configuration, scoring-version, and provenance identity are required.

History

Latest AIQ snapshot

Compare every published configuration now; trend lines begin after the next Official cycle.

Published evidence

History: Values come from RLS-protected public reads.

Measure

Period

Family

Reasoning

ViewSnapshot

First published observation. Compare the configurations in this snapshot; trend lines begin after the next published cycle.

17 published series · scoring 1.0.8 · first observation; trend begins with the next published cycle

Latest visible date: 2026-08-28 UTC · 17 observations · calibrated ability range 69.4–77.6. Highest point estimate: sol-medium · conditional 95% interval 65.0–86.6 · strict pass 27.8% (n=72) · n=72 · coverage Unavailable · scoring 1.0.8 · published

Evidence notes and visible values

Showing 17 configurations in canonical matrix order. Family and reasoning are explicit filters, not point-estimate cutoffs. Published buckets retain the latest exact Official run; synthetic fixture buckets expose no run detail. Lines connect observations for continuity only; they do not interpolate or estimate values between dates. Absent buckets remain gaps, bars use a zero baseline, and the server returns at most 20 buckets per configuration. Each point requires matching run, configuration, scoring version, and provenance identity. Scoring versions: 1.0.8.

Published evidence unavailable

Exact trend run context: The live public read could not be read. No synthetic measurements were substituted.

Exact run, configuration, scoring-version, and provenance identity could not be verified.

Recorded	Series	Primary metric	Primary interval	Strict pass	n	Run / bucket	Coverage	Runtime	Missing	Summed adapter duration	API-equivalent cost	Scoring	Evidence
2026-08-28	sol-low	76.4 · Calibrated ability	63.5–85.7 · Conditional 95% interval	26.4% (n=72)	72	run_602248d32fb259e607d90a9fc14707df48e308f4ede3921cc6324341cde06900 Latest of 1 run	Unavailable	Unavailable	Unavailable	Unavailable	Unavailable	1.0.8	Published
2026-08-28	sol-medium	77.6 · Calibrated ability	65.0–86.6 · Conditional 95% interval	27.8% (n=72)	72	run_5a2a4eca1cb69ceb798ada0d14415156ac4ab27f458ab646760fecac602b31b1 Latest of 1 run	Unavailable	Unavailable	Unavailable	Unavailable	Unavailable	1.0.8	Published
2026-08-28	sol-high	73.2 · Calibrated ability	59.9–83.4 · Conditional 95% interval	25.0% (n=72)	72	run_437b45de053bfca2ff1f69ad7dc57c006818f1ab4203786cc7db5cb6cde076ac Latest of 1 run	Unavailable	Unavailable	Unavailable	Unavailable	Unavailable	1.0.8	Published
2026-08-28	sol-xhigh	75.9 · Calibrated ability	63.0–85.4 · Conditional 95% interval	27.8% (n=72)	72	run_b2372624b77abd0598e1c4f3b783a3726ec32241eacc3396d803f4ae0641b430 Latest of 1 run	Unavailable	Unavailable	Unavailable	Unavailable	Unavailable	1.0.8	Published
2026-08-28	sol-max	75.2 · Calibrated ability	62.2–84.9 · Conditional 95% interval	26.4% (n=72)	72	run_366299d303a738a36a1289a57a3234cc7059d0b057e620f6139ac02787899be3 Latest of 1 run	Unavailable	Unavailable	Unavailable	Unavailable	Unavailable	1.0.8	Published
2026-08-28	sol-ultra	75.2 · Calibrated ability	62.2–84.9 · Conditional 95% interval	27.8% (n=72)	72	run_395ab5e170b1b9653667e306f6c7ffd93d118175c19adacb0a7c5d5b8bfca7b1 Latest of 1 run	Unavailable	Unavailable	Unavailable	Unavailable	Unavailable	1.0.8	Published
2026-08-28	terra-low	70.0 · Calibrated ability	56.2–80.9 · Conditional 95% interval	29.2% (n=72)	72	run_0333c6971437abbd8e524383310d57e9441e80de282d6cd0ba85a4183602d1be Latest of 1 run	Unavailable	Unavailable	Unavailable	Unavailable	Unavailable	1.0.8	Published
2026-08-28	terra-medium	74.5 · Calibrated ability	61.3–84.3 · Conditional 95% interval	27.8% (n=72)	72	run_8832ca59be523e36b05b5516ff7281b3f711eaf10cfe3581e45b8b95d1c4992b Latest of 1 run	Unavailable	Unavailable	Unavailable	Unavailable	Unavailable	1.0.8	Published
2026-08-28	terra-high	74.6 · Calibrated ability	61.4–84.4 · Conditional 95% interval	23.6% (n=72)	72	run_bde48d35c4ab92dc81c7d7b05f60687b45850990211853d6674a97b37d9c56fe Latest of 1 run	Unavailable	Unavailable	Unavailable	Unavailable	Unavailable	1.0.8	Published
2026-08-28	terra-xhigh	71.6 · Calibrated ability	58.0–82.1 · Conditional 95% interval	26.4% (n=72)	72	run_81e97c23f7c4ad230af28152a1c110db8164e0a8769f2dc78ceadb049cf0d294 Latest of 1 run	Unavailable	Unavailable	Unavailable	Unavailable	Unavailable	1.0.8	Published
2026-08-28	terra-max	71.6 · Calibrated ability	58.0–82.1 · Conditional 95% interval	26.4% (n=72)	72	run_29b8598df64fabb999b6f034956e2a10f13758a151d32272cb8cf8746b2d0c52 Latest of 1 run	Unavailable	Unavailable	Unavailable	Unavailable	Unavailable	1.0.8	Published
2026-08-28	terra-ultra	72.8 · Calibrated ability	59.4–83.0 · Conditional 95% interval	25.0% (n=72)	72	run_8dc4463121678365f69e79c11278f844c6b2acdd49f76da099a46e852be079ef Latest of 1 run	Unavailable	Unavailable	Unavailable	Unavailable	Unavailable	1.0.8	Published
2026-08-28	luna-low	69.4 · Calibrated ability	55.6–80.4 · Conditional 95% interval	25.0% (n=72)	72	run_bdc18132086ad72aa7b1f858785bf0673abb4507c9f6633eaad13d06e1b84ad0 Latest of 1 run	Unavailable	Unavailable	Unavailable	Unavailable	Unavailable	1.0.8	Published
2026-08-28	luna-medium	70.6 · Calibrated ability	56.9–81.4 · Conditional 95% interval	23.6% (n=72)	72	run_92a9f7939a2dc7b961a3cbd6b13f719907ae2adfbaca013b6f086514266747cc Latest of 1 run	Unavailable	Unavailable	Unavailable	Unavailable	Unavailable	1.0.8	Published
2026-08-28	luna-high	72.7 · Calibrated ability	59.2–82.9 · Conditional 95% interval	27.8% (n=72)	72	run_9c6cefcef6a726dc2ab54d6b851f06629e45097f8de1b6d47f709ff1cccee3b4 Latest of 1 run	Unavailable	Unavailable	Unavailable	Unavailable	Unavailable	1.0.8	Published
2026-08-28	luna-xhigh	75.7 · Calibrated ability	62.8–85.3 · Conditional 95% interval	29.2% (n=72)	72	run_652a8005e9e3066cf1987e27a140a421e256f7961aca372489ff0686f885ed83 Latest of 1 run	Unavailable	Unavailable	Unavailable	Unavailable	Unavailable	1.0.8	Published
2026-08-28	luna-max	76.3 · Calibrated ability	63.4–85.7 · Conditional 95% interval	27.8% (n=72)	72	run_cc3cfefe0e6c96deb38f020dcb5154bda98b62333adc00bd816f9c9f6c7232ab Latest of 1 run	Unavailable	Unavailable	Unavailable	Unavailable	Unavailable	1.0.8	Published

Transport measurement · not scoring

Normal vs Fast

Same fixed response task, paired modes, 5 trials per mode. AIQ is not recomputed from speed, time, tokens, or credits. TTFT is not shown because the current CLI does not expose first-token timestamps.

2026-08-28 UTC

Median Fast speedup: 1.35×
Fast wins: 17/17
Unavailable modes: 0
Paired configurations: 17/17

Measure

View

Family

Reasoning

Exact paired measurements

Configuration	Normal completion	Fast completion	Normal time	Fast time	Speedup	Normal output	Fast output	Normal credits	Fast credits
sol-low	100.0%	100.0%	21.7 s	16.1 s	1.35×	37.4 tok/s	51.3 tok/s	8.103 credits	20.261 credits
sol-medium	100.0%	100.0%	22.0 s	16.0 s	1.38×	36.5 tok/s	51.4 tok/s	8.150 credits	18.677 credits
sol-high	100.0%	100.0%	21.3 s	16.1 s	1.33×	38.9 tok/s	51.3 tok/s	8.165 credits	18.735 credits
sol-xhigh	100.0%	100.0%	23.3 s	16.6 s	1.40×	36.5 tok/s	50.3 tok/s	7.525 credits	20.468 credits
sol-max	100.0%	100.0%	22.3 s	16.4 s	1.36×	37.9 tok/s	50.9 tok/s	8.229 credits	18.945 credits
sol-ultra	100.0%	100.0%	22.4 s	16.2 s	1.38×	38.4 tok/s	52.4 tok/s	8.297 credits	20.657 credits
terra-low	80.0%	100.0%	24.5 s	18.5 s	1.32×	33.7 tok/s	43.3 tok/s	3.130 credits	6.786 credits
terra-medium	100.0%	100.0%	29.9 s	19.2 s	1.56×	26.8 tok/s	42.3 tok/s	3.240 credits	6.772 credits
terra-high	100.0%	80.0%	21.1 s	16.8 s	1.26×	39.0 tok/s	49.1 tok/s	2.669 credits	5.901 credits
terra-xhigh	100.0%	100.0%	21.4 s	16.2 s	1.33×	38.7 tok/s	51.5 tok/s	2.732 credits	6.181 credits
terra-max	100.0%	100.0%	21.6 s	15.9 s	1.36×	38.8 tok/s	53.0 tok/s	2.491 credits	5.577 credits
terra-ultra	100.0%	100.0%	21.2 s	16.7 s	1.27×	39.4 tok/s	50.5 tok/s	2.575 credits	5.791 credits
luna-low	100.0%	100.0%	21.0 s	15.4 s	1.36×	39.4 tok/s	53.0 tok/s	0.291 credits	0.595 credits
luna-medium	100.0%	100.0%	21.1 s	15.5 s	1.36×	38.9 tok/s	52.7 tok/s	0.269 credits	0.661 credits
luna-high	100.0%	100.0%	20.8 s	16.1 s	1.29×	39.7 tok/s	51.1 tok/s	0.270 credits	0.662 credits
luna-xhigh	100.0%	100.0%	20.8 s	16.2 s	1.28×	40.0 tok/s	50.9 tok/s	0.268 credits	0.598 credits
luna-max	100.0%	100.0%	21.6 s	16.5 s	1.30×	40.1 tok/s	54.4 tok/s	0.297 credits	0.754 credits

Aggregate output rate uses total elapsed time because the current Codex event stream does not expose a trustworthy first-token timestamp. Fast credit estimates use the published 2.5× ChatGPT credit multiplier. Neither measure changes AIQ, confidence intervals, or ranking.

Evidence availabilitypublished + unavailable · 4 sources

Published evidence

Matrix entries: Values come from RLS-protected public reads.

Published evidence

Trend points: Values come from RLS-protected public reads.

Published evidence unavailable

Historical run context: The live public read could not be read. No synthetic measurements were substituted.

Cannot read public_runs: invalid response shape

Published evidence

Historical efficiency: Values come from RLS-protected public reads.

Loading history…

Evidence archive

Run archive

Open any configuration to inspect its 72 task results, failures, timing, cost, and provenance.

Published evidence unavailable

Run archive: The live public read could not be read. No synthetic measurements were substituted.

Cannot read public_runs: invalid response shape

Loading run evidence…

Method

How AIQ is scored

See what AIQ measures, how uncertainty is calculated, and which operational metrics stay outside the score.

Published evidence unavailable

Benchmark method: The live public read could not be read. No synthetic measurements were substituted.

Cannot read public_scoring_versions: stale active tuple

06 · Calibration efficiency estimates

Observed Codex adapter elapsed time and estimated token cost stay distinct

Time is observed Codex adapter elapsed time. Cost is a versioned estimate from covered token aggregate usage and the official OpenAI API pricing documentation, accessed 2026-08-02. It is not actual Codex subscription billing or necessarily an exact API invoice.

Source / as of / estimator version: Official OpenAI API pricing · USD · standard processing tier · 2026-08-02 · aiq.standard-api-equivalent-usd.v1
Formula: Uncached input = total input − cache-read input − cache-write input. Multiply uncached input, cache-read input, cache-write input, and output by their versioned standard per-token rates, then sum them. Reasoning tokens are nested within output and are not added twice.
Coverage: Raw token counters are provider-reported. The verifier recomputes aggregates and the exact cost estimate from those counters when token coverage is complete. Prompts above 272,000 input tokens use 2× input and 1.5× output rates. Aggregate counters cannot identify which requests crossed that threshold, so AIQ displays a published-rate range: all-standard rates as the lower bound and the full long-context uplift as the upper bound. The range is not an exact invoice. Missing, invalid, or JCS-overflowed aggregate usage remains unavailable, never zero.
Observed Codex adapter elapsed time: Sum, median, and p95 of Codex adapter elapsed time: model plus allowed tools. It excludes workspace setup, artifact sealing, and evaluator replay. Runtime-issue tasks consume time; missing or non-invoked cells do not. Full-matrix timings are operational resource-profile evidence under the recorded node, execution order, and concurrency shown with each run. Model, tool, network, and local contention vary. This is not pure task latency or an isolated API-frontier latency test. Cell durations can overlap under concurrency. Signed matrix-stage start and finish times provide the full batch wall-clock, which is counted once across the 17 configurations. TTFT and TPS are unavailable and are not inferred.
Interpretation: AIQ, observed adapter elapsed time, and estimated API-equivalent USD remain separate. Scatter and Pareto context do not create a combined ranking.

Loading method…

Distributed radar

Runner network

Trace where published results came from. AIQ has 3 registered production identities. Radar telemetry is not enabled, so live online or offline state is unknown.

Published evidence

Runner network: Values come from RLS-protected public reads.

Exact retained records

Network status

Counts come from retained registry and signed observation records.

Nodes: 3
Reporting telemetry: 0/3
Recently observed: 0/3
Verified observations: 0
Verified trusted results: 0

Trust model

A signature proves origin, not result quality.

Node identity, signature verification, and receiver trust are separate checks. Only receiver-verified evidence enters the trusted layer; synthetic records prove none of these states.

Exact registry, trust, recency, signature, and provenance state for each public node.
Node	Registry	Trust	Telemetry	Provenance
AIQ Official Runnerofficial	active	trusted verified	Registered identityRadar telemetry not enabled · live state unknown	Published
AIQ Independent Verifierverifier	active	independently reproduced	Registered identityRadar telemetry not enabled · live state unknown	Published
AIQ Official Publisherofficial	active	trusted verified	Registered identityRadar telemetry not enabled · live state unknown	Published

Node evidenceCapabilities, observations, assignments, receipts, and hashes

Registry: active

Registry trust: trusted verified

AIQ Official Runner

official

Identity: sha256:2316ef9ab5fca992036fbce464e10499e3726b375df965e2e9c70ca5d01aa83c
Registry last-seen record: No registry telemetry record
Evidence: Published

Latest signed capability record

Capability telemetry is not enabled for this identity.

Latest signed observation record

Observation telemetry is not enabled for this identity.

Assignment lifecycle

total: 0
offered: 0
accepted: 0
running: 0
completed: 0
revoked: 0
expired: 0

Result receipts

total: 0
received: 0
accepted: 0
rejected: 0

Trust-layer aggregation

Receiver-verified trusted: 0
Signed untrusted: 0
Rejected: 0
Missing: 0
Aggregated at: Not aggregated

Registry: active

Registry trust: independently reproduced

AIQ Independent Verifier

verifier

Identity

Loading radar evidence…

06 · Calibration efficiency estimates

Observed Codex adapter elapsed time and estimated token cost stay distinct

Source / as of / estimator version: Official OpenAI API pricing · USD · standard processing tier · 2026-08-02 · aiq.standard-api-equivalent-usd.v1
Formula: Uncached input = total input − cache-read input − cache-write input. Multiply uncached input, cache-read input, cache-write input, and output by their versioned standard per-token rates, then sum them. Reasoning tokens are nested within output and are not added twice.
Coverage: Raw token counters are provider-reported. The verifier recomputes aggregates and the exact cost estimate from those counters when token coverage is complete. Prompts above 272,000 input tokens use 2× input and 1.5× output rates. Aggregate counters cannot identify which requests crossed that threshold, so AIQ displays a published-rate range: all-standard rates as the lower bound and the full long-context uplift as the upper bound. The range is not an exact invoice. Missing, invalid, or JCS-overflowed aggregate usage remains unavailable, never zero.
Observed Codex adapter elapsed time: Sum, median, and p95 of Codex adapter elapsed time: model plus allowed tools. It excludes workspace setup, artifact sealing, and evaluator replay. Runtime-issue tasks consume time; missing or non-invoked cells do not. Full-matrix timings are operational resource-profile evidence under the recorded node, execution order, and concurrency shown with each run. Model, tool, network, and local contention vary. This is not pure task latency or an isolated API-frontier latency test. Cell durations can overlap under concurrency. Signed matrix-stage start and finish times provide the full batch wall-clock, which is counted once across the 17 configurations. TTFT and TPS are unavailable and are not inferred.
Interpretation: AIQ, observed adapter elapsed time, and estimated API-equivalent USD remain separate. Scatter and Pareto context do not create a combined ranking.

What this page claims

Latest AIQ snapshot

Run archive

How AIQ is scored

Observed Codex adapter elapsed time and estimated token cost stay distinct

Runner network

Network status

A signature proves origin, not result quality.

AIQ Official Runner

Latest signed capability record

Latest signed observation record

Assignment lifecycle

Result receipts

Trust-layer aggregation

AIQ Independent Verifier

Run archive

Runner network

Network status

A signature proves origin, not result quality.

AIQ Official Runner

Latest signed capability record

Latest signed observation record

Assignment lifecycle

Result receipts

Trust-layer aggregation

AIQ Independent Verifier

Latest signed capability record

Latest signed observation record

Assignment lifecycle

Result receipts

Trust-layer aggregation

AIQ Official Publisher

Latest signed capability record

Latest signed observation record

Assignment lifecycle

Result receipts

Trust-layer aggregation

How AIQ is scored

Observed Codex adapter elapsed time and estimated token cost stay distinct

Latest AIQ snapshot