Skip to content
SingularityOBSERVATORY
Menu

OBSERVATORY / SYNCHRONIZED EVIDENCE

Timeline

Five evidence tracks · independent units · historical knowledge states

Date definitions
OBSERVED AT

The date or period the evidence describes. Forecast and plan dates remain future source statements, not observations.

PUBLISHED AT

The source publication date when the source establishes one; null otherwise.

KNOWLEDGE AT

The date Vardark first recorded the admitted source revision used by this view; exact UTC instants preserve multiple reviewed revisions on one day.

ASSESSED AT

The date of a Vardark assessment or editorial review; null for unassessed measurements.

SYNCHRONIZED SMALL MULTIPLES

2022—2030

Dates align; values are comparable only within a row.

COMPUTE

Compute capacity

H100e, estimated ownership

2022-03-31 = 100 within Compute only

Quarterly global ownership estimate. Location, access and planned facilities are separate.

VALID
CHIP SUPPLY

Advanced-node supply

thousand 300 mm wafers/month

2024-12-31 = 100 within Chip Supply only

One compatible front-end path. HBM, packaging, equipment and geography remain separate gaps or signals.

LIMITED
AGENT HORIZON

Agent task horizon

expert-minutes by reliability level

No cross-release index

Model- and suite-specific P50/P80 intervals. Release-aligned points are not a single longitudinal leaderboard.

NOT ESTABLISHED
EVENT ANNOTATIONS

Milestones, methods, benchmarks and facilities

Dated events with linked evidence.

HISTORICAL KNOWLEDGE STATE

What Vardark knew by 06 Sep 2026

Corrections append a replacement; they do not erase the statement that was previously published.

CORRECTED STATE

Agent Horizon shows 4 compatible P50/P80 estimates across 2 public model cohorts under the same TH1.1 contract. They are concurrent model and reliability observations, not a time series; suite coverage and the 16-hour boundary limit generalisability.

Show revision history
TRANSITION 1 · 23 Aug 2026, 00:52:22 UTC
Agent Horizon shows two compatible reliability levels, P50 and P80, for the same public frontier cohort. They are concurrent choices of success requirement and do not constitute a time series; suite coverage and the 16-hour boundary limit generalisability.

Agent Horizon shows 4 compatible P50/P80 estimates across 2 public model cohorts under the same TH1.1 contract. They are concurrent model and reliability observations, not a time series; suite coverage and the 16-hour boundary limit generalisability.

agent-horizon/claim/evidence-state ↗
CORRECTED STATE

RExBench v3 reports that the best result, even with human-written hints, is below 44% on realistic research extensions.

Show revision history
TRANSITION 1 · 13 Aug 2026, 09:42:11 UTC
RExBench reports that the best result, even with human-written hints, is below 40% on realistic research extensions.

RExBench v3 reports that the best result, even with human-written hints, is below 44% on realistic research extensions.

research-automation/claim/rexbench-capability ↗