ANALYSIS
Methodology note: Historical index values on this Pulse are normalized to the current methodology for comparability. The original as-published record remains unchanged in the canonical archive.
Verdict: three material events entered the verified candidate set after the previous Daily. The strongest is a preprint claiming that GPT-5.6 Sol Ultra autonomously generated the core construction and proof for a negative answer to a century-old Nevanlinna question. PRAXIST adds a separate execution-grounded autonomous-R&D system result on MLE-bench, while OpenAI’s planned wind-down of direct model supply to SpaceX-owned Cursor is a strategic distribution change. The breadth is meaningful, but none of the capability claims has independent reproduction sufficient to move the indices. ASI Readiness remains 59.03/100, Singularity Readiness remains 53.19/100, and the ASI forecast remains 30% at 5 years / 81% at 10 years with a 2032–33 central estimate.
AI for Science — direct ASI component
GPT-5.6 Sol Ultra / Nevanlinna counterexample — Impact 8.6/10, evidence 63/100, primary confirmed. A primary preprint constructs a real meromorphic-function counterexample and describes it as a full negative answer to a question originating in Nevanlinna’s 1925 work. The paper explicitly attributes the core construction and proof to an autonomous GPT-5.6 Sol Ultra run. This is a high-significance AI-for-Science signal, but theorem correctness and the autonomous process remain author-reported: no peer review, formal verification, independent proof check, reproduction or run transcript was found. The capability penalty remains active; AI-for-Science maturity, ASI Readiness and forecast stay unchanged.
Intelligence and autonomous R&D
PRAXIST cumulative R&D lineages — Impact 8.0/10, evidence 67/100, primary confirmed. The PRAXIST paper reports a lineage-centered system that carries validated artifacts across generations of executable experiments. On its 75-task MLE-bench sweep it records 60 medals, including 49 gold, versus 55 medals and 34 gold for a Claude Code + Claude Opus 4.8 baseline, at much lower reported model spend. The result is a system/scaffold comparison rather than a base-model intelligence win. It is provider-run, based on a single finalized sweep and not independently reproduced; the current product page also differs from the paper on the gold count. Intelligence, Readiness and forecast remain unchanged.
Frontier-model distribution
OpenAI / Cursor contract wind-down — Impact 7.4/10, evidence 82/100, independently confirmed. OpenAI says it plans to end direct model supply to Cursor after the coding platform’s acquisition by SpaceX, proposing November 12, 2026 as the cutoff. Cursor access continues during the transition and the final termination date remains unresolved while discussions continue. This is a material ecosystem-access and distribution state change, not evidence of new model capability. No readiness or forecast movement follows.
Readiness and forecast
ASI READINESS: 59.03/100 — Δ0.00 today
SINGULARITY READINESS: 53.19/100 — Δ0.00 today
ASI FORECAST: 5Y 30% | 10Y 81% — Δ0 pp
Central estimate: 2032–33 — unchanged.
Bottom line: frontier AI is showing more execution-grounded work in mathematics and autonomous R&D, but the evidence is still concentrated in first-party or author-run evaluations. Independent proof checking and benchmark reproduction remain the decisive gates before these signals can justify a Readiness or forecast increase.