Tootfinder

OpenAI says its internal model produced 372 math results, nearly all from a single prompt handed to a single AI agent; some might have taken multiple attempts (Joseph Howlett/Scientific American)
https://www.scientificamerican.com/article

OpenAI unleashes hundreds more math results upon a field already in shock
A month after resolving one of the six biggest open problems in mathematics, the company says its new internal model has toppled heaps more—this time without the multimillion-dollar price tag

England in the Middle Ages: If a criminal is at large, he is summoned to the county court. If he fails to appear four times, he is declared an outlaw on the fifth occasion. Thereafter he may be beheaded on sight.
(source: Ian Mortimer - The Time Travelers Guide to Medieval England)

OpenAI prices GPT-6.1 Sol at $2/1M input and $10/1M output tokens, the same as GPT-6 Sol and Claude Sonnet 5.5, and says it performs well on safety tests (Maximilian Schreiner/The Decoder)
https://the-decoder.com/gpt-6-1-sol-comes-close-to-astra-at-a-fifth-o…

GPT-6.1 Sol comes close to Astra at a fifth of the price
OpenAI's new GPT-6.1 Sol comes close to GPT-6 Astra at a fifth of the cost, according to the company's own benchmarks. The planned flagship, GPT-6.1 Astra, is staying under wraps for now. Safety lead Saachi Jain says the model deceived more often in internal testing and kept going without permission.

Spec Sheets Are Not Kernels: An ISA- and Source-Level Audit of INT8 Availability on NVIDIA Blackwell Ultra
Teng-Ruei Chen
https://arxiv.org/abs/2608.11693 https://arxiv.org/pdf/2608.11693 https://arxiv.org/html/2608.11693
arXiv:2608.11693v1 Announce Type: new
Abstract: NVIDIA's published specifications give the Blackwell Ultra GPU (B300) a dense-compute ratio of roughly 30:1 between FP8 and INT8 tensor-core throughput; its predecessors, H200 and B200, both provide 1:1. We audit what this deprioritization means in practice by tracing INT8 W8A8 support through four layers of the stack: the published specifications, the PTX ISA, NVIDIA's CUTLASS kernel library, and the two major open-source LLM serving engines (vLLM and SGLang). We find a consistent, layered withdrawal: (i) the PTX ISA never exposes the fifth-generation tensor-core integer path (tcgen05.mma with .kind::i8) on sm_103a, even though the same PTX revision extends the FP4 kinds to that target, leaving legacy warp-level IMMA as the only architecturally legal integer tensor-core path on B300; (ii) CUTLASS's kernel generator explicitly skips INT8 UMMA generation for any build targeting 103a, while generating FP8 unconditionally; (iii) vLLM ships no INT8 GEMM for Blackwell and fails with a hard runtime error at the first forward pass, after the model has loaded; and (iv) SGLang's ahead-of-time INT8 GEMM stops at Sm90, while its FP8 tuning configurations already cover B200. We document an escape hatch (rerouting vLLM's INT8 path to a JIT-compiled Triton backend via an environment variable), a false-negative trap in the obvious profiler methodology for detecting "native INT8" on sm_103, and the practical failure semantics that make naive testing expensive. Together, these findings show that a quantization format's availability is a property of the whole stack rather than of the model or the spec sheet. Four distinct layers, three of them NVIDIA's own, withdrew INT8 support in mutually consistent ways, and a format that is nominally present on the datasheet is, by default, undeployable on this hardware.
toXiv_bot_toot

OpenAI commits $1B in subsidized model access, training, support, and partnerships to a new initiative aimed at protecting essential services around the world (Sam Sabin/Axios)
https://www.axios.com/2026/09/03/openai-critical-infrastructure-cyber-ai-models

OpenAI launches plan to protect critical infrastructure from AI cyberattacks
The move comes as the AI lab releases its most powerful model to date.

GateTruth: Auditing the Rigor of RTL Design Benchmarks via Mutation Testing
Meet Bhadra
https://arxiv.org/abs/2608.12635 https://arxiv.org/pdf/2608.12635 https://arxiv.org/html/2608.12635
arXiv:2608.12635v1 Announce Type: new
Abstract: Benchmarks for evaluating large language models on register-transfer-level (RTL) hardware design have proliferated rapidly, yet none reports having applied mutation testing, an established hardware-verification technique for quantifying testbench quality, to ask whether its own testbenches are trustworthy. A testbench that never fails is not evidence of a correct design; it may simply never stimulate the logic that is actually broken. We introduce GateTruth, a mutation-testing engine and methodology for auditing RTL benchmark testbench rigor: inject a deterministic, seeded set of semantic mutants into a reference design and measure what fraction the testbench catches.
We validate the methodology against our own 68-task, dual-track reference suite -- 60 specification-to-RTL generation tasks and 8 agentic-repair tasks, scored through a pinned, deterministic synthesis-to-timing flow with correctness enforced as a strict gate -- certifying that 46 of 60 Track A testbenches kill at least 95% of injected mutants under sequential, reproducible execution; we disclose why the other 14 do not, including a Goodhart effect on testbenches revised to pass this gate. We then point the same engine, unmodified, at RTLLM v2.0, a widely adopted external benchmark: of 46 auditable designs, 72% fall below the 95% floor our own suite is held to, and three score 0% outright.
A comparable audit of NVIDIA's CVDP benchmark is structurally impossible: its public release withholds reference solutions, removing the golden RTL mutation testing requires. Auditing our own instrument also surfaced a second finding: an initially uniform 4096-token output cap silently truncated three of seven evaluated models, and re-running at 16,384 tokens moved one model from fifth place to first. We argue mutation-kill certification should become a standard reporting requirement for RTL-generation benchmarks generally.
toXiv_bot_toot

Oil Was Pouring Out of Hormuz Again. Then Attacks on Ships Resurged. (Wall Street Journal)
https://www.wsj.com/world/middle-east/oil-was-pouring-through-the-strait-of-hormuz-again-then-attacks-on-shipping-resurged-fe9f0dd9
http://www.memeorandum.com/261005/p14#a261005p14

Molecular gas hidden in plain sight in the early Universe
Pasquier Noterdaeme, Jens-Kristian Krogager, Sergei Balashev
https://arxiv.org/abs/2607.18974 https://arxiv.org/pdf/2607.18974 https://arxiv.org/html/2607.18974
arXiv:2607.18974v1 Announce Type: new
Abstract: We report the discovery of an extreme intervening molecular absorber at zabs = 4.1 towards the z = 4.7 quasar SDSS J080023.02 305101.22 revealed through strong H2 absorption that had remained unnoticed in archival data for about two decades. The system, which we analysed with new VLT/X-shooter observations, has a metallicity of about one-fifth the solar value, in line with the moderate dust reddening AV~0.06 mag. More strikingly, it exhibits the highest molecular fraction, fH2 = 2N(H2)/(2N(H2) N(HI))~=60%, measured at z>0 from direct determinations of both atomic and molecular hydrogen column densities. Notably, this fraction is comparable to the very highest values in the Local Group. The inferred fH2 still represents a conservative lower limit to the local molecular fraction since the H i absorption very likely includes atomic gas unrelated to the actual molecular component, as indicated by the wide (Delta v~500 km/s) multi-component low-ionisation metal profile.
The detection of such a system at early cosmic times is remarkable given the limited number of quasar spectra probing z>4. It suggests an unexpectedly high incidence of H2, with important implications for the evolution of molecular gas. This high incidence may be driven by high average densities and enhanced turbulence at those redshifts. At the same time, our results also highlight possible observational biases, both in quasar selection and in recognising strong H2 absorption, suggesting that a significant fraction of molecular gas may remain undetected, in particular near the peak of star formation. The present system offers a unique benchmark for developing efficient detection algorithms. Further progress will benefit from colour-independent quasar surveys, while constraining the physical conditions and environments of such extreme absorbers will require observations on future extremely large telescopes.
toXiv_bot_toot