AI Wiki

Explore AI Wiki

Corrections

When we get something substantively wrong, we fix it and log the fix here; small copyedits live in each article's revision history instead. Found an error? Use the "Report issue" button on the article (no account needed). So far we have received 2 reports and reviewed 1, taking a median of 4.8 days. 1 is waiting, the oldest since September 26, 2026. That is our record rather than a promise. See the verifiability policy for how checking works.

321 articles were corrected in the last 30 days. Every correction note and resolved report is listed below, newest first. A fix applied to several articles with the same note is shown once, listing each article; "See the change" opens that version in the article's revision history.

  1. editorial

    Feynman TSMC A16 node was stated as an NVIDIA roadmap disclosure without a source (it is a press report; now attributed to TrendForce/Chosun Biz); Q1 FY27 earnings described as upcoming or by analyst consensus, replaced with NVIDIA's reported results

  2. editorial

    Edition numbering was off by one from 2020 on (no 2020 edition in the official count; 2025 is the 10th, 2026 the 11th) and the venue's English name is Beiren Etrong, not Beiren Yichuang

  3. editorial

    Rubin Ultra was first put on the roadmap at Computex 2024, not GTC 2025

  4. editorial

    2026 edition described as upcoming; it ran August 19 to 23, 2026 in Beijing E-Town (dates, theme, Beijing government results figures, NeodraDynamics Neo1 debut added)

  5. editorial

    Production shipments began August 2026 (not 'second half of 2026'); TrendForce April 2026 report misdescribed (it reported a dual-die Rubin Ultra plan); Kyber/NVL576 naming updated to NVIDIA's GTC 2026 scheme (Kyber NVL144, NVL576 = eight 72-GPU racks)

  6. editorial

    Availability updated from 'fall 2026' to NVIDIA's October 7, 2026 announcement (laptops October 16, desktops November); added Nemotron 3.5 Bolt release for RTX Spark

  7. editorial

    TorchTitan TP degree was "1 to 16"; its paper says TP is usually capped at 8 within a node; removed unsupported "selective TP"; linked /wiki/torchtitan

  8. editorial

    Fixed broken /wiki/ctc link (now connectionist_temporal_classification); added sourced successor note on Omnilingual ASR (Nov 2025)

  9. editorial

    Added Reflection's Beam 1.0 DFlash2 drafter (October 2026)

  10. editorial

    Added Zephon, the open-source data loader DatologyAI released in October 2026

  11. editorial

    Hy3 preview was no longer the latest open release ("culminating"); added Hy4 Preview pointer and Hy Memory 2.0 / Hy-Memory-30B-A3B (October 2026)

  12. editorial

    Replaced stale "weights remained forthcoming at the August 22 cutoff" wording; noted Muse Spark Open shares its base model with Muse Spark 1.3 (PyTorch blog)

  13. editorial

    Removed stale claim that Reflection had not published an open-weight model; added Beam 1.0 open-weight release (October 16, 2026)

  14. editorial

    Added Anthropic's April 23, 2026 postmortem on Claude Code quality reports affecting Opus 4.6 (API not affected)

  15. editorial

    Added Anthropic's October 2026 quantum-circuit design research using Claude Mythos 5 (arXiv:2610.22367)

  16. editorial

    TSMC 2024 production wording, 2023 vs 2024 speedup figures, citation

  17. editorial

    NVIDIA was not reported as an investor in the 2026 round; quote attributions and partner availability fixed

  18. editorial

    ICLR 2025 acceptance date (decisions 22 January 2025, not April)

  19. editorial

    Audit + October 2026 update (GTC 2026 NVL72/NVL144/NVL576 recast, reported dual-die and HBM despec, Kyber delay report, NVIDIA responses); removed unverified 2.5M-parts and 120-140 kW claims

  20. editorial

    Add Beam open-weight release (Oct 2026), 2026 funding/compute/government deals, NVIDIA talks; fix Series A leads, HQ, founder bios, Asimov models; remove unsourced claims

  21. editorial

    Fact-check against primary sources; fixed PyTorch reference title, removed unverified venue claims; added day-0 vLLM/SGLang/TensorRT-LLM support, DSpark and DFlash V2 details, architecture breakdown, quantized formats, expanded benchmarks and safety results

  22. editorial

    Claim-by-claim audit against model cards, technical report and Unsloth guide; reworded unsupported Unsloth/LiLiCorr claims; added 24GB packaging, RTX Spark release, Bolt family, architecture, training, recipes, speed comparators

  23. editorial

    Fact-check of rover autonomy, Starlink collision avoidance, debris tracking and other sections against primary sources; replaced references that did not exist; added a section on AI and robots aboard the International Space Station

  24. editorial

    Separate Zhejiang Lab's Three-Body constellation from ADA Space's Star-Compute program, update Cowboy Space (Reason-1 launched, FCC record), Lonestar and Axiom Space

  25. editorial

    NASA began operating on October 1, 1958; July 29, 1958 is when the Space Act was signed