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); we aim to review every report within 72 hours. See the verifiability policy for how checking works.

1. MLE-benchSeptember 7, 2026editorial

   the 31.6% figure was filed under AIDE + o3 but is the AIRA-greedy + o3 result on the full benchmark; clarified which AIRA numbers are MLE-bench Lite, added an AIRA2 MLE-bench-30 Percentile Rank subsection and the AIRA3 live-competition note with two new footnotes

2. Inception LabsSeptember 7, 2026editorial

   Inception's 'Next Step for dLLMs' post dates to November 6, 2025, not May 14, 2026

3. Mercury (Inception Labs)September 7, 2026editorial

   Inception's 'Next Step for dLLMs' post dates to November 6, 2025, not May 14, 2026

4. Diffusion Language ModelsSeptember 7, 2026editorial

   Mercury Coder's February 2025 launch did not coincide with the $50M round (announced November 2025). Added Nemotron-Labs-TwoTower, Nemotron-Labs-Diffusion, DFlash/DSpark block-diffusion drafters, LLaDA 2.1/2.0-Uni/2.2, Mercury 2 follow-ups, and a September 2026 adoption status.

5. NVIDIA Vera RubinSeptember 7, 2026editorial

   Groq 3 LPX SRAM is 500 MB per LPU per NVIDIA, not 512 MB per die; added the LPX blog as a source

6. Cerebras WSE-4 and 2026 IPOSeptember 7, 2026editorial

   CS-4 was announced Aug 18, 2026 on the WSE-3 Turbo, not a WSE-4; added WSE-3 Turbo and CS-4 specs, Cerebras roadmap statements, IPO closing total, founding year fix

7. Mustafa SuleymanSeptember 6, 2026editorial

   MAI-Thinking-1 context window is 256,000 tokens, not 128,000 (per Microsoft announcement, keynote, technical report and Foundry catalog); 35B is the active-parameter count of a ~1T MoE; added Aug 12, 2026 public preview and Foundry pricing

8. MAI-Thinking-1September 6, 2026editorial

   MAI-Thinking-1 moved to public preview on Microsoft Foundry on Aug 12, 2026 at $2/$8 per 1M tokens; SWE-Bench Pro tightened to 52.8%; Sonnet 4.6 link fixed; technical-report architecture, benchmark table, human-eval results and EU data summary added

9. NVIDIA Isaac LabSeptember 6, 2026editorial

   PhysX described as the original and default physics backend and Newton as the alternative backend introduced in the Isaac Lab 3.0 beta (17 March 2026), with the multi-backend architecture, kit-less mode, presets, and 3.0 timeline rows added

10. NVIDIA NewtonSeptember 6, 2026editorial

    Newton 1.0 dates separated by event (v1.0.0 tag, changelog and PyPI wheel March 10, 2026; GTC 2026 GA announcement March 16; GitHub release page April 13); v1.5.1 re-dated to its August 27 tag and PyPI upload with the August 28 GitHub page noted; v1.4.0 and v1.5.0 artifact dates confirmed; v1.6.0rc1 (September 3) added

11. Normal ComputingSeptember 6, 2026editorial

    CN101 is a digital CMOS chip (accumulator dynamics, stochastic computing), not an analog RC network; added the August 2026 CN101 paper results, Hot Chips 2026 listing, tape-out date discrepancy, 2026 roadmap shift, and fixed misattributed publications

12. LightmatterSeptember 5, 2026editorial

    fixed the Passage M1000 and L200 launch date (March 31, 2025, not December 2024), the L20 and L200 first-disclosure dates, headquarters (Mountain View), the CFO/Graham roles, the Series C investor list, and two reference bylines/dates; added 2025-2026 developments (16-wavelength link, Guide VLSP light engine, GUC/Synopsys/Cadence partnerships, Passage L20, vClick, 1.6 Tbps per fiber, XPO MSA, OCP CPO initiative and August 2026 workstream, Roy Kim, Guide DR, NVLink Fusion), reference-platform and Guide specs, a timeline, and leadership/board from the company site.

13. Mustafa SuleymanSeptember 5, 2026editorial

    Build 2026 MAI launch dated June 2, 2026; added MAI-Image-2.5, 2.5-Pro, 2.6 and 2.6-Flash entries to the MAI model timeline

14. Normal ComputingSeptember 5, 2026editorial

    Extropic is headquartered in Waltham, Massachusetts (not California) and announced its $14.1 million seed in December 2023 (not March 2024), per PR Newswire and SiliconANGLE

15. WayveSeptember 4, 2026editorial

    fixed the February 2026 financing (1.2B Series D, not a Mercedes-led round), the UK government MoU date and the Tokyo robotaxi attribution; added the September 2026 London launch on Uber, 2026 partnerships, Series D extension and Wayve Labs

16. Clément DelangueSeptember 4, 2026editorial

    added the September 2026 agreement to sell Hugging Face to NVIDIA from Delangue's perspective (approach to Huang, his post, CNBC interview, reporters' call, incident statements, terms and role); fixed the misattributed $50-100 billion remark (Lux Capital's Brandon Reeves in Forbes, not Sequoia), the stale private-company sentence, the Miami residence, and an unsourced headcount

17. Hugging FaceSeptember 4, 2026editorial

    NVIDIA agreed to acquire Hugging Face on September 2, 2026 (announced September 3, pending close in H1 2027); rewrote the August acquisition-report paragraphs as history, added a deal-terms subsection with the openness commitments and risk factor, updated the lead and financing text, added the criticism section's acquisition concerns, and fixed the stale claim that no final incident report had been published (OpenAI and METR published on August 26).

18. GPT-6 AstraSeptember 4, 2026editorial

    internal ExploitBench (June-August 2026) Sol figure is 11.5% (launch-day table cell read 5.5% against the post's own chart); Fast-mode speed and Bedrock wording changes on the live post noted

19. LLM API Pricing ComparisonSeptember 4, 2026editorial

    GPT-6 Astra API pricing added ($10/$1/$12.50/$50, Fast mode 2x, Bedrock); GPT-5.6 Sol promotional cut of August 21 ($4/$0.40/$20 through at least Nov 21) replaces the stale $5/$30; DeepSeek peak/off-peak repricing in force since August 16; Claude Sonnet 5 September 1 increase cancelled; Claude Fable 5.1, Gemini 3.6/3.7/3.8 Flash promotional rate, and Grok 4.6 added; every OpenAI row re-verified September 3, 2026.

20. ChatGPTSeptember 4, 2026editorial

    GPT-5.6 API prices were stale (Sol is $4/$20 promotional since August 21, Terra $2/$12 and Luna $0.20/$1.20 since July 30). Added the September 3, 2026 GPT-6 Astra rollout in ChatGPT: plan availability, Astra Pro, Enterprise opt-in, credits, computer use and Sites, safety pauses and cyber refusals, Codex context notes, API pricing, model table row, and timeline entries.

21. Preparedness Framework (OpenAI)September 4, 2026editorial

    adds the GPT-6 Astra sequence (August-September 2026, the first Critical designation, mapped onto the framework's thresholds, Safeguards Report, and development-phase rules); fixes the severe-harm gloss, a High-capability misquote, and the claim that Version 2 dropped the halt-development instruction; converts footnote citations to numbered markers without renumbering.

22. GPT-5.6September 4, 2026editorial

    GPT-5.6 Sol's API price has been a promotional $4/$20 per million tokens since August 21, 2026 (the article said its launch rates were unchanged); adds the September 3, 2026 GPT-6 Astra launch as successor, with OpenAI's Sol-versus-Astra benchmark and alignment comparison, pricing ratio, and Aidan Clark's training-run statements

23. Claude Agent SDKSeptember 2, 2026editorial

    fixed SDK sessions, permissions, setting defaults, links, and unsupported adoption claims

24. Tendon-driven actuationSeptember 2, 2026editorial

    Qualified prevalence and backdrivability claims and added the verified 2026 1X NEO tendon hand

25. Claude Mythos PreviewSeptember 2, 2026editorial

    moved detailed Claude Mythos 5.1 coverage to the Claude Mythos 5 article while retaining the Preview lifecycle note

26. SpaceX StarmindAugust 18, 2026editorial

    Fixed AI1's compute payload at 150 kW peak and 120 kW average throughout, removed a false current-versus-earlier distinction, and refreshed the official SpaceX source access date.

27. Space-based data centersAugust 18, 2026editorial

    Updated the SpaceX AI1 power and late-2027 schedule, replaced obsolete Aetherflux material with Cowboy Space, added the Cowboy and Orbital FCC filings plus Aethero's flown edge-compute missions, qualified launch-cost model inputs and broad first/inventory claims, and preserved the verified ESA study material.

28. GPU ClusterAugust 17, 2026editorial

    Corrected cluster scale, infrastructure status, cooling, and reliability claims

29. AI DatacenterAugust 17, 2026editorial

    update infrastructure, cooling, power, water, project, and capital expenditure claims through August 17, 2026

30. NVIDIA L40SAugust 17, 2026editorial

    Qualified historical performance and market comparisons and removed unsupported availability and efficiency claims.

31. OpenAIAugust 17, 2026editorial

    updated the Ohio infrastructure project from reported talks to announced lease and guarantee terms

32. CME Compute FuturesAugust 17, 2026editorial

    add pending CFTC approval status, proposed settlement formula, index scope, and filed position controls

33. Space-based data centersAugust 12, 2026editorial

    Starcloud-2 launch date updated from October 2026 to January 2027 per SpaceNews (May 26, 2026); China Three-Body constellation second/third-group claim reattributed to ADA Space EVP Zhao Hongjie via China SCIO and the SatNews February 2026 testing claim reattributed to CASIC's separate CubeSat testbed; added supporting references for the 744 TOPS / 1,000 POPS / 2,400-400 figures, the Gigasat Factory and TeraFab claims, and Lonestar's in-transit operations (now attributed to the company); corrected the 5 GW concept attribution to Starcloud, Google's next-nearest-neighbor spacing, the Altman quote wording, the Deutsche Bank row's unverified figures, and Starcloud's total-funding wording.

34. AI watermarkingAugust 12, 2026editorial

    same copy-paste vs model-level causal fix as claude page

35. Nemotron 3August 12, 2026editorial

    Nemotron Open Model License commercial terms apply to Nano and Super, not OpenMDW-licensed Ultra

36. Claude (language model)August 12, 2026editorial

    copy-paste persistence comes from the watermark being part of the text; model-level application means presence across products

37. Nemotron 3August 12, 2026editorial

    Ultra shipped under OpenMDW-1.1, not the Nemotron Open Model License; add Nemotron 3.5 Lightning (Aug 11, 2026)

38. C2PA (Coalition for Content Provenance and Authenticity)August 12, 2026editorial

    update AI Omnibus from May 2026 provisional agreement to enacted Regulation (EU) 2026/1744; add Anthropic/Claude C2PA adoption (Aug 2026)

39. WeirdMLAugust 7, 2026editorial

    Claude Opus 4.6 (high) score is 77.9% per the leaderboard CSV

40. NVIDIA RTX SparkAugust 7, 2026editorial

    OEM follow-on claim matched to the cited source (no timing, Dell included)

41. NVIDIA IsingAugust 7, 2026editorial

    QCalEval author count, decoder license and gating details, press-release quote wording

42. Muse SparkAugust 7, 2026editorial

    attribute Arena's Pareto-frontier remark to the model's placement, not its pricing

43. MiniMax CodeAugust 7, 2026editorial

    check-in point tiers per the changelog; cite the dedicated Token Plan pages

44. WeirdMLAugust 7, 2026editorial

    refresh stale v2 leaderboard (top now Claude Fable 5 max at 91.9%, Aug 8 2026), fix o3/o1 and Mixtral row mixups and v1-only timeout/run counts, add Muse Spark 1.2 result and Agentic WeirdML section

45. World action modelAugust 7, 2026editorial

    Cosmos 3 DROID policy models first shipped with the May 31, 2026 family release, not July

46. Vals AIAugust 7, 2026editorial

    remove two editorial scaffold sentences that leaked into the published text

47. Situational AwarenessAugust 7, 2026editorial

    Bloom Energy is a fuel-cell power company, not a Bitcoin miner; the $5.5 billion figure is 13F-disclosed U.S. equity exposure, not total assets, and the $225M-to-$5.5B framing was not in the cited source

48. Sakana AIAugust 7, 2026editorial

    Ren Ito is listed as Co-Founder and Chairman on sakana.ai/company-info as of August 2026; infobox Key people and body updated (COO retained as his founding-era role).

49. Safe Superintelligence IncAugust 7, 2026editorial

    recast the TPU-as-main-substrate claim in dated past framing and add the July 27, 2026 NVIDIA strategic partnership and reported $5 billion investment (NVIDIA Newsroom, Bloomberg, TechCrunch); paraphrase the Gavin Baker remark whose verbatim wording differs across renderings.

50. RogoAugust 7, 2026editorial

    source the $2 billion Series D valuation, re-cite the customer list to Rogo's own Series B post, and remove claims unsupported by their citations

51. Prime IntellectAugust 7, 2026editorial

    December 2025 database entry is the Series A's initial close, not a separate ~$49.9M Series B; fix the commoditizing tagline quote

52. Prime AgentAugust 7, 2026editorial

    align quoted README wording ("are not a security sandbox")

53. Open Secure AI AllianceAugust 7, 2026editorial

    CoSAI partner count is more than 40 per the cited OASIS release, not 45

54. NVIDIA Vera RubinAugust 7, 2026editorial

    200 petaFLOPS figure appears only in the July post; drop unsupported cable-free attribution for the Pegatron tray

55. Muse SparkAugust 7, 2026editorial

    DeepSWE 1.1 prose (GPT-5.6 Terra leads Muse Spark 1.2 there), three reference bylines, Wang 'rough edges' quote re-cited to his April 8 X thread, unsupported Bloomberg attribution removed, Avocado code name re-cited to CNBC.

56. Muse CodeAugust 7, 2026editorial

    Muse Spark 1.2 does not beat GPT-5.6 Terra on every coding chart (Terra leads DeepSWE 1.1, 64.8% vs 59.3%, per Meta's own chart); Terminal-Bench 2.1 submissions run through the harbor framework, not a benchmark-owned harness.

57. Meta Superintelligence LabsAugust 7, 2026editorial

    replaced 13 dead or invented reference URLs with verified real sources, re-pointed the anachronistic Wang layoff-memo and LeCun citations to correctly dated sources, and tightened claims to match the verified sources.

58. Isomorphic LabsAugust 7, 2026editorial

    Companies House citation pointed to the wrong record (13207520 is unrelated GG-895-389 LIMITED); the real record is Isomorphic Labs Limited, company number 13223825, and that is the registered name (no previous names on file).

59. Higgsfield AIAugust 7, 2026editorial

    fix unsupported user/video figures, missourced complaint and founder-background citations, seed-round details, and the dead AMD/TensorWave reference

60. Hell GrindAugust 7, 2026editorial

    note that the festival's own materials disagree on the submission opening date (rules say Aug 7, timeline widget says Aug 10)

61. FluidStackAugust 7, 2026editorial

    the December 2025 ~$700M raise was in-talks reporting only; the round closed as an $830M Series A at a $7.5B valuation in January 2026, led by Situational Awareness, announced July 20, 2026

62. Conduit (neurotechnology company)August 7, 2026editorial

    match the cited source's competition name (North American Women's Under-20 Championship)

63. Cloudflare OSAugust 7, 2026editorial

    the term gadgets comes from the project README, not the launch post

64. CloudflareAugust 7, 2026editorial

    cite TechCrunch for Human Native background; AI Gateway routing in Cloudflare OS is optional, not universal

65. BittensorAugust 7, 2026editorial

    dTAO-era subnet growth started from about 64 subnets, not the old 32 cap; update Apex description; fix GTAO first-product attribution

66. BigFinanceBenchAugust 7, 2026editorial

    the GPT-5.6 figures on OpenAI's page sit inside a Rogo testimonial quote, not OpenAI's own results

67. BigBang-v1August 7, 2026editorial

    remove wrong day-after-Dean-departure linkage (weights shipped Aug 2, Dean announced Aug 5); fix quantization timing and loosen same-day counts

68. Situational AwarenessAugust 7, 2026editorial

    47% gain was H1 2025, not H2 2024; add July 2026 fund deleveraging (Citadel sale, $45B to $10B) and August 2026 $400M private investment, per WSJ/CNBC/FT/Bloomberg

69. Leopold AschenbrennerAugust 7, 2026editorial

    47% gain was first-half 2025, not first-half 2026; add July 2026 deleveraging (Citadel sale, $45B to $10B) and August 2026 $400M private investment, per WSJ/CNBC/FT/Bloomberg

70. Isomorphic LabsAugust 7, 2026editorial

    fix leadership errors (Jaderberg joined May 2022 as ML director, not May 2024 as chief AI officer; no AlphaFold role; Yakneen was ex-SOPHiA GENETICS, not Roche/Microsoft; Cambridge US site opened June 2025, not 2024); refresh leadership table to August 2026 roster incl. CMO Ben Wolf; add February 2025 Novartis expansion and July 2026 joint bioresilience program

71. Higgsfield AIAugust 7, 2026editorial

    fix 11 'Highsfield' misspellings; add Higgsfield Originals/Hell Grind section (Cannes screening and dispute, Aug 2026 open-source release, $1M Global Film Festival) and May 2026 $400M run-rate figure

72. SpaceX StarmindAugust 5, 2026editorial

    reference access date aligned with the as-of date of the vendor-agnostic language claim

73. Sanjay GhemawatAugust 5, 2026editorial

    the New Yorker's ten-month gap refers to their DEC departures, not their Google start dates; Senior Fellow citation fixed

74. Koray KavukcuogluAugust 5, 2026editorial

    Dean's title attributed to The Verge rather than the Google announcement; co-cited the August 5 post for the former CTO title after Google silently updated its author page

75. Jeff DeanAugust 5, 2026editorial

    harmonized the early-Google employee count with the New York Times' 30th-employee figure

76. Google DeepMindAugust 5, 2026editorial

    co-cited Wired for Dean's dual chief-scientist title and Dean's announcement for the Discovery Loop name, which the Google post does not contain

77. Discovery LoopAugust 5, 2026editorial

    removed an unsourced superlative, restored Wired's question verbatim, re-attributed the NYT framing in its own terms, and removed an embellished 'first meeting' detail

78. Demis HassabisAugust 5, 2026editorial

    Discovery Loop name and Dean's title attributed to the right sources, an uncited superlative removed, and a mislink of 'conviction' to a VC firm's page removed

79. Recommender SystemAugust 5, 2026editorial

    GEM is a ranking-stage teacher model, not an example of autoregressive generative recommendation

80. GEM (Meta)August 5, 2026editorial

    the ranking-stage claim is from the Q2 2025 earnings call, citation added

81. TranshumanismAugust 5, 2026editorial

    restored dropped apostrophes in the Sandberg quote and unbundled three book topics the cited page lists separately

82. GPT-6 AstraAugust 5, 2026editorial

    restored the hedge in Alpoge's wording, he wrote "i think ehrhart is the only one so far"

83. NVIDIA Vera (CPU)August 5, 2026editorial

    the 164 MB L3 cache and Armv9.2 instruction set are now NVIDIA-published figures (August 2026 storage post), no longer only third-party reporting

84. Long-context language modelsAugust 5, 2026editorial

    the ALiBi slope formula was garbled; the paper defines the slopes as the geometric sequence m_h = 2^(-8h/H)

85. GPT-LiveAugust 5, 2026editorial

    OpenAI attributes single-UDP-packet session start to Instant Connect and WARP together, not to Instant Connect alone

86. Generative BionicsAugust 5, 2026editorial

    the fourth-quarter 2026 production target is in the CES release, not the July 20 announcement; citations split accordingly

87. Frontis-MA1August 5, 2026editorial

    removed an unsourced framing about third-party descriptions of the release; the cited sources state the license facts only

88. Claude Fable 5August 5, 2026editorial

    restored the hedge in Alpoge's wording, he wrote "i think ehrhart is the only one so far"

89. DeepSeek V4-FlashAugust 4, 2026editorial

    replaced two references to a source on the site's excluded list with IT之家 and Jiemian News coverage; the V4-Pro price reduction began as a late-April 2026 limited-time promotion, not a May permanent cut, and the peak-pricing plan was communicated in a June 29, 2026 email to users

90. Galbot G1August 4, 2026editorial

    funding table now matches the body's June 2025 date for the CATL round; the release sentence now matches its cited source, which dates the G1 release to March 2025; Ramesohl's title per the cited release is managing partner

91. GalbotAugust 4, 2026editorial

    aligned the yuan-to-dollar conversion with the cited Global Times figure, reworded a reporter narrative that had been presented as a company quote, added the reference that actually supports the Stanford PhD, and matched TechNode's wording on the factory deployments

92. Benchmark (AI)August 4, 2026editorial

    the claim that LMArena disputed the paper's figures now cites LMArena's own response post rather than the paper it responds to

93. WayveAugust 4, 2026editorial

    restored words silently dropped from the Kendall Series C quote

94. Tesla AI5August 4, 2026editorial

    the Terafab sentence cited an August 2025 article that predates the March 2026 announcement; recited to Tom's Hardware's announcement coverage and updated the stale 'sometimes referred to as' phrasing, Terafab is the project's official name

95. KAI SchedulerAugust 4, 2026editorial

    six (not five) v0.16.x releases shipped in July 2026, and v0.10.0 was published November 18, 2025 per the GitHub releases page, which contradicts the README's own 2025/10 label

96. TerafabAugust 4, 2026editorial

    the AI5 sentence was cited to sources that do not mention the tape-out; recited to Electrek's April 15, 2026 report, which describes TSMC handling AI5 while Samsung's 2 nm line is tied to AI6

97. Galbot G1August 4, 2026editorial

    GraspVLA first author is Deng Shengliang (was misnamed Jia); removed an unverifiable Peking University bachelor claim (sources conflict); fixed the March 2026 round investors per TechNode/Caixin; linked the new Galbot company page; June 2025 round date precision

98. SpaceX StarmindAugust 4, 2026editorial

    fixed citation integrity flagged in review (merger and IPO facts now cited to CNBC and NPR rather than the Space.com naming piece; Starcloud is not described as NVIDIA-backed, its Series A was led by Benchmark and EQT; unquoted a Tom's Hardware paraphrase presented as a Musk quote)

99. Shenzhen Robot ValleyAugust 4, 2026editorial

    removed compass directions for the flanking mountains that no cited source states

100. Qwen3.8-MaxAugust 4, 2026editorial

     attributed the open-weights reversal analysis to Ben Thompson (relayed by Simon Willison), and pointed the Artificial Analysis reference at the archived snapshot after the live page 404ed

101. NVIDIA H100August 1, 2026reader report

     Confirmed and fixed. A reader reported that the Tensor Core throughput figures in the H100 performance table were roughly double the correct values. They were: the dense column had been populated with NVIDIA's sparsity-assisted numbers, and the sparsity column with those numbers doubled again. The table now reports NVIDIA's shipping-product figures with dense and 2:4-structured-sparsity values in separate columns for both H100 SXM5 and H100 PCIe. Corrected dense peaks for the SXM5 are 494.7 TFLOPS TF32, 989.4 TFLOPS BF16 and FP16, and 1,978.9 TFLOPS/TOPS FP8 and INT8, with the sparsity-assisted figures at twice each value. The DGX H100 FP8 aggregate was corrected in the same pass. Thanks to the reader who spotted the inconsistency against our own NVIDIA B200 page.