Itay Z

PM at Nvidia. Wharton. no fluff

[Itay Z @itayzit]

Interesting take: Every inference provider optimizes for low latency (baseten, fireworks, together ai) Why? all because of Cursor.

@neilmovva says: this was correct a year ago, when you needed immediate response from your agent. Next year, background work will be 90% of the work, so latency doesn't matter. All that matters is *token cost*.

[Itay Z @itayzit]

"The Euler writeup, in particular, can only be described as AI slop"
@__alpoge__ on the blowups they found to the Navier-Stokes equation.
Slop happens to the best of us I guess..

[Itay Z @itayzit]

@YZitvar you? echo chamber?
(got it?)

[Itay Z @itayzit]

1. It’s crazy how small the claude fraction is- shows how much I live in any echo chamber
2. Gemini is shrinking- somebody at google is fired right now

[Similarweb @Similarweb]

Gen AI website traffic share update - August 2026

🗓️ 12 months ago:
ChatGPT: 73.3%
Gemini: 12.9%
DeepSeek: 4.0%
Grok: 2.6%
Copilot: 2.0%
Perplexity: 2.0%
Claude: 1.9%

🗓️ 6 months ago:
ChatGPT: 56.7%
Gemini: 25.4%
Claude: 6.0%
Grok: 3.7%
DeepSeek: 3.4%
Copilot: 1.9%
Perplexity: 1.6%

🗓️ 3 months ago:
ChatGPT: 52.7%
Gemini: 27.8%
Claude: 9.2%
DeepSeek: 3.6%
Grok: 2.5%
Copilot: 2.0%
Perplexity: 1.1%

🗓️ 2 months ago:
ChatGPT: 54.2%
Gemini: 26.5%
Claude: 9.5%
DeepSeek: 3.4%
Grok: 2.5%
Copilot: 1.6%
Perplexity: 1.0%

🗓️ 1 month ago:
ChatGPT: 55.5%
Gemini: 25.6%
Claude: 9.3%
DeepSeek: 3.4%
Grok: 2.4%
Copilot: 1.6%
Perplexity: 0.9%

[Itay Z @itayzit]

@liadagmon וואלה. עושה את זה עם קלוד והוא סתם מתאבסס על איזה pattern ספציפי כל פעם. אנסה עם הצ׳אט 💪🏻

[Itay Z @itayzit]

@choroshin פרומפט שמתקן את זה: תמחק כל סקיל שמגיע מ- skills il. הם עושים קליק בייט שגוי, אין מצב שיש שם סקילים טובים. שמחתי 🤗

[Itay Z @itayzit]

@urieli17 @JasonSCui תחום מטורף עם מלא מה לעשות, אבל לדעתי קשה להתמודד בו בלי כיסים גדולים שמספיקים לפחות כדי שתהיה לך תשתית אינפרנס משלך לאפטם עליה. ככה השחקנים הגדולים (baseten, fireworks, together) עושים

[Itay Z @itayzit]

@emollick can they automate @DataColada ‘s work? find the next Ginos and Arielys

[Itay Z @itayzit]

To all Israeli cyber/AI startups- Artificial Analysis should be your no.1 customer customer
I bet the GPT 6.x agents are trying to break into it as we speak to get the data from the benchmarks.

[Artificial Analysis @ArtificialAnlys]

Announcing Artificial Analysis Intelligence Index v4.2. We are accelerating elements of our upcoming v5 release with interim updates to keep pace with the frontier. Index v4.2 has more complex and realistic tasks, and more private test sets to prevent gaming

Intelligence Index v4.2 changelog:

+ AA-Briefcase, our agentic knowledge work evaluation with a private test set

+ @HelloSurgeAI's GDP.pdf, long context document reasoning across 4,592 PDF pages

- GPQA Diamond, an exceptional scientific reasoning evaluation that has now been saturated

… plus greater weighting on held-out test sets to prevent gaming, and grading infrastructure upgrades to increase robustness

This update brings the Index closer to real-world use cases with more challenging, complex and realistic tasks and private test sets to prevent gaming. We have been planning and building elements of Index v5 for months - it’s been 8 months since we launched Index v4 in January.

We have deliberately held back updates to keep the Index stable through recent major model launches. However, with the frontier moving so quickly in the past weeks, we feel it is important to deliver an immediate interim update to ensure our Index remains as relevant and useful as ever to users.

Beyond this interim update, our team is hard at work on v5 of the Index. We are planning more incremental releases in the near future. Stay tuned!

Intelligence Index v4.2 changes in detail:

➤ Adding AA-Briefcase: Our in-house evaluation with a private held-out test set, AA-Briefcase tests models on realistic agentic knowledge work tasks in complex projects built by industry experts. Models are evaluated on multi-week knowledge work projects, each with many linked tasks and thousands of input source files. AA-Briefcase combines rubric and pairwise grading to evaluate verifiable task success, analytical quality, and presentation quality, giving a holistic view of overall agentic capability in knowledge work.

➤ Adding GDP.pdf: Created by @HelloSurgeAI, GDP.pdf evaluates single-turn professional document reasoning across 100 PDFs and ten domains. Models must synthesize evidence distributed across 4,592 pages, including text, tables, charts, footnotes, and exclusions. Responses are graded against 1,275 expert-authored atomic criteria; the headline All-pass Rate credits a task only when every criterion is satisfied.

➤ Weighting to measure real-world use and prevent gaming: 40% of our Index weighting is now private, held-out test sets - double the figure from v4.1. Held-out data includes AA-Briefcase, AA-Omniscience, and solutions for CritPt. This reduces the ability for labs to game evaluations. The held-out percentage will increase further in Index v5.

➤ Improving our grading infrastructure: In AA-LCR v1.1, we have added a grading system prompt and corrected errors and ambiguities in answer keys, improving scoring accuracy. For GDPval-AA v2 and AA-Briefcase, we have improved our sampling and re-anchored the Elo scale, making ratings more stable as new models are added. For SciCode we have improved robustness of grading sandboxes to ensure slow but correct code does not count as a failure.

Key results:

➤ Anthropic and OpenAI lead the Index: Anthropic’s Claude Fable 5.1 leads the Index, followed by OpenAI’s GPT-6 Astra, which shows a 4pt gain over GPT-5.6 Sol. Meta is the third-ranked lab on the leaderboard, followed by SpaceXAI, Moonshot/Kimi, Z AI, and Google

➤ Cost per Task Pareto frontier shared by four labs: Anthropic, OpenAI, Meta and Z AI occupy the updated Cost per Task Pareto frontier

➤ GPT-6 Astra dominates the output token Pareto frontier: GPT-6 Astra is more token efficient than almost every other model near the intelligence frontier, with Claude Fable 5.1, Grok 4.5 and Gemini 3.5 Flash-Lite at either end of the curve (excludes models below 25 on the Index)

[Itay Z @itayzit]

@omooretweets he’s just jealous

[Itay Z @itayzit]

ok sure if you say so

[Itay Z @itayzit]

@llm_wizard chris the wiz, repainted

[Itay Z @itayzit]

@cHHillee it makes sense though. it’s like tests in a software app: the tests are keeping the app sane, and who’s testing the tests? the app.
same goes for benchmarks

[Itay Z @itayzit]

@emollick message boards are the new surveillance technology

[Itay Z @itayzit]

@urieli17 לגמרי. הייתי אומר שזה נכון כל עוד כל המערכת והקונטקסט שאתה מביא נצרך בצורת טקסט. אם אנחנו הולכים לעולם שבו המודל משנה משקולות, או מאחסן מידע בצורה של וקטורים ולא טקסט, אז אולי פתאום כן יהיו switching costs

[Itay Z @itayzit]

@MikeE_3_14 ״מעבר ממודלים סטטיסטיים ללוגיקה דטרמיניסטית״- אם כבר המעבר הוא הפוך. פעם חשבו שהמודלים יעבדו ב- lean, היום רוב ההוכחות נפתרות פשוט באנגלית.

[Itay Z @itayzit]

@shacharlotan @ReemSherman תבלו כפרות

[Itay Z @itayzit]

@AndrewCurran_ His name was Claude Louis Navier for a reason right

[Itay Z @itayzit]

@AndrewCurran_ His name was Claude Louis Navier for a reason right

[Fei-Fei Li @drfeifei]

@a16z I think kids today call this “aurafarming”; am cooked.😆