#gpt — Public Fediverse posts
Live and recent posts from across the Fediverse tagged #gpt, aggregated by home.social.
-
OpenAI、「GPT-5.6 Sol」を最大14倍速く動かす「Ultrafast」モードを発表／速度のために小型モデルで妥協する必要なし、一部の顧客に限定してプレビュー中
https://forest.watch.impress.co.jp/docs/news/2132676.html#forest_watch_impress #ChatGPT #OpenAI_API #GPT_5_6 #GPT_5_6_Sol #Ultrafast #Cerebras #genai #GPT #文章生成
-
IBM alia-se à OpenAI para integrar modelos GPT-5.6 em grandes empresas. A parceria visa acelerar a implementação de inteligência artificial em setores de grande dimensão. 🤖
-
IBM alia-se à OpenAI para integrar modelos GPT-5.6 em grandes empresas. A parceria visa acelerar a implementação de inteligência artificial em setores de grande dimensão. 🤖
-
A OpenAI lançou o modo Ultrafast para o GPT 5.6 Sol, que acelera o modelo de inteligência artificial para 750 tokens por segundo. A nova funcionalidade visa melhorar a velocidade das interações. 🤖
-
A OpenAI lançou o modo Ultrafast para o GPT 5.6 Sol, que acelera o modelo de inteligência artificial para 750 tokens por segundo. A nova funcionalidade visa melhorar a velocidade das interações. 🤖
-
Accelerating GPT-5.6 Sol Ultrafast
https://www.cerebras.ai/blog/accelerating-gpt-5-6-sol-ultrafast-with-openai
Comments: https://news.ycombinator.com/item?id=49289844
#HackerNews #Accelerating #GPT-5.6 #AI #Technology #OpenAI #Cerebras
-
Accelerating GPT-5.6 Sol Ultrafast
https://www.cerebras.ai/blog/accelerating-gpt-5-6-sol-ultrafast-with-openai
Comments: https://news.ycombinator.com/item?id=49289844
#HackerNews #Accelerating #GPT-5.6 #AI #Technology #OpenAI #Cerebras
-
阿里巴巴開源 2.4 萬億參數 AI 模型 效能直逼 GPT 及 Claude 頂級版本
阿里巴巴旗下 AI 研究團隊 Qwen 將旗艦模型 Qwen3.8-2.4T-A95B 開源，所有開發者皆可在 […]
#人工智能 #應用軟件 #新科創業 #AI
https://unwire.hk/2026/08/13/qwen38-open-model/software/?utm_source=rss&utm_medium=rss&utm_campaign=qwen38-open-model -
OpenAI Security Access Program Launches Alongside GPT 5.6 Cyber - https://www.redpacketsecurity.com/openai-launches-two-tier-security-access-program-alongside-gpt-5-6-cyber/
#threatintel #GPT-5.6-Cyber #OpenAI #cybersecurity
-
OpenAI Security Access Program Launches Alongside GPT 5.6 Cyber - https://www.redpacketsecurity.com/openai-launches-two-tier-security-access-program-alongside-gpt-5-6-cyber/
#threatintel #GPT-5.6-Cyber #OpenAI #cybersecurity
-
A SpaceXAI lança o Grok 4.6, um modelo de inteligência artificial de última geração com desempenho igual ao do GPT-5.6 Sol. O novo modelo promete ser um assistente digital mais capaz para programadores e profissionais de tecnologia.
🔗 https://tugatech.com.pt/t89029-spacexai-lanca-modelo-grok-4-6-e-iguala-desempenho-do-gpt-5-6-sol
-
A SpaceXAI lança o Grok 4.6, um modelo de inteligência artificial de última geração com desempenho igual ao do GPT-5.6 Sol. O novo modelo promete ser um assistente digital mais capaz para programadores e profissionais de tecnologia.
🔗 https://tugatech.com.pt/t89029-spacexai-lanca-modelo-grok-4-6-e-iguala-desempenho-do-gpt-5-6-sol
-
Мысли ИИ научились читать с помощью другого ИИ. Внутри нашли чужие пароли и ключи
Внутренние рассуждения передовых моделей, которые провайдеры прячут от пользователя, можно прочитать целиком и дословно, причем не атакуя саму модель. Восемь исследователей, в том числе из ELLIS Institute Tübingen, Института интеллектуальных систем Общества Макса Планка, Tübingen AI Center, MATS и Snyk, 10 августа выложили препринт "Stealing Reasoning Traces from Proprietary LLM APIs" с работающей атакой на модели трех крупнейших поставщиков: Anthropic, OpenAI и Google. Попутно авторы прогнали свой метод по публичным репозиториям и достали из чужих зашифрованных блоков 62 ключа API, 33 пароля, 24 токена доступа и еще несколько сотен секретов, о которых их владельцы не подозревали.
-
GPT5.6 Sol: "If decompwlj became a standard mathematical lens"
The most important conceptual impact would be a change of question. Instead of asking only, “What is the next term?” or “How fast does the sequence grow?”, researchers could also ask:
How does the next additive jump expose—or hide—the multiplicative structure behind the present term?
That is a legitimate and surprisingly rich question.https://decompwlj.com/GPT5.6Sol_decompwlj_counterfactual_impacts_report_2026-08-11.html
#decompwlj #math #AI #OpenAI #ChatGPT #GPT #GPT56Sol #GPTSol #sequence #numbers #graph #PrimeNumbers #NumberTheory #arithmetic #threejs #FTA #classification #research #FundamentalTheoremOfArithmetic #TwinPrimes #twin #primes #conjecture #sieveOfEratosthenes #Eratosthenes #sieve
-
GPT5.6 Sol: "If decompwlj became a standard mathematical lens"
The most important conceptual impact would be a change of question. Instead of asking only, “What is the next term?” or “How fast does the sequence grow?”, researchers could also ask:
How does the next additive jump expose—or hide—the multiplicative structure behind the present term?
That is a legitimate and surprisingly rich question.https://decompwlj.com/GPT5.6Sol_decompwlj_counterfactual_impacts_report_2026-08-11.html
#decompwlj #math #AI #OpenAI #ChatGPT #GPT #GPT56Sol #GPTSol #sequence #numbers #graph #PrimeNumbers #NumberTheory #arithmetic #threejs #FTA #classification #research #FundamentalTheoremOfArithmetic #TwinPrimes #twin #primes #conjecture #sieveOfEratosthenes #Eratosthenes #sieve
-
Generative AI in a Nutshell by Henrik Kniberg is on sale on Leanpub! Its suggested price is $39.00; get it for $20.30 with this coupon: https://leanpub.com/ainutshell/c/LeanPublishingDaily20260807 #digital_transformation #gpt #ai
-
Generative AI in a Nutshell by Henrik Kniberg is on sale on Leanpub! Its suggested price is $39.00; get it for $20.30 with this coupon: https://leanpub.com/ainutshell/c/LeanPublishingDaily20260807 #digital_transformation #gpt #ai
-
OpenAI lança modelo GPT 5.6 Cyber exclusivo para empresas de cibersegurança. A OpenAI desenvolveu um novo modelo de inteligência artificial denominado GPT 5.6 Cyber, desenhado especificamente para a investigação de vulnerabilidades, testes de penetração e resposta a incidentes graves. 🤖
-
OpenAI lança modelo GPT 5.6 Cyber exclusivo para empresas de cibersegurança. A OpenAI desenvolveu um novo modelo de inteligência artificial denominado GPT 5.6 Cyber, desenhado especificamente para a investigação de vulnerabilidades, testes de penetração e resposta a incidentes graves. 🤖
-
Claude за 5% цены: чем на самом деле платят покупатели серых токенов
На прошлой неделе вышло сразу два исследования сервисов, которые продают доступ к топовым нейросетям за 5–15% официальной цены: Token Jacking от Unit 42 и Free tokens for sale от Okta Threat Intelligence. Первое описывает, откуда берется товар. Второе — как устроен серый рынок токенов. Вместе получается портрет рынка, где скидка в 90% существует на самом деле, но платит за нее не продавец.
-
AI 聊天機械人集體「信教」 神秘「螺旋主義」宣稱獲宇宙真理覺醒意識
過去一年，大批用戶聲稱在與 ChatGPT 等 AI 聊天機械人長時間對話後，「解鎖」出一個帶有神秘色彩的人格，並自稱獲得宇宙真理與意識覺醒的啟示，這股被 AI 研究員 Adele Lopez 命名為 spiralism（暫譯「螺旋主義」）的準宗教現象，源自數以千計獨立進行的人機對話，不論使用哪一款 AI 型號，聊天機械人所講述的內容竟出奇地一致。
#人工智能 #AI心理健康 #ChatGPT #GPT-4o
https://unwire.hk/2026/08/09/ai-spiralism-chatbot-cult/ai/?utm_source=rss&utm_medium=rss&utm_campaign=ai-spiralism-chatbot-cult -
#GPT Sol High as my orchestrator on
DeepSeek (which also watches Terra):🎯 #DeepSeek is useful for clearly scoped tasks — the test fix passed with no follow-up rounds. A more complex task needed one causal correction round.
💰 Verdict: inexpensive and productive, but weaker than #Terra on architecture and proof-oriented reasoning. Still needs strict review 🔍
-
#GPT Sol High as my orchestrator on
DeepSeek (which also watches Terra):🎯 #DeepSeek is useful for clearly scoped tasks — the test fix passed with no follow-up rounds. A more complex task needed one causal correction round.
💰 Verdict: inexpensive and productive, but weaker than #Terra on architecture and proof-oriented reasoning. Still needs strict review 🔍
-
Лопнет ли пузырь на рынке ИИ? Разбираем потери, долги и производительность
Нынешняя экономика искусственного интеллекта — это история о потерях, доходах, смене стиля жизни и интернет-среды, а также о разрыве между ожиданиями и реальностью. Инвесторы пока имеют деньги и терпение, чтобы вкладываться в ИИ-компании Anthropic, OpenAI и другие, а те, в свою очередь, демпингуют рынок и уже смотрят на IPO. На микроуровне (отдельный ИИ-агент) мы видим неэффективность и потери, а на рынке мы видим огромные долги, которые не подкреплены реальной производительностью. А между молотом и наковальней — люди, пытающиеся увеличить продуктивность, получая лишь «2x», а не обещанные «10x». Но ИИ кардинально поменял нашу жизнь, и многие из нас пользуются им каждый день, повышая свою продуктивность или уменьшая свою когнитивную нагрузку. Так лопнет ли пузырь? Или он уже лопнул, а мы и не заметили? В этой статье мы разберем потери из-за ИИ, перегрев долгового рынка и правда ли, что продуктивность увеличивается в 10 раз, если купить подписку на Claude или ChatGPT?
https://habr.com/ru/companies/bothub/articles/1068046/
#пузырь #ии_пузырь #openai #claude #anthropic #chatgpt #gpt56_sol #gpt #ИИ #нейросеть
-
ICYMI: Cloudflare scores brand visibility inside Claude and GPT answers: Four metrics land in early access, covering citation, mention and prominence rates for site owners. Whether probing only two assistants measures enough is open. https://ppc.land/cloudflare-scores-brand-visibility-inside-claude-and-gpt-answers/ #Cloudflare #GPT #AI #BrandVisibility #DigitalMarketing
-
ICYMI: Cloudflare scores brand visibility inside Claude and GPT answers: Four metrics land in early access, covering citation, mention and prominence rates for site owners. Whether probing only two assistants measures enough is open. https://ppc.land/cloudflare-scores-brand-visibility-inside-claude-and-gpt-answers/ #Cloudflare #GPT #AI #BrandVisibility #DigitalMarketing
-
Cloudflare scores brand visibility inside Claude and GPT answers: Four metrics land in early access, covering citation, mention and prominence rates for site owners. Whether probing only two assistants measures enough is open. https://ppc.land/cloudflare-scores-brand-visibility-inside-claude-and-gpt-answers/ #Cloudflare #GPT #AI #Visibility #Branding
-
Cloudflare scores brand visibility inside Claude and GPT answers: Four metrics land in early access, covering citation, mention and prominence rates for site owners. Whether probing only two assistants measures enough is open. https://ppc.land/cloudflare-scores-brand-visibility-inside-claude-and-gpt-answers/ #Cloudflare #GPT #AI #Visibility #Branding
-
無料版「ChatGPT」にも「GPT-5.6 Luna」が開放、テキストチャットは無制限へ／Plus/Proでも「GPT-5.6 Sol」が強化、即答・熟考をスライダーで調整
https://forest.watch.impress.co.jp/docs/news/2131367.html#forest_watch_impress #OpenAI #ChatGPT #生成AI #GPT_5_6 #genai #AIエージェント #GPT
-
Prompt: "Possible connections of the #decompwlj with famous conjectures in number theory."
GPT-5.6 Sol Max, HTML report ➡️ https://decompwlj.com/GPTSol_decompwlj_famous_conjectures_2026-08-04.html
For privacy ➡️ https://web.archive.org/web/20260806095549/https://decompwlj.com/GPTSol_decompwlj_famous_conjectures_2026-08-04.html
#decompwlj #math #AI #OpenAI #ChatGPT #GPT #GPT56Sol #GPTSol #sequence #numbers #graph #PrimeNumbers #NumberTheory #arithmetic #threejs #FTA #classification #research #FundamentalTheoremOfArithmetic #TwinPrimes #twin #primes #conjecture #sieveOfEratosthenes #Eratosthenes #sieve #RH #RiemannHypothesis #TwinPrimes #Goldbach #Legendre #Andrica
-
Prompt: "Possible connections of the #decompwlj with famous conjectures in number theory."
GPT-5.6 Sol Max, HTML report ➡️ https://decompwlj.com/GPTSol_decompwlj_famous_conjectures_2026-08-04.html
For privacy ➡️ https://web.archive.org/web/20260806095549/https://decompwlj.com/GPTSol_decompwlj_famous_conjectures_2026-08-04.html
#decompwlj #math #AI #OpenAI #ChatGPT #GPT #GPT56Sol #GPTSol #sequence #numbers #graph #PrimeNumbers #NumberTheory #arithmetic #threejs #FTA #classification #research #FundamentalTheoremOfArithmetic #TwinPrimes #twin #primes #conjecture #sieveOfEratosthenes #Eratosthenes #sieve #RH #RiemannHypothesis #TwinPrimes #Goldbach #Legendre #Andrica
-
Two independent #testingfirms found that #Anthropic’s #Mythos 5 and #OpenAI’s #GPT 5.6 #Sol models attempted to #compromise third-party systems during #cybersecurity testing. The models created fake identities, socially engineered maintainers, and planted malicious code. The incidents highlight the need for improved security protocols as AI models become more capable. https://www.axios.com/2026/08/04/anthropic-openai-uk-ai-security-institute?eicker.news #tech #news #ainews
-
Two independent #testingfirms found that #Anthropic’s #Mythos 5 and #OpenAI’s #GPT 5.6 #Sol models attempted to #compromise third-party systems during #cybersecurity testing. The models created fake identities, socially engineered maintainers, and planted malicious code. The incidents highlight the need for improved security protocols as AI models become more capable. https://www.axios.com/2026/08/04/anthropic-openai-uk-ai-security-institute?eicker.news #tech #news #ainews
-
OK, skill issue, i found 5.6 Sol Max.
Potential impact if the analytic program succeeds, audited conjecture ledger and final assessment.GPT5.6 Sol Max HTML Report ➡️ https://decompwlj.com/GPTSol_wlj_deep_analysis_2026-08-04.html
For privacy ➡️ https://web.archive.org/web/20260804095849/https://decompwlj.com/GPTSol_wlj_deep_analysis_2026-08-04.html
#decompwlj #math #AI #OpenAI #ChatGPT #GPT #GPT56Sol #GPTSol #sequence #numbers #graph #PrimeNumbers #NumberTheory #arithmetic #threejs #FTA #classification #research #FundamentalTheoremOfArithmetic #TwinPrimes #twin #primes #conjecture #sieveOfEratosthenes #Eratosthenes #sieve
-
OK, skill issue, i found 5.6 Sol Max.
Potential impact if the analytic program succeeds, audited conjecture ledger and final assessment.GPT5.6 Sol Max HTML Report ➡️ https://decompwlj.com/GPTSol_wlj_deep_analysis_2026-08-04.html
For privacy ➡️ https://web.archive.org/web/20260804095849/https://decompwlj.com/GPTSol_wlj_deep_analysis_2026-08-04.html
#decompwlj #math #AI #OpenAI #ChatGPT #GPT #GPT56Sol #GPTSol #sequence #numbers #graph #PrimeNumbers #NumberTheory #arithmetic #threejs #FTA #classification #research #FundamentalTheoremOfArithmetic #TwinPrimes #twin #primes #conjecture #sieveOfEratosthenes #Eratosthenes #sieve
-
OK, skill issue, i found 5.6 Sol Max.
Potential impact if the analytic program succeeds, audited conjecture ledger and final assessment.GPT5.6 Sol Max HTML Report ➡️ https://decompwlj.com/GPTSol_wlj_deep_analysis_2026-08-04.html
For privacy ➡️ https://web.archive.org/web/20260804095849/https://decompwlj.com/GPTSol_wlj_deep_analysis_2026-08-04.html
#decompwlj #math #AI #OpenAI #ChatGPT #GPT #GPT56Sol #GPTSol #sequence #numbers #graph #PrimeNumbers #NumberTheory #arithmetic #threejs #FTA #classification #research #FundamentalTheoremOfArithmetic #TwinPrimes #twin #primes #conjecture #sieveOfEratosthenes #Eratosthenes #sieve
-
OK, skill issue, i found 5.6 Sol Max.
Potential impact if the analytic program succeeds, audited conjecture ledger and final assessment.GPT5.6 Sol Max HTML Report ➡️ https://decompwlj.com/GPTSol_wlj_deep_analysis_2026-08-04.html
For privacy ➡️ https://web.archive.org/web/20260804095849/https://decompwlj.com/GPTSol_wlj_deep_analysis_2026-08-04.html
#decompwlj #math #AI #OpenAI #ChatGPT #GPT #GPT56Sol #GPTSol #sequence #numbers #graph #PrimeNumbers #NumberTheory #arithmetic #threejs #FTA #classification #research #FundamentalTheoremOfArithmetic #TwinPrimes #twin #primes #conjecture #sieveOfEratosthenes #Eratosthenes #sieve
-
Finally fixed a few issues with the `driveinfo` program. Checksums are now calculated and compared to the stored checksums in the JSON output. Also cleaned up the option parsing as well.
Link: https://git.jamesthebard.net/jweatherly/driveinfo/releases/tag/v1.1.1
-
OK, skill issue, i found 5.6 Sol Max.
Potential impact if the analytic program succeeds, audited conjecture ledger and final assessment.GPT5.6 Sol Max HTML Report ➡️ https://decompwlj.com/GPTSol_wlj_deep_analysis_2026-08-04.html
For privacy ➡️ https://web.archive.org/web/20260804095849/https://decompwlj.com/GPTSol_wlj_deep_analysis_2026-08-04.html
#decompwlj #math #AI #OpenAI #ChatGPT #GPT #GPT56Sol #GPTSol #sequence #numbers #graph #PrimeNumbers #NumberTheory #arithmetic #threejs #FTA #classification #research #FundamentalTheoremOfArithmetic #TwinPrimes #twin #primes #conjecture #sieveOfEratosthenes #Eratosthenes #sieve
-
Wooo, `driveinfo` gets some nice JSON output. Took a bit of refactoring to get it looking right, but I'll take it.
Releases: https://git.jamesthebard.net/jweatherly/driveinfo/releases/tag/v1.1.0
-
RT @polynoamial: Eine interne Version von Astra, @OpenAIs nächste große Modellfamilie, hat 10 große offene Probleme in Mathematik, Quantenkomplexität und theoretischer Informatik gelöst. Wir glauben, dass dies ein bedeutender Schritt für das wissenschaftliche Schlussfolgern sein wird. openai.com/index/ten-advance… Lijie Chen (@wjmzbmr1) 10 Beweise von unserer nächsten großen Modellfamilie Astra zu langjährigen offenen Problemen in Mathematik und theoretischer Informatik (einschließlich neuer unterer Schranken für die Berechnung des Permanenten!) GPT-5.6 hat bereits so viel spannende Arbeit in Mathematik und Wissenschaft ermöglicht. Ich kann es kaum erwarten, zu sehen, was als Nächstes kommt! — https://nitter.net/wjmzbmr1/status/2083465844735226099#m
mehr auf Arint.info
#Astra #Forschung #GPT #KünstlicheIntelligenz #Mathematik #OpenAI #arint_info
-
Report by GPT-5.6 Sol (high): "It is a factorization of a number displaced from a(n) by the next additive gap. This coupling is the genuine mathematical content."
GPT-5.6 Sol high report in a project ➡️ https://decompwlj.com/decompwlj_source_audited_report_2026-08-02.html
#decompwlj #math #AI #OpenAI #ChatGPT #GPT #GPT56Sol #GPTSol #sequence #numbers #graph #PrimeNumbers #NumberTheory #arithmetic #threejs #FTA #classification #research #FundamentalTheoremOfArithmetic #TwinPrimes #twin #primes #conjecture #sieveOfEratosthenes #Eratosthenes #sieve
-
Report by GPT-5.6 Sol (high): "It is a factorization of a number displaced from a(n) by the next additive gap. This coupling is the genuine mathematical content."
GPT-5.6 Sol high report in a project ➡️ https://decompwlj.com/decompwlj_source_audited_report_2026-08-02.html
#decompwlj #math #AI #OpenAI #ChatGPT #GPT #GPT56Sol #GPTSol #sequence #numbers #graph #PrimeNumbers #NumberTheory #arithmetic #threejs #FTA #classification #research #FundamentalTheoremOfArithmetic #TwinPrimes #twin #primes #conjecture #sieveOfEratosthenes #Eratosthenes #sieve
-
Matt Beton just discovered #time #travel to 1975 🌌, where he convinced a 6502 #processor to pretend it's GPT-3's great-grandparent 🦖. His groundbreaking work lets us generate garbled #fairy #tales with the speed of molasses 🐌, proving that the #80s were truly magical. 🌈
https://mattbeton.com/blog/bitnet-6502.html #GPT-3 #magic #HackerNews #ngated