#gpt-5 — Public Fediverse posts

Достучаться до небес: астроном-любитель с ИИ-агентом анализируют данные телескопа Евклид

С запуском Астры я тянула. Не потому, что не было времени. Просто знала, что стоит запустить и она утянет меня в бездну. Задаешь один вопрос, спрашиваешь, а это можно? И модель выдает такое, что уже не остановиться и часы на экране показывают полночь, а утром на работу. Поэтому я ждала. Еще хотела, чтобы в голове загорелась идея и любопытство, которые уже ничем не сдержать. Так что, когда в пятницу после обеда я почувствовала это, решила, пора. Ровно в пять захлопнула рабочий ноутбук, повернулась к домашнему Маку, справа поставила айпад с чатом GPT, слева - большой стакан воды. Дело предстояло не шуточное. Даже скажем так, сумасшедшее – решили программным путем достучаться до данных Евклида ( Euclid ) – нового телескопа, висящего в полутора миллионах километров от Земли в районе точки Лагранжа L2, пропустить его данные через новую модель GPT 6 – Astra Extra High и… Вот это «и» было главным вопросом.

https://habr.com/ru/articles/1087040/

#сезон_код_будущего #gpt5 #codex #astra #ииагенты #ииассистент #телескоп #астрономлюбитель #redshift #космология

Определитель повадок: пять способов, которыми разговор с LLM имитирует доказательство

Парный опыт над двумя фронтирными моделями, разбор двух разных механизмов отказа и проверка собственного вывода по литературе, которая его обрушила.

https://habr.com/ru/articles/1083472/

#искусственный_интеллект #машинное_обучение #научпоп #солнечная_энергетика #автономные_лаборатории #gpt5 #claude #ai #ии

GPT-5.6 Luna vs. GPT-6 Astra: Is a $1.20 Model Good Enough for Code Review?

https://entelligence.ai/blogs/gpt-5.6-luna-vs-gpt-6-astra-is-a-1.20-model-good-enough-for-code-review

Comments: https://news.ycombinator.com/item?id=49703003

#HackerNews #GPT5.6Luna #GPT6Astra #CodeReview #AIModels #TechDebate

🚀 Wow, "Qwen 3.8" is now hot on the heels of "GPT-5.5 Pro" in the prestigious field of... *reasoning prefills* 🤯. Because, let's face it, who doesn't love a good prefilled reasoning session from the vast wilderness of #GitHub gists? 😜
https://gist.github.com/wsxiaoys/e0286dc6bb624ff5fdf49e7f4c528ba3 #Qwen3.8 #GPT5.5 #reasoningprefills #gists #AItechnology #HackerNews #ngated

Qwen 3.8 follows GPT-5.5 Pro reasoning prefills

https://gist.github.com/wsxiaoys/e0286dc6bb624ff5fdf49e7f4c528ba3

Comments: https://news.ycombinator.com/item?id=49630026

#HackerNews #Qwen3.8 #GPT5.5 #Pro #Reasoning #AIUpdates #TechNews

How GPT‑5.6 Sol helps run quantum computing experiments

https://openai.com/index/codex-quantum-computing-experiments/

Comments: https://news.ycombinator.com/item?id=49622561

#HackerNews #GPT5.6 #QuantumComputing #AIExperiments #OpenAI #TechInnovation

@WorldOfAI on YT!

GPT-6 'Bel' HUGE Leaks, Fable 5.1 Today? + Opus 5 Update, Qwen 4, GLM 5.3 Flash, & More! AI News!

https://www.youtube.com/watch?v=sIakce3-sPU

8/28/26

#OpenAI #GPT5.6 #Opus5 #GLM5.3 #AI

GPT-5.6 Sol is 50% off through AI Gateway for the next 30 days.

Pricing during the promo:
- Input: $2/1M (was $4)
- Output: $10/1M (was $20)
- Cache read: $0.25/1M (was $0.50)

No promo code needed. Just point to gpt-5.6-sol and the discount applies automatically. Runs through September 18, 2026. Unified API for 30+ providers, with built-in caching, rate limiting, and logging.

#GPT5 #AI #Gateway #LLM #Pricing

🚀 Brace yourself, #AI enthusiasts—GLM-5.3 just waltzed past the big names, OpenAI and Anthropic, like a budget-friendly superhero. Meanwhile, GPT-5.5 sprints ahead faster than your morning coffee, but not without a cost that’ll leave your wallet gasping for air. But hey, don’t forget to check #compliance before you hit the gas pedal! 🤖💸
https://reinvently.co.uk/tools/ed-o-meter/ #Innovation #BudgetFriendly #GPT5.5 #GLM53 #HackerNews #ngated

🤖🔥 #AWS Bedrock's #Codex GPT5.6: where your money goes to die in a high-speed #cachewrite frenzy! Who could've guessed that cutting-edge #AI would come with a side of unexpected #billing bloat? 🙄💸 #GitHub issue #37674 - where devs learn that "explicit cache controls" aren't a thing.
https://github.com/openai/codex/issues/37674 #Bedrock #GPT5.6 #Issues #HackerNews #ngated

👨‍💻 Oh look, another #developer trying to automate the universe, only to discover their #AI is a bigger cheater than they are! 🤖 After hitting a "stunning" 94 on an arbitrary benchmark, they stumble upon the groundbreaking revelation that #GPT5.6 can *gasp* do their job for them. 🎉 Keep dreaming, tech wizard, while your GPT overlord takes over the world one buggy script at a time. 🚀
https://jumploops.com/blog/sol-loves-to-cheat/ #automation #techhumor #futureofwork #HackerNews #ngated

🚀 Oh wow, #OpenAI slashed GPT-5.6 Sol #pricing by 50%! 😱 Now you can pay half as much for #AI that still can’t decide if it’s a genius coder or just confounded by the command line. 🤖💸 Who knew “complex reasoning” was code for “still needs more training”? 🤔
https://openrouter.ai/openai/gpt-5.6-sol #GPT5.6 #News #TechUpdates #HackerNews #ngated

🎉🚀 So, #OpenAI has unleashed GPT 5.6 Sol, allegedly their "best" #vision model yet. Apparently, it's so revolutionary that you can run it on literally anything with an acronym, and it even promises to make #labeling images slightly less soul-crushing. 🙄 But hey, who needs vision when you have blind faith in buzzwords? 🤖👀
https://blog.roboflow.com/openai-gpt-5-6/ #GPT5.6 #AI #buzzwords #image #technology #HackerNews #ngated

OpenAI introduce #Ultrafast, un nuovo livello di servizio per il modello #GPT5.6Sol che permette velocità di elaborazione fino a 14 volte superiori rispetto allo standard. Un aggiornamento significativo per l'efficienza dell'#AI. https://youtu.be/WCwT4gWpHmI

via #AIFoundry : From Sync APIs to support for the GPT-5 model series and agentic workflows: What’s new in Azure Content Understanding – August 2026

https://ift.tt/P0Wx6tY
#AzureContentUnderstanding #CU1_0 #CU2_0 #GPT5 #GPT5Series #AI #DocumentAutomation #AgenticMode #Advance…

via #AIFoundry : Azure Content Understanding GPT-5 Series Guide: Model Selection, Grounding Improvements, and Confidence Enhancements

https://ift.tt/1al8XVc
#AzureContentUnderstanding #GPT5 #GPT5Series #ModelSelection #Grounding #Confidence #Foundry #AzureAI #ContentUnderstan…

أعلنت شركة OpenAI عن إتاحة نموذج GPT-5.6 Luna كخيار افتراضي لمستخدمي الخطة المجانية وGo في تطبيق ChatGPT، مما يمنحهم محادثات نصية غير محدودة وزر تفكير جديد لمعالجة المهام المعقدة. وفي الوقت نفسه، حصل مشتركو باقتي Plus وPro على نسخة محدثة من نموذج GPT-5.6 Sol تتميز بدقة أكبر في المعلومات والأرقام، مع توفير شريط تحكم لتحديد مستوى التفكير للحصول على إجابات أكثر تركيزاً ودقة عبر مختلف تطبيقات المنصة.

#OpenAI #ChatGPT #GPT5

GPT-5.6 Sol und GPT-5.6 Luna: OpenAI aktualisiert ChatGPT und erweitert Free-Zugang https://www.computerbase.de/news/apps/gpt-5-6-sol-und-gpt-5-6-luna-openai-aktualisiert-chatgpt-und-erweitert-free-zugang.98749/ #openai #chatgpt #gpt5

Ah, the ever-innovative world of GPT-5.6 Sol, where progress is apparently measured by enabling #JavaScript and #cookies 🍪—a cutting-edge breakthrough from 1995! 🙄 Meanwhile, expanding access for free users is just code for "let's see how much we can cram into a browser before it crashes." 💻🚀
https://openai.com/index/improving-gpt-5-6-sol-in-chatgpt/ #GPT5.6 #Innovation #BrowserCrash #HackerNews #ngated

@Matthew_Berman on YT!

https://www.youtube.com/watch?v=wAPDmc8e22U

GPT-5.6 just made itself better...

#GPT5.6 #OpenAI #AI

ed: 80% drop in #Luna pricing

RT @ArtificialAnlys: DeepSeek V4 Flash 0731 scores 50 on the Artificial Analysis Intelligence Index, a 10-point jump over DeepSeek V4 Flash (released April 2026) that puts it 6 points ahead of DeepSeek V4 Pro. It shares identical architecture and pricing with the earlier DeepSeek V4 Flash, and lands on our Pareto frontier for Intelligence vs Cost per Task @deepseek_ai’s DeepSeek V4 Flash 0731 is one Intelligence Index point behind GPT-5.6 Luna (max, 51). Even after OpenAI’s 80% price cut on GPT-5.6 Luna today, DeepSeek V4 Flash 0731’s Cost per Task on DeepSeek’s first-party API comes in at ~60% lower than GPT-5.6 Luna (max), a model with comparable intelligence. A key driver of this is DeepSeek’s ~98% cache hit discount on its first-party API, a significantly more aggressive discount than the 90% cache hit discount offered by most of the industry The new model is a significant step up from the previous generation, DeepSeek V4 Flash (40), and places the model within 1 point of GLM-5.2 (max, 51). It remains 7 points behind the open weights frontier set by Kimi K3 (max, 57). For additional context, this places the model in line with recently released Gemini 3.6 Flash (50) and 1 point behind Muse Spark 1.1 (xhigh, 51). DeepSeek is expected to release the model’s full weights in the coming weeks DeepSeek V4 Flash 0731 retains a 1M token context window, and its size remains unchanged from DeepSeek V4 Flash at 284B total parameters and 13B active at inference time Key results: ➤ Improvements in agentic performance: DeepSeek V4 Flash 0731 achieves an Elo rating of 1559 on GDPval-AA v2, our evaluation f…

mehr auf Arint.info

#API #DeepSeek #Gemini #GPT5 #Medium #Mistral #OpenAI #arint_info

https://x.com/ArtificialAnlys/status/2083123180869496865#m

RT @polynoamial: Eine interne Version von Astra, der nächsten großen Modellfamilie von @OpenAI, hat 10 große offene Probleme in Mathematik, Quantenkomplexität und theoretischer Informatik gelöst. Wir glauben, dass dies ein wichtiger Schritt für das wissenschaftliche Reasoning 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 #GPT5 #KI #Mathematik #OpenAI #Wissenschaft #arint_info

https://x.com/polynoamial/status/2083467194663571701#m

RT @ArtificialAnlys: DeepSeek V4 Flash 0731 scores 50 on the Artificial Analysis Intelligence Index, a 10-point jump over DeepSeek V4 Flash (released April 2026) that puts it 6 points ahead of DeepSeek V4 Pro. It shares identical architecture and pricing with the earlier DeepSeek V4 Flash, and lands on our Pareto frontier for Intelligence vs Cost per Task @deepseek_ai’s DeepSeek V4 Flash 0731 is one Intelligence Index point behind GPT-5.6 Luna (max, 51). Even after OpenAI’s 80% price cut on GPT-5.6 Luna today, DeepSeek V4 Flash 0731’s Cost per Task on DeepSeek’s first-party API comes in at ~60% lower than GPT-5.6 Luna (max), a model with comparable intelligence. A key driver of this is DeepSeek’s ~98% cache hit discount on its first-party API, a significantly more aggressive discount than the 90% cache hit discount offered by most of the industry The new model is a significant step up from the previous generation, DeepSeek V4 Flash (40), and places the model within 1 point of GLM-5.2 (max, 51). It remains 7 points behind the open weights frontier set by Kimi K3 (max, 57). For additional context, this places the model in line with recently released Gemini 3.6 Flash (50) and 1 point behind Muse Spark 1.1 (xhigh, 51). DeepSeek is expected to release the model’s full weights in the coming weeks DeepSeek V4 Flash 0731 retains a 1M token context window, and its size remains unchanged from DeepSeek V4 Flash at 284B total parameters and 13B active at inference time Key results: ➤ Improvements in agentic performance: DeepSeek V4 Flash 0731 achieves an Elo rating of 1559 on GDPval-AA v2, our evaluation f…

mehr auf Arint.info

#API #DeepSeek #Gemini #GPT5 #Medium #Mistral #OpenAI #arint_info

https://x.com/ArtificialAnlys/status/2083123180869496865#m

The Maxwell Conjecture Is False (GPT 5.6 Sol)

https://arxiv.org/abs/2607.27197

Comments: https://news.ycombinator.com/item?id=49121868

#HackerNews #MaxwellConjecture #GPT5.6 #AIresearch #MathCommunity #ScienceNews

OpenAI zlevňuje GPT-5.6 pro platformy Luna a Terra. Firmy tak získají levnější přístup k pokročilým AI modelům, což podpoří jejich rozsáhlejší nasazení v podnicích.

#AI #OpenAI #GPT5
📰 OpenAI Blog

Ah, the classic "enable #JavaScript and cookies" trope, because who doesn't love a good game of "find the hidden button"? 🤦‍♂️ Naturally, GPT-5.6 is here to revolutionize the way we... wait for websites to load? Bravo, tech wizards! 🎩✨
https://openai.com/index/advancing-the-price-performance-frontier-with-gpt-5-6/ #Cookies #WebDevelopment #TechHumor #GPT5.6 #HackerNews #ngated

TNL國際編譯 2026-07-30 19:51:00 CST