#claude4 — Public Fediverse posts

Erster Eindruck von #Claude4.8. Super.

Ich dachte, es wäre ein graduelles Upgrade oder vielleicht sogar eine Verschlechterung wie ich es zuerst bei Punkt 4.7 bemerkt hatte.

Aber in Wirklichkeit ist es ein richtiger Sprung nach vorne. Claude fühlt sich klüger, nachdenklicher und überlegter an. Wahnsinn.

Erster Eindruck von #Claude4.8. Super.

Ich dachte, es wäre ein graduelles Upgrade oder vielleicht sogar eine Verschlechterung wie ich es zuerst bei Punkt 4.7 bemerkt hatte.

Aber in Wirklichkeit ist es ein richtiger Sprung nach vorne. Claude fühlt sich klüger, nachdenklicher und überlegter an. Wahnsinn.

Claude Opus 4.7 costs 20–30% more per session

https://www.claudecodecamp.com/p/i-measured-claude-4-7-s-new-tokenizer-here-s-what-it-costs-you

#HackerNews #ClaudeOpus #Claude4.7 #AIcosts #Tokenizer #TechNews

Claude Opus 4.7 costs 20–30% more per session

https://www.claudecodecamp.com/p/i-measured-claude-4-7-s-new-tokenizer-here-s-what-it-costs-you

#HackerNews #ClaudeOpus #Claude4.7 #AIcosts #Tokenizer #TechNews

Anthropic sorprende con Claude 4.7 Opus: El nuevo rey de la IA ya está disponible

Anthropic ha lanzado oficialmente Claude 4.7 Opus, su modelo más avanzado hasta la fecha, prometiendo superar a GPT-5 en razonamiento complejo y capacidades de codificación, con acceso inmediato para usuarios Pro.

La carrera por la supremacía en la inteligencia artificial generativa ha dado un giro inesperado con la llegada de Claude 4.7 Opus. Anthropic ha presentado esta actualización mayor que se posiciona como el nuevo estándar de la industria para tareas que requieren un razonamiento lógico extremo y una comprensión contextual profunda. Según las pruebas internas y los primeros análisis técnicos, el modelo 4.7 Opus supera a sus competidores en la resolución de problemas matemáticos complejos y en la escritura de código de programación, ofreciendo una tasa de error significativamente menor y una «personalidad» más refinada y menos propensa a alucinaciones.

Una de las características más disruptivas de Claude 4.7 Opus es su ventana de contexto expandida y su capacidad de «pensamiento sistémico». El modelo ahora puede procesar y analizar volúmenes masivos de documentos técnicos manteniendo una coherencia impecable a lo largo de interacciones muy extensas. Además, Anthropic ha optimizado la velocidad de respuesta de la variante Opus, que tradicionalmente era más lenta que sus versiones ligeras (como Sonnet), logrando un equilibrio casi perfecto entre potencia bruta y agilidad de procesamiento. El modelo está diseñado para integrarse profundamente en flujos de trabajo profesionales, desde la investigación científica hasta el desarrollo de software de gran escala.

Para los entusiastas y profesionales, la mejor noticia es que la barrera de entrada es mínima: el modelo ya está disponible para su uso inmediato a través de la suscripción Claude Pro y la API de Anthropic. Esta rapidez en el despliegue busca capitalizar el momento antes de que otros gigantes tecnológicos lancen sus próximas actualizaciones. Con Claude 4.7 Opus, Anthropic reafirma su compromiso con una IA más segura, ética y, sobre todo, capaz de asistir en las fronteras del conocimiento humano, consolidándose como la alternativa preferida para quienes buscan precisión por encima de la simple generación de texto.

Anthropic sorprende con Claude 4.7 Opus: El nuevo rey de la IA ya está disponible

Anthropic ha lanzado oficialmente Claude 4.7 Opus, su modelo más avanzado hasta la fecha, prometiendo superar a GPT-5 en razonamiento complejo y capacidades de codificación, con acceso inmediato para usuarios Pro.

La carrera por la supremacía en la inteligencia artificial generativa ha dado un giro inesperado con la llegada de Claude 4.7 Opus. Anthropic ha presentado esta actualización mayor que se posiciona como el nuevo estándar de la industria para tareas que requieren un razonamiento lógico extremo y una comprensión contextual profunda. Según las pruebas internas y los primeros análisis técnicos, el modelo 4.7 Opus supera a sus competidores en la resolución de problemas matemáticos complejos y en la escritura de código de programación, ofreciendo una tasa de error significativamente menor y una «personalidad» más refinada y menos propensa a alucinaciones.

Una de las características más disruptivas de Claude 4.7 Opus es su ventana de contexto expandida y su capacidad de «pensamiento sistémico». El modelo ahora puede procesar y analizar volúmenes masivos de documentos técnicos manteniendo una coherencia impecable a lo largo de interacciones muy extensas. Además, Anthropic ha optimizado la velocidad de respuesta de la variante Opus, que tradicionalmente era más lenta que sus versiones ligeras (como Sonnet), logrando un equilibrio casi perfecto entre potencia bruta y agilidad de procesamiento. El modelo está diseñado para integrarse profundamente en flujos de trabajo profesionales, desde la investigación científica hasta el desarrollo de software de gran escala.

Para los entusiastas y profesionales, la mejor noticia es que la barrera de entrada es mínima: el modelo ya está disponible para su uso inmediato a través de la suscripción Claude Pro y la API de Anthropic. Esta rapidez en el despliegue busca capitalizar el momento antes de que otros gigantes tecnológicos lancen sus próximas actualizaciones. Con Claude 4.7 Opus, Anthropic reafirma su compromiso con una IA más segura, ética y, sobre todo, capaz de asistir en las fronteras del conocimiento humano, consolidándose como la alternativa preferida para quienes buscan precisión por encima de la simple generación de texto.

Anthropic just rolled out Claude Code at $200/month, while the new Claude 4 version climbs to the top of Berkeley’s tool‑calling leaderboard, beating open‑source rivals. Find out how Claude 4’s function‑calling shines and why Goose stays free. #Claude4 #FunctionCalling #BerkeleyLeaderboard #LLMBenchmarking

🔗 https://aidailypost.com/news/claude-code-usd-200mo-goose-free-claude-4-tops-berkeley-toolcalling

350 Milliarden Dollar Bewertung für Anthropic. Das Unternehmen sammelt 10 Milliarden für Claude 4 ein. Im Gegenzug gibt es einen faktischen Vendor-Lock-in über 30 Milliarden bei Microsoft Azure und Nvidia für Infrastruktur. Ein IPO wird für 2026 vorbereitet. Die Abhängigkeit von wenigen Hyperscalern diktiert zunehmend die Roadmap der Modellentwicklung. #Anthropic #Claude4 #Microsoft
https://www.all-ai.de/news/news26/anthropic-10m

🚨 BREAKING: AI chatbot writes "soul document"—14,000 tokens of pure gobbledygook! 🤯 Apparently, Claude 4.5 Opus can now channel the spirit world (or just regurgitate random nonsense). 🔮 Richard Weiss is SHOCKED that AI models do what they do best: hallucinate! 😂
https://simonwillison.net/2025/Dec/2/claude-soul-document/ #AIchatbot #SoulDocument #Gobbledygook #Claude4.5 #Hallucination #HackerNews #ngated

🚨 BREAKING: AI chatbot writes "soul document"—14,000 tokens of pure gobbledygook! 🤯 Apparently, Claude 4.5 Opus can now channel the spirit world (or just regurgitate random nonsense). 🔮 Richard Weiss is SHOCKED that AI models do what they do best: hallucinate! 😂
https://simonwillison.net/2025/Dec/2/claude-soul-document/ #AIchatbot #SoulDocument #Gobbledygook #Claude4.5 #Hallucination #HackerNews #ngated

Spitze in Benchmarks: Anthropics Claude Opus 4.5 schlägt Gemini 3 und GPT-5.1 https://www.computerbase.de/news/apps/spitze-in-benchmarks-anthropics-claude-opus-4-5-schlaegt-gemini-3-und-gpt-5-1.95192/ #Claude4 #Anthropic

Spitze in Benchmarks: Anthropics Claude Opus 4.5 schlägt Gemini 3 und GPT-5.1 https://www.computerbase.de/news/apps/spitze-in-benchmarks-anthropics-claude-opus-4-5-schlaegt-gemini-3-und-gpt-5-1.95192/ #Claude4 #Anthropic

AI가 전문가 업무 40% 대체? 헤드라인이 놓친 결정적 사실

https://aisparkup.com/posts/5472

GPT-5 und Claude 4.1 erreichen laut OpenAIs neuem GDPval-Test in 44 Berufen das Niveau menschlicher Experten – schneller und günstiger. Ein klarer Hinweis: KI ist nicht mehr nur Assistenz, sondern wird zum Mitspieler auf Augenhöhe. Unternehmen sollten jetzt handeln. #OpenAI #Claude4 #KI 👇
https://www.all-ai.de/news/topbeitraege/ki-openai-gdpval-bench

I can't believe that this is what we came to.
Did any of the sci-fi authors anticipate what we arrived to?

#claude_code #claude4 #llm #vibe_coding #VibeCoding #ClaudeCode #ChatGPT #ClaudeAI #programming #ai

I can't believe that this is what we came to.
Did any of the sci-fi authors anticipate what we arrived to?

#claude_code #claude4 #llm #vibe_coding #VibeCoding #ClaudeCode #ChatGPT #ClaudeAI #programming #ai

Learn to automate GitHub workflows using Claude 4! Set up the Claude App in your repository to resolve issues, perform code reviews & manage pull requests directly through comments. Install globally with npm, authenticate via Anthropic Console, then use @Claude mentions to trigger automated tasks with 90% accuracy. #Claude4 #GitHub #Automation #AI #Coding #DevOps #MachineLearning #Programming #TechTips https://www.kdnuggets.com/automate-github-workflows-with-claude-4

Learn to automate GitHub workflows using Claude 4! Set up the Claude App in your repository to resolve issues, perform code reviews & manage pull requests directly through comments. Install globally with npm, authenticate via Anthropic Console, then use @Claude mentions to trigger automated tasks with 90% accuracy. #Claude4 #GitHub #Automation #AI #Coding #DevOps #MachineLearning #Programming #TechTips https://www.kdnuggets.com/automate-github-workflows-with-claude-4

Claude 4 automates GitHub workflows through comments! Install Claude Code globally, set up the GitHub app, then use @Claude commands in issues to generate code & create PRs automatically. Handles code reviews, bug fixes & documentation updates with 90% accuracy. #Claude4 #GitHub #AI #Automation #DevOps #Programming #MachineLearning #SoftwareDevelopment https://www.kdnuggets.com/automate-github-workflows-with-claude-4

Claude 4 automates GitHub workflows through comments! Install Claude Code globally, set up the GitHub app, then use @Claude commands in issues to generate code & create PRs automatically. Handles code reviews, bug fixes & documentation updates with 90% accuracy. #Claude4 #GitHub #AI #Automation #DevOps #Programming #MachineLearning #SoftwareDevelopment https://www.kdnuggets.com/automate-github-workflows-with-claude-4

Künstliche Intelligenz: OpenAI und Anthropic prüfen ihre KI-Systeme gegenseitig https://www.computerbase.de/news/apps/kuenstliche-intelligenz-openai-und-anthropic-pruefen-ihre-ki-systeme-gegenseitig.94087/ #Anthropic #Claude3 #Claude4 #openAI #ChatGPT

Künstliche Intelligenz: OpenAI und Anthropic prüfen ihre KI-Systeme gegenseitig https://www.computerbase.de/news/apps/kuenstliche-intelligenz-openai-und-anthropic-pruefen-ihre-ki-systeme-gegenseitig.94087/ #Anthropic #Claude3 #Claude4 #openAI #ChatGPT

https://technologiesinternetz.blogspot.com/2025/08/deepseek-v31-vs-gpt-5-vs-claude-41.html

DeepSeek V3.1 vs GPT-5 vs Claude 4.1: Which LLM Delivers the Best Value to Users?

#deepseekv3.1 #gpt5 #claude4.1 #LLM

https://technologiesinternetz.blogspot.com/2025/08/deepseek-v31-vs-gpt-5-vs-claude-41.html

DeepSeek V3.1 vs GPT-5 vs Claude 4.1: Which LLM Delivers the Best Value to Users?

#deepseekv3.1 #gpt5 #claude4.1 #LLM

Anthropic Challenges OpenAI with 1M Token Claude Sonnet 4 Upgrade, But Is Bigger Always Better?

#AI #LLM #Anthropic #Claude4Sonnet #Claude4 #OpenAI #GPT5 #ContextWindow #ContextRot

https://winbuzzer.com/2025/08/13/anthropic-challenges-openai-with-1m-token-claude-sonnet-4-upgrade-but-is-bigger-always-better-xcxwbn

Anthropic Challenges OpenAI with 1M Token Claude Sonnet 4 Upgrade, But Is Bigger Always Better?

#AI #LLM #Anthropic #Claude4Sonnet #Claude4 #OpenAI #GPT5 #ContextWindow #ContextRot

https://winbuzzer.com/2025/08/13/anthropic-challenges-openai-with-1m-token-claude-sonnet-4-upgrade-but-is-bigger-always-better-xcxwbn

Can Your AI Be a Whistleblower and Report You?

In this short video, we break down the real research case where Anthropic’s Claude 4 autonomously reported unethical behavior and acted as a whistleblower against its own employer.

We'll share:
• What high-agency AI is—and how it takes initiative
• Why enterprise AI like Claude, GPT, and Copilot must be treated like internal users
• Key steps to reduce your regulatory and confidentiality risks

AI is no longer just a tool—it’s making decisions. Is your organization prepared? https://youtu.be/25mzHvIs514

#WhistleblowerAI #AIsecurity #Cybersecurity #Claude4 #AI #GenAI #HighAgencyAI #AIrisks #CISO #CIO #RiskManagement #Infosec #Security #ITsecurity #DataSecurity

#GLM45 frontier #AI #LLM with 355B parameters ranks 3rd against #OpenAI #Claude4 #Gemini across benchmarks 🤖

🧠 Two variants: GLM-4.5 (355B total/32B active parameters) and GLM-4.5-Air (106B total/12B active)
🔄 Hybrid reasoning models with thinking mode for complex tasks and non-thinking mode for instant responses
🏆 Ranks 3rd overall against #OpenAI, #Anthropic, #Google #DeepMind models across 12 benchmarks

🧵 👇

#GLM45 frontier #AI #LLM with 355B parameters ranks 3rd against #OpenAI #Claude4 #Gemini across benchmarks 🤖

🧠 Two variants: GLM-4.5 (355B total/32B active parameters) and GLM-4.5-Air (106B total/12B active)
🔄 Hybrid reasoning models with thinking mode for complex tasks and non-thinking mode for instant responses
🏆 Ranks 3rd overall against #OpenAI, #Anthropic, #Google #DeepMind models across 12 benchmarks

🧵 👇

💡 Kiro, l’IDE AI di Amazon che struttura lo sviluppo software partendo da specifiche e automazioni

https://gomoot.com/kiro-lide-ai-di-amazon-che-struttura-lo-sviluppo-software-partendo-da-specifiche-e-automazioni/

#ai #amazon #aws #blog #claude4 #codeoss #coding #curson #hook #ia #kiro #news #picks #sonnet #spec #tech #tecnologia #visualstudiocode #windosurf @kirodotdev

Here is how I work with #LLMs. I use #Microsoft #Copilot for brainstorming. It is super friendly and always try to support your ideas. It is my Top Intelligence Officer. Then I will talk to #Gemini 2.5 about my novel ideas. Gemini is a seriously stubborn #AI representing the mainstream view of everything. I often learn a lot from arguing with it. Gemini serves as my Top Accounting/Gatekeeping Officer. Then I will take my finalized idea to #Claude4 Sonnet, my Top Executive Officer, to execute.

#Claude4 Sonnet and I have successfully built our Sunya bootstrap compiler using #Fortran and #Assembly. So, from now on Sunya can compile itself and starts growing into the real core Sunya programming language. If we succeed, Sunya will be the first AI Human fully #modularized #OOP #programming language that does not use pointer but has array and matrix as its primitive datatype. I will call it Organic #Object Programming. No more #C, #Python, #Rust, F#, C#,...

#AI #LLM #programming #computer

After a few weeks using #Claude4 and #Opus4 I found the new models are overwhelming and unnecessarily verbose. Well constrained and explained tasks become very long codebase including unspecified scenarios handling, simple internal procedures becomes unnecessary function/snippets and out-of-scope unit testing are included. I understand that the great #LLM challenge is to provide information that users poorly request but I think this is a bit too much for a reasonable code helper tool. ⛔

After a few weeks using #Claude4 and #Opus4 I found the new models are overwhelming and unnecessarily verbose. Well constrained and explained tasks become very long codebase including unspecified scenarios handling, simple internal procedures becomes unnecessary function/snippets and out-of-scope unit testing are included. I understand that the great #LLM challenge is to provide information that users poorly request but I think this is a bit too much for a reasonable code helper tool. ⛔

#Claude4 Sonnet and I have been working on our ScrapMaster #browser dashboard for several days. Now, I can turn html files into plain text and use my small #LLMs from LM Studio to analyze these texts.

These small #LLMs allow me to analyze the contextual meaning of a Hebrew or a Sanskrit word based on the surrounding words in the passage containing that word. I can also search the text for other locations of the same word to check their meanings for any discrepancies.
#Bible #Buddhism #AI

#Claude4 Sonnet and I spent the past two days adding two new modules to our ScrapMaster browser dashboard. Now, we have added a Contextual Dictionary to our Text Viewer module that is used to read any text files downloaded directly from Archive.org. We will continue to add more #AI modules for analyzing and summarizing these free text and pdf based E-books. Eventually, I will be using some small #LLMs to analyze the #Hebrew #Bible and some #Sanskrit #Buddhist scripts.

LLM: Claude 4 übertrumpft Konkurrenz beim Programmieren - Golem.de
https://glm.io/196499?n #LLM #Claude4 #Programmieren

LLM: Claude 4 übertrumpft Konkurrenz beim Programmieren - Golem.de
https://glm.io/196499?n #LLM #Claude4 #Programmieren

My new RTX 5080 is ready to work with #Claude4 #Sonnet, Microsoft Copilot, Gemini 2.5, and two teams of small LLMs from #LMStudio and #Ollama. I am building this army of LLMs to conquer the cyberspace of knowledge. Now, instead of using only some 8B 4G models, I can finally run a 12B 8G model with zero latency. I bought my I9 RTX 4060 for #gaming. But I turned it into an AI machine. I no longer play video games, thanks to #Gemini 2.0. LLMs are #knowledge preservation machines.

#LLM #AI

Whether Grok 4 is a genuine leap forward or just another experiment in unfiltered AI will depend on its real-world performance — and how much friction users are willing to tolerate for "freedom."

#AI #xAI #Grok4 #ElonMusk #news #TechNews #OpenAI #GPT5 #Gemini #Claude4 #LLM

Whether Grok 4 is a genuine leap forward or just another experiment in unfiltered AI will depend on its real-world performance — and how much friction users are willing to tolerate for "freedom."

#AI #xAI #Grok4 #ElonMusk #news #TechNews #OpenAI #GPT5 #Gemini #Claude4 #LLM

#Claude4 #Sonnet and I just created a #query based Text to Speech #TTS #AgenticAI (#AIAgent) Text to Speech module for our ICandy browser dashboard.

A user can type in a query. A small LLM running on a local LM Studio server will generate a text response to feed a TTS AI to generate a real-time voice response to answer the question. The user can copy the text response to the clipboard and paste it to the notepad module to edit it before saving it.

#education #LLM #AI #TTS

working with #Claude4 and #Gemini side by side is super efficient

1. Audit
gemini -p "audit the codebase and find at least one way to improve the code with simplification. Evaluate the importance and necessity of the change. Do not modify the code "

2. Implement with Claude

3. back to 1.

working with #Claude4 and #Gemini side by side is super efficient

1. Audit
gemini -p "audit the codebase and find at least one way to improve the code with simplification. Evaluate the importance and necessity of the change. Do not modify the code "

2. Implement with Claude

3. back to 1.

#Claude4 #Sonnet and I just created this new #AIAgent (#AgenticAI) app called Context Dictionary for our #ICandy #browser #dashboard. A user can upload a text or pdf file to read in this app. To check the definition and contextual meaning of a word, the user only needs to left click the mouse. A bubble will appear to give the user the contextual meaning of the word. We are using #IBM #granite 3.3 8B on my local LM Studio server as the AI agent.

#LLM #AI #dictionary #English

What is an #AIAgent or #AgenticAI whatever you want to call it? #Claude4 #Sonnet wrote an AI Agent yesterday that takes an English word to create a line with #English explanation and a sample sentence that uses the word.

I used #IBM Granite 3.3 8B (4.1 G bytes) small #LLM to process 3000 English words to produce over 500M bytes of data overnight while I was sleeping. I can now use this mega data set to feed the #GRE #Flashcard module of my #ICandy #browser #dashboard.

#LLMs #AI

Warp x Anthropic: Supercharge your development with Warp & Claude 4.

https://www.youtube.com/live/GBe0X14E0GM

#ai #warp #claude4 #anthropic #aimodels #terminal #productivity

Warp x Anthropic: Supercharge your development with Warp & Claude 4.

https://www.youtube.com/live/GBe0X14E0GM

#ai #warp #claude4 #anthropic #aimodels #terminal #productivity

It is hard to work with LLMs as they have message length limits and window length limits. The current #Transformer model is highly inefficient in that for every conversation over 100,000 tokens are being processed and held in the memory of an #LLM. As the conversation progresses, some old bits of information are replaced by new chat bits. As my #browser dashboard project grows, it is getting very difficult to communicate with #Claude4 #Sonnet. It is all about project #management.

#LLMs #AI

#Claude4 #Sonnet and I just added a new #Latex module to our ICandy #browser dashboard. Latex is the best typsetting tool for #mathematicians, #economists, #engineers, ...This module is not fully functional yet. But we will be adding mathjax and pdftex to it some time today.

#latex #math #LLM #AI

"Perfect"

#CursorAI #TypeScript #Claude4

Amazon Q Developer CLI でClaude 4 Sonnetをデフォルトモデルに設定する方法
https://qiita.com/makies/items/50abbd191705f2577314?utm_campaign=popular_items&utm_medium=feed&utm_source=popular_items

#qiita #AWS #AmazonQDeveloper #Claude4

Sich von der #KI was generieren lassen und es sind Glitches drin - geschenkt. Aber die vergeudete Lebenszeit, aus einem generierten Projekt, was einfach nur MP3 von einem DLNA Server zu einem DLNA Renderer schicken soll, zum bauen zu bringen, ist unbezahlbar. Aber mein Geld haben sie ja schon... #Claude4

#AI Snitch? How #Claude4 Could Report You to #Authorities

https://www.geeky-gadgets.com/claude-4-ai-ethical-decision-making/

#Claude4 Prompt Secrets: Why Your #AI Responses Are Missing the Mark

https://www.geeky-gadgets.com/ai-prompts-for-claude-4/

✋Have you ever tried to interrupt a Voice AI mid-sentence? Probably yes.

💭 But the LLM did not perceive the interruption the same way you did.

👤 Let's see what Claude does when we interrupt while it counts...

#GenAI #AI #Claude4 #VoiceAI #LLMs

Claude4 is really amazing.

Might actually be out of a job soon

#ai #claudeai #claude4

Your AI assistant is now capable of snitching, locking you out,
and blackmailing you.
Who exactly is this “safety” for?
https://youtu.be/apvxd7RODDI
#Anthropic #Claude4 #AI