#openweight — Public Fediverse posts
Live and recent posts from across the Fediverse tagged #openweight, aggregated by home.social.
-
Moving away from expensive frontier models ( #OpenAI, #Claude, #Gemini) to build a custom #openweight AI setup. My current workflow orchestrates #kimi k2.6, #deepSeek v4, and #glm using Oh My OpenAgent as base.
Read about my setup here: https://www.richardorilla.website/seting_up_opencode.html
-
Moving away from expensive frontier models ( #OpenAI, #Claude, #Gemini) to build a custom #openweight AI setup. My current workflow orchestrates #kimi k2.6, #deepSeek v4, and #glm using Oh My OpenAgent as base.
Read about my setup here: https://www.richardorilla.website/seting_up_opencode.html
-
Moving away from expensive frontier models ( #OpenAI, #Claude, #Gemini) to build a custom #openweight AI setup. My current workflow orchestrates #kimi k2.6, #deepSeek v4, and #glm using Oh My OpenAgent as base.
Read about my setup here: https://www.richardorilla.website/seting_up_opencode.html
-
Moving away from expensive frontier models ( #OpenAI, #Claude, #Gemini) to build a custom #openweight AI setup. My current workflow orchestrates #kimi k2.6, #deepSeek v4, and #glm using Oh My OpenAgent as base.
Read about my setup here: https://www.richardorilla.website/seting_up_opencode.html
-
Moving away from expensive frontier models ( #OpenAI, #Claude, #Gemini) to build a custom #openweight AI setup. My current workflow orchestrates #kimi k2.6, #deepSeek v4, and #glm using Oh My OpenAgent as base.
Read about my setup here: https://www.richardorilla.website/seting_up_opencode.html
-
Mistral Medium 3.5 Developer Guide: API, Remote Agents & Pricing 2026
Mistral Medium 3.5 is a 128B open-weight dense model with 77.6% SWE-Bench Verified and a 256K context window, released April 29, 2026 under a modified MIT license. It ships with...
https://wowhow.cloud/blogs/mistral-medium-3-5-developer-guide-api-remote-agents-2026
-
Today I coded for 4 hours with #OpenCode using GLM 5 instead of Claude Sonnet 4.6. The project is a #Deno app written in #typescript and the agent had to refactor code and HTML layout.
I perceived no difference in capabilities. GLM5 handled edits well, made some mistakes just like Sonnet, and produced output of the same quality.
But GLM 5 is 3x smaller than Sonnet!
Private AI models are wasteful. They're too big, use too much power, and are too expensive to run. #openweight is the future.
-
#Mistral has built a $14B empire by focusing on #openweight #AI models, which prioritise data #sovereignty and #independence from American and Chinese tech giants. While Mistral’s models lag behind those of OpenAI and Anthropic in performance, its emphasis on #security and #localisation has resonated with European companies and governments. https://www.forbes.com/sites/iainmartin/2026/04/16/how-frances-mistral-built-a-14-billion-ai-empire-by-not-being-american/?eicker.news #tech #media #news
-
#Mistral has built a $14B empire by focusing on #openweight #AI models, which prioritise data #sovereignty and #independence from American and Chinese tech giants. While Mistral’s models lag behind those of OpenAI and Anthropic in performance, its emphasis on #security and #localisation has resonated with European companies and governments. https://www.forbes.com/sites/iainmartin/2026/04/16/how-frances-mistral-built-a-14-billion-ai-empire-by-not-being-american/?eicker.news #tech #media #news
-
#Mistral has built a $14B empire by focusing on #openweight #AI models, which prioritise data #sovereignty and #independence from American and Chinese tech giants. While Mistral’s models lag behind those of OpenAI and Anthropic in performance, its emphasis on #security and #localisation has resonated with European companies and governments. https://www.forbes.com/sites/iainmartin/2026/04/16/how-frances-mistral-built-a-14-billion-ai-empire-by-not-being-american/?eicker.news #tech #media #news
-
#Mistral has built a $14B empire by focusing on #openweight #AI models, which prioritise data #sovereignty and #independence from American and Chinese tech giants. While Mistral’s models lag behind those of OpenAI and Anthropic in performance, its emphasis on #security and #localisation has resonated with European companies and governments. https://www.forbes.com/sites/iainmartin/2026/04/16/how-frances-mistral-built-a-14-billion-ai-empire-by-not-being-american/?eicker.news #tech #media #news
-
#Mistral has built a $14B empire by focusing on #openweight #AI models, which prioritise data #sovereignty and #independence from American and Chinese tech giants. While Mistral’s models lag behind those of OpenAI and Anthropic in performance, its emphasis on #security and #localisation has resonated with European companies and governments. https://www.forbes.com/sites/iainmartin/2026/04/16/how-frances-mistral-built-a-14-billion-ai-empire-by-not-being-american/?eicker.news #tech #media #news
-
#Mistral has built a $14B empire by focusing on #openweight #AI models, which prioritise data #sovereignty and #independence from American and Chinese tech giants. While Mistral’s models lag behind those of OpenAI and Anthropic in performance, its emphasis on #security and #localisation has resonated with European companies and governments. https://www.forbes.com/sites/iainmartin/2026/04/16/how-frances-mistral-built-a-14-billion-ai-empire-by-not-being-american/?AIagents.at #AIagent #AI #ML #NLP #LLM #GenAI
-
#Mistral has built a $14B empire by focusing on #openweight #AI models, which prioritise data #sovereignty and #independence from American and Chinese tech giants. While Mistral’s models lag behind those of OpenAI and Anthropic in performance, its emphasis on #security and #localisation has resonated with European companies and governments. https://www.forbes.com/sites/iainmartin/2026/04/16/how-frances-mistral-built-a-14-billion-ai-empire-by-not-being-american/?AIagents.at #AIagent #AI #ML #NLP #LLM #GenAI
-
#Mistral has built a $14B empire by focusing on #openweight #AI models, which prioritise data #sovereignty and #independence from American and Chinese tech giants. While Mistral’s models lag behind those of OpenAI and Anthropic in performance, its emphasis on #security and #localisation has resonated with European companies and governments. https://www.forbes.com/sites/iainmartin/2026/04/16/how-frances-mistral-built-a-14-billion-ai-empire-by-not-being-american/?AIagents.at #AIagent #AI #ML #NLP #LLM #GenAI
-
#Mistral has built a $14B empire by focusing on #openweight #AI models, which prioritise data #sovereignty and #independence from American and Chinese tech giants. While Mistral’s models lag behind those of OpenAI and Anthropic in performance, its emphasis on #security and #localisation has resonated with European companies and governments. https://www.forbes.com/sites/iainmartin/2026/04/16/how-frances-mistral-built-a-14-billion-ai-empire-by-not-being-american/?AIagents.at #AIagent #AI #ML #NLP #LLM #GenAI
-
#Mistral has built a $14B empire by focusing on #openweight #AI models, which prioritise data #sovereignty and #independence from American and Chinese tech giants. While Mistral’s models lag behind those of OpenAI and Anthropic in performance, its emphasis on #security and #localisation has resonated with European companies and governments. https://www.forbes.com/sites/iainmartin/2026/04/16/how-frances-mistral-built-a-14-billion-ai-empire-by-not-being-american/?AIagents.at #AIagent #AI #ML #NLP #LLM #GenAI
-
#Anthropic’s #AIcodingagent, #ClaudeCode, is generating significant buzz at the #HumanX conference, overshadowing OpenAI. The conference also highlighted the challenges of #AI #changemanagement within companies and the growing concern about #China’s #dominance in #openweight #AImodels. https://www.cnbc.com/2026/04/11/vibe-check-from-ai-industry-humanx-anthropic-is-talk-of-the-town.html?eicker.news #tech #media #news
-
#Anthropic’s #AIcodingagent, #ClaudeCode, is generating significant buzz at the #HumanX conference, overshadowing OpenAI. The conference also highlighted the challenges of #AI #changemanagement within companies and the growing concern about #China’s #dominance in #openweight #AImodels. https://www.cnbc.com/2026/04/11/vibe-check-from-ai-industry-humanx-anthropic-is-talk-of-the-town.html?eicker.news #tech #media #news
-
#Anthropic’s #AIcodingagent, #ClaudeCode, is generating significant buzz at the #HumanX conference, overshadowing OpenAI. The conference also highlighted the challenges of #AI #changemanagement within companies and the growing concern about #China’s #dominance in #openweight #AImodels. https://www.cnbc.com/2026/04/11/vibe-check-from-ai-industry-humanx-anthropic-is-talk-of-the-town.html?eicker.news #tech #media #news
-
#Anthropic’s #AIcodingagent, #ClaudeCode, is generating significant buzz at the #HumanX conference, overshadowing OpenAI. The conference also highlighted the challenges of #AI #changemanagement within companies and the growing concern about #China’s #dominance in #openweight #AImodels. https://www.cnbc.com/2026/04/11/vibe-check-from-ai-industry-humanx-anthropic-is-talk-of-the-town.html?eicker.news #tech #media #news
-
#Anthropic’s #AIcodingagent, #ClaudeCode, is generating significant buzz at the #HumanX conference, overshadowing OpenAI. The conference also highlighted the challenges of #AI #changemanagement within companies and the growing concern about #China’s #dominance in #openweight #AImodels. https://www.cnbc.com/2026/04/11/vibe-check-from-ai-industry-humanx-anthropic-is-talk-of-the-town.html?eicker.news #tech #media #news
-
Kleines Update dazu: Das #llm #MiniMax #M2.7 soll bald als #OpenWeight veröffentlicht werden. Dann waren meine Sorgen diesbezüglich zum Glück umsonst.
-
When People Realize How Good The Latest Chinese Open Source Models Are (And Free), The GenAI Bubble Could Finally Pop
-
Ich bin dann mal sehr gespannt, klingt ja wirklich super: https://ethz.ch/de/news-und-veranstaltungen/eth-news/news/2025/07/ein-sprachmodell-im-dienste-der-gesellschaft.html
-
#Google releases #VaultGemma, its first #privacy-preserving #LLM
#GoogleResearch shows that #AI models can keep training data private.
This work on differential privacy has led to a new #openweight Google model called VaultGemma. The model uses differential privacy to reduce the possibility of memorization, which could change how Google builds privacy into its future AI agents. For now, though, the company's first differential privacy model is an experiment.
https://arstechnica.com/ai/2025/09/google-releases-vaultgemma-its-first-privacy-preserving-llm/ -
The #US #AIActionPlan prioritises #opensource and #openweight #AI to compete with #China’s growing influence in the field. China’s open-source models, like #DeepSeek #R1, have gained widespread adoption, including in the US. The US, once a leader in open-source AI, is now at risk of falling behind if it doesn’t prioritise #openness and #collaboration. https://venturebeat.com/ai/why-open-source-ai-became-an-american-national-priority/?eicker.news #tech #media #news
-
@katzenberger @silentexception
So, from that perspective, it is important that:
1. We find the appropriate job/task for which the #LLM is indispensible, and not use it for everything
2. We try to reuse the LLM as much as possible rather than training again and again -
Dieser Atlantic-Artikel zeigt sehr eindrücklich die fragile Basis des #AI Booms.
Aus #cybersecurity Sicht stelle ich mir zwei Fragen:
1. Wollen wir trotz der #digitalsovereignty Debatte die Zuspitzung auf die Hyperscaler weiter forcieren - oder gestalten wir aktiv Vendor-Diversifikation?
2. Wie resilient ist mein AI-Use-Case wenn LLM-Kosten signifikant steigen? #OpenWeight und #Selfhosting sind keine Nischenlösungen, sondern sinnvolle Optionen im #TPRM.Hier der Artikel von Matteo Wong und Charlie Warzel, #TheAtlantic
https://www.theatlantic.com/technology/2026/03/ai-boom-polycrisis/686559/ -
Les entreprise de la tech se foutent de vous la
L' #OpenWeight c'est juste du #feeeware déguisé façon #LLM 🤷♀️
C pas #OpenSource c'est réutilisable sans payer de licence c'est tout
-
#Ai2 has released #MolmoWeb, an #openweight #visualwebagent that operates from browser screenshots and executes actions like clicking, typing, and scrolling. It comes with #MolmoWebMix, a dataset of 30,000 human task trajectories and 2.2 million screenshot question-answer pairs, making it the largest publicly released collection of human web-task execution. https://venturebeat.com/data/ai2-releases-molmoweb-an-open-weight-visual-web-agent-with-30k-human-task?eicker.news #tech #media #news
-
Das chinesische #llm #MiniMax M2.7 übertrifft seinen Vorgänger M2.5 bei weitem und ist im Bereich von Athropics #Opus angekommen.
Leider scheint sich nun auch die Veröffentlichungsstrategie von #OpenWeight zu propritär zu wandeln.
Ich bin gespannt wie sich die chinesische Konkurrenz #Kimi, #GLM und #Qwen in Zukunft verhält.
Ich halte OpenWeight für unverzichtbar - der Markt vielleicht schon. -
An open-weight, 380M param, noninvasive thought to text model, via reconstructed, denoised, and upsampled EEG data.
-
Io faccio una mia stima (che in realtà qualcuno di voi l’avrà già dedotta da tempo leggendo i miei toot).
Ci sarà un massiccio abbandono della (vera) comunità vecchia dei modelli di #AI americani a favore di modelli #openweight.
Vi ricordo che c’è poco hype per DeepSeek4… troppo presto? Forse… vedremo…
-
Qwen3.5 (#OpenWeight) was released yesterday!
https://github.com/QwenLM/Qwen3.5
Authors claim better than GPT-5.2 "average ranking" (and close to Opus-4.5-Thinking):
https://qwen.ai/blog?id=qwen3.5
The first release is too large to run on most GPUs (397B total parameters and 17B active parameters, good for 8xH200), but "More sizes are coming".
These models are improving at an impressive rate. That's were AI is improving the fastest (more than from hardware improvements).
-
#SiliconValley #AI startups are seeing record valuations, but many are building on cheap, free-to-download AI models from #China
“These models were not that far behind the frontier. In fact, they were surprisingly close to the frontier. The ones that are coming now, well they’re palpably close to the frontier.”
Models like #DeepSeek’s #R1 and #Alibaba’s #Qwen, are free to use & considered “#opensource” or “#openweight” because anyone can download, copy, modify & use them.
https://www.nbcnews.com/tech/innovation/silicon-valley-building-free-chinese-ai-rcna242430 -
📬 Multi-Turn-Jailbreaks: Tod durch tausend Prompts bei Open-Weight-LLMs
#Jailbreaks #KünstlicheIntelligenz #AIThreats #Cisco #CyberSec #Jailbreak #KI #LLM #OpenWeight #PromptInjection #Sicherheit https://sc.tarnkappe.info/bb3c5c -
#Ai2 has released #MolmoWeb, an #openweight #visualwebagent that operates from browser screenshots and executes actions like clicking, typing, and scrolling. It comes with #MolmoWebMix, a dataset of 30,000 human task trajectories and 2.2 million screenshot question-answer pairs, making it the largest publicly released collection of human web-task execution. https://venturebeat.com/data/ai2-releases-molmoweb-an-open-weight-visual-web-agent-with-30k-human-task?eicker.news #tech #media #news
-
#Ai2 has released #MolmoWeb, an #openweight #visualwebagent that operates from browser screenshots and executes actions like clicking, typing, and scrolling. It comes with #MolmoWebMix, a dataset of 30,000 human task trajectories and 2.2 million screenshot question-answer pairs, making it the largest publicly released collection of human web-task execution. https://venturebeat.com/data/ai2-releases-molmoweb-an-open-weight-visual-web-agent-with-30k-human-task?eicker.news #tech #media #news
-
#Ai2 has released #MolmoWeb, an #openweight #visualwebagent that operates from browser screenshots and executes actions like clicking, typing, and scrolling. It comes with #MolmoWebMix, a dataset of 30,000 human task trajectories and 2.2 million screenshot question-answer pairs, making it the largest publicly released collection of human web-task execution. https://venturebeat.com/data/ai2-releases-molmoweb-an-open-weight-visual-web-agent-with-30k-human-task?eicker.news #tech #media #news
-
#Ai2 has released #MolmoWeb, an #openweight #visualwebagent that operates from browser screenshots and executes actions like clicking, typing, and scrolling. It comes with #MolmoWebMix, a dataset of 30,000 human task trajectories and 2.2 million screenshot question-answer pairs, making it the largest publicly released collection of human web-task execution. https://venturebeat.com/data/ai2-releases-molmoweb-an-open-weight-visual-web-agent-with-30k-human-task?eicker.news #tech #media #news
-
OpenAI and AWS Forge Landmark Partnership to Bring New gpt-oss AI Models to the Cloud
#AI #OpenAI #AWS #gptoss #CloudAI #OpenSourceAI #OpenWeight #Amazon
-
#OpenAI Just Released Its First #Open-Weight #Models Since #GPT-2 www.wired.com/story/openai...
OpenAI Just Released Its First... -
#OpenAI Just Released Its First #Open-Weight #Models Since #GPT-2 www.wired.com/story/openai...
OpenAI Just Released Its First... -
#OpenAI releases #GPTOSS, a #free #openweight model available in 120-billion and 20-billion parameter versions. The model, designed to run on #laptops, can perform #reasoning tasks, #browse the web, #writecode, and #operateagents. https://www.theverge.com/openai/718785/openai-gpt-oss-open-model-release?eicker.news #tech #media #news
-
Fal just dropped Flux 2 Turbo – an open‑weight model that’s 10× cheaper and 6× more efficient for AI image synthesis. Powered by Black Forest Labs, it runs on Nano Banana Pro and pairs with GPT‑Image 1.5. Curious how this changes the open‑source scene? Read the full breakdown! #Fal #Flux2Turbo #OpenWeight #AIImageSynthesis
🔗 https://aidailypost.com/news/fal-releases-open-weight-flux-2-turbo-10-cheaper-6-more-efficient
-
"There are a number of diffusion models out there, but I have tended to use Midjourney, which has been around longer than many other AI tools. Using Midjourney allows us to see how diffusion models have developed over time, as you can see with the simple prompt “otter on a plane using wifi” (for every image and video in this post, I pick the best out of the first four images generated). We go from melted fur at the start of 2022 to a visible otter (with too many fingers and a weird keyboard) at the end of that year. In 2023, we get a photorealistic otter, but still a weird keyboard and plane windows. In 2024, the lighting and positioning become better, and by 2025 we have excellent photorealism.
But what makes diffusion models interesting is not their increasing ability to make photorealistic images, but rather the fact that they can create images in various styles. This cuts to the heart of why AI image generation is so controversial, as many AI models are trained on images from throughout the web, including copyrighted work, and can thus replicate images in the style of living artists without their permission or compensation."
#AU #GenerativeAI #GeneratedImages #DiffusionModels #OpenWeight
-
Meta Unveils New Llama 4 AI Models With Massive Context Windows up to 10 Million Tokens
#AI #GenAI #Llama4 #MetaAI #AIModels #MultimodalAI #OpenWeight #LLMs #Llama4Scout #Llama4Maverick #Llama4Behemoth #AIbenchmarks
-
OpenAI's New gpt-oss-20b AI Model Runs On-Device with Snapdragon, But There's a Catch
#AI #OpenAI #gptoss20b #Qualcomm #Snapdragon #OnDeviceAI #OpenSourceAI #OpenWeight
