Meta, Google, NVIDIA go open source; OpenAI stays closed — strategies diverge
The gap between companies that sell hardware and companies that profit from APIs over open-source strategy has become a talking point in the community
00
Meta AI
Meta's AI organization behind Llama. In August 2026, it acknowledged the failure of Llama 4 and open-sourced its successor, "Muse." Muse Glimmer 30B is an open model with about 30 billion parameters that Meta AI released for local agent use. According to reports from August 2026, Muse Glimmer 30B was praised for successfully completing a repository code review in local-run tests. That same month, Mark Zuckerberg published a 6,500-word AI manifesto, which was said to have only deepened public distrust instead of easing it.
Current rank (1M)
7
- Change
- Prev.
- 10
Rank over the last 30 days
2026.08.18 – 2026.09.16 · High 7 · Low 10 · Now 7
Meta AI · Related stories
Unsloth releases desktop app with local training support
MLX, GGUF, diffusion, and audio in one app — connects Claude Code and Codex to local models
890
llama.cpp adds CI build target for AMD ROCm 7.14
First official release built on TheRock build system expands Linux and Windows support
170
memcode, an open-source coding agent that remembers repositories, released
MIT-licensed CLI and gateway store work history and past failures in a .memcode folder
20
Zuckerberg's 6,500-word AI manifesto only deepened public distrust
TechCrunch pushes back on Meta CEO's "personal superintelligence" essay
20
Meta AI's 'Muse Glimmer' 30B passes local repo code review test
AI influencer runs it offline, finds it understands existing code and even generates new examples
220
Meta releases 30B-parameter open model for local agents
Muse Glimmer is an agent-specialized model that runs on 24GB-class consumer GPUs with 4-bit quantization
180
DeepSeek-V4-Flash-0731 reported to stall during long-context tasks
Generation halts past 100K tokens when run locally with OpenCode; resuming with a command restores work
50
DeepSeek v4 Flash 0731 run locally with RTX 4090 and Tesla P40 combo
Adding two GPUs to 128GB of DDR4 yields 176GB total, with 2–3 tokens per second depending on quantization
70
xAI unveils image generator Imagine 2.0, ranks second behind GPT-Image-2 on Arena benchmark
New version adds editing tools and templates, Elo score trails OpenAI's model by a narrow margin
10
Mistral Releases Open-Weight Safety Classifier Shieldstral 1.0 3B
A lightweight model that turns policies into text-based questions, letting operators change content moderation standards without retraining
10
Meta unveils beta of terminal coding agent "Muse Code"
Built on new model Muse Spark 1.2, it plans, implements, and validates multi-file changes across large repositories, the company said
20
Meta AI · Recent announcements
- 611Muse: Meta's personal AI agent, features and capabilities
- 2,833As a test of our progress to advance the frontier of AI research, in June we entered the next generation of our autonomous AI research system, AIRA₃, in a live Kaggle competition run by NVIDIA to fine-tune a 30B Nemotron model. The challenge was to teach the model to reason
- 2,008Introducing Muse Voice Transcribe, the first real-time audio perception model from Meta Superintelligence Labs.
- 941Muse Spark 1.3 with max reasoning is now available on Muse Code and Meta Model API 👇
- 192메타 개인 에이전트 뮤즈, 신뢰를 설계로 판다
- 41메타 뮤즈 스파크 1.3, 코딩 벤치마크서 GPT-5.6·오퍼스5 앞질렀다
- 18Today we launched https://Muse.ai, a personal AI agent, and published a deep dive on how we built safety into its system.
- –facebook/omnilingual-gaia2-results · dataset
- –Présentation de Meta One : Un service d’abonnement offrant davantage de fonctionnalités et d’IA pour créer, se connecter et se démarquer
- –메타, 구독 서비스 Meta One 전 세계 출시
- –Introducing Muse: The World’s First Personal AI Agent Built for Everyone
- –메타, 멘로파크 인프라 랩 AI 하드웨어 공개
- –Wzmacniamy w Polsce ochronę przed oszustwami
- –Closed-Loop Cooling Explained: The Plumbing Behind Meta’s AI