π€ HuggingFace New Models
This week’s HuggingFace releases showcase growing diversity in specialized AI applications:
- LLM-OS-Models/gemma-4-E2B-Terminal-SFT-Native-Liquid β Fine-tuned Gemma 4 models optimized for terminal/CLI text generation. Available in 1-epoch and 2-epoch variants, both instruction-tuned and base. Downloads ranging from 35β46.
- mradermacher/WebWorld-32B-i1-GGUF β A 32B parameter web-agent model in GGUF format, tagged for world-model and web-agent use cases.
- amethyst9/1194928 β A US-region hosted model entry.
- NamanSoni78/chota-model, sundaycoiL/contact-book, animemakerai/loras-may1126, maximso/ct-1 β Various community-uploaded models with minimal download activity.
Trend: Lightweight, task-specific fine-tunes (especially terminal/CLI tools) are gaining traction as local AI deployment matures.
π° OpenRouter Free & Low-Cost Models
Free Tier:
- inclusionAI: Ring-2.6-1T β 262K context, absolutely free. The standout free offer this week.
Limited Free Tier (expiring ~May 14β15):
- Z.ai: GLM 4.6 β 204.8K context, $0.00000039/$0.0000019 per token, expires in 2 days.
- MoonshotAI: Kimi K2 0905 β 262K context, $0.0000004/$0.000002 per token, expires in 2 days.
- xAI: Grok 4.1 Fast β 2M context (!), $0.0000002/$0.0000005 per token, expires in 3 days.
- xAI: Grok 4 Fast β 2M context, same pricing as Grok 4.1 Fast.
- xAI: Grok Code Fast 1 β 256K context, $0.0000002/$0.0000015 per token.
- xAI: Grok 4 β 256K context, $0.000003/$0.000015 per token.
- xAI: Grok 3 Mini β 131K context, $0.0000003/$0.0000005 per token.
- xAI: Grok 3 β 131K context, $0.000003/$0.000015 per token.
- xAI: Grok 3 Mini Beta / Grok 3 Beta β Same pricing as Grok 3 variants.
Trend: xAI dominates the limited free tier with 8 models. The emergence of 2M context windows (Grok 4 Fast series) signals a new benchmark war.
π¦ GitHub Trending Repos
| Repo | Description | Stars | Language |
|---|---|---|---|
| strukto-ai/mirage | Unified Virtual Filesystem For AI Agents | 1,805 β | TypeScript |
| yaojingang/yao-open-prompts | Chinese AI prompt library β work, study, content, marketing, life | 1,580 β | Python |
| lightseekorg/tokenspeed | Speed-of-light LLM inference engine | 922 β | Python |
| huangserva/3DCellForge | AI-powered interactive 3D cell generation studio | 868 β | JavaScript |
| alchaincyf/huashu-md-html | md/html bidirectional pipeline β markdown βη²ΎηΎHTML | 366 β | CSS |
Trend: AI agent infrastructure (virtual filesystems, inference engines) and domain-specific tools (3D, Chinese prompts) are hot categories this week.
π Summary
- 10 HF models discovered
- 1 free OpenRouter model (Ring-2.6-1T)
- 10 limited free tier models (mostly xAI Grok series)
- 5 GitHub trending repos
- 0 new research papers
Key Theme: The AI landscape this week is defined by ultra-low-cost inference (sub-$1/M tokens) and massive context windows (up to 2M tokens), making long-document AI applications more accessible than ever.