Skip to content
LFM2.5-8B and Qwen3.6 trending on Hugging Face, Replicate film tool, Marqly 5.0 (three model drops plus daily tools) | Daily AI roundup cover

LFM2.5-8B and Qwen3.6 trending on Hugging Face, Replicate film tool, Marqly 5.0 (three model drops plus daily tools)

By Harsh Desai
Share

TL;DR

Hugging Face saw three new models trend today while Replicate added a film photo converter and Product Hunt highlighted an updated bookmark manager, giving builders immediate access to text, speech, and workflow tools.

What shipped

On 31 May three models appeared on the Hugging Face trending list and one new image tool launched on Replicate. A bookmark manager update also surfaced on Product Hunt. These releases focus on ready-to-run text, speech, and simple productivity features.

Hugging Face trending

LiquidAI, Nvidia and OpenMOSS-Team each placed a model on the Hugging Face trending list. The releases cover compact text generation, an optimized larger text model, and a speech option. All three can be downloaded or run directly on the platform for quick tests.

  • LFM2.5-8B-A1B-GGUF LiquidAI released LFM2.5-8B-A1B-GGUF, a text-generation model now trending on Hugging Face. It supports direct fine-tuning and inference on the Hub. Vibe builders can test compact local text tasks without extra infrastructure.
  • Qwen3.6-35B-A3B-NVFP4 Nvidia released Qwen3.6-35B-A3B-NVFP4, a text-generation model built with Model Optimizer and trending on Hugging Face. It allows immediate fine-tuning through the platform. Developers can compare it against other large text models for production use.
  • MOSS-TTS-v1.5 OpenMOSS-Team released MOSS-TTS-v1.5, a text-to-speech model trending on Hugging Face. It supports download and inference for voice output projects. Creators gain a fast way to add natural speech to apps.

Replicate new models

ai-vintage-film-photo: Colin Hughes launched ai-vintage-film-photo on Replicate to turn digital photos into 35mm film looks such as Kodak Portra. Photographers apply organic grain effects through the API. This speeds up aesthetic edits for social posts.

Product Hunt picks

Marqly 5.0: Marqly released version 5.0, an AI-powered bookmark manager featured on Product Hunt. Users organize and search links with AI help. This gives non-technical teams a simple daily workflow upgrade.

What this means for you

For Vibe Builders: You can now pull the trending LFM2.5-8B and MOSS-TTS models straight from Hugging Face to add text or voice features to side projects. The Replicate film tool and Marqly 5.0 give quick wins for photo edits and link management without code changes. Test one model and one workflow tool this week to see what ships fastest.

For Non-techies: For daily work the new Hugging Face models and Marqly update mean you can try voice output or better bookmark search right away. The Replicate photo converter handles film looks for social posts without extra software. Start with Marqly 5.0 to clean up research links before testing a speech model.

For Developers: The Nvidia Qwen3.6 model and LiquidAI release on Hugging Face let you benchmark optimized text models against current stacks. Replicate adds a narrow image task while Marqly shows AI in simple productivity apps. Run the Qwen variant through your existing inference setup this week to measure speed gains.

What to watch next

Watch for follow-up benchmarks on the Qwen3.6 and LFM2.5 models later this week. Check Replicate for any new film or photo variants. Track Marqly updates for added AI search features.

Harshs take

The day shows more models landing on existing hubs than fresh platforms. This pattern favors builders who already use Hugging Face over those waiting for new runtimes. The practical edge goes to teams that test one trending model against their current stack before the next wave arrives.

A contrarian view is that most of these releases add incremental options rather than clear performance jumps. The real signal sits in how quickly each model moves from trending list to production use.

Run the LFM2.5-8B model on a small internal task today and compare output quality against your last deployed text model.

by Harsh Desai

More AI news

Everything AI. One email.
Every Monday.

New tools. Model launches. Plugins. Repos. Tactics. The moves the sharpest builders are making right now, before everyone else.

No spam. Unsubscribe anytime.