TokenVibe

Telegram @tokendiff

⚔ Compare ☆ Follow Own this channel?

🔄 Data last refreshed 1 day ago

🤝

Make a deal with TokenVibe

🔒 Paid proposals held safely in escrow — released only when the work's approved.

Subscribers

442

🎁 Free analysisfor TokenVibe

🕵️Fake Followers CheckReal-audience % & bot risk→ 📊Engagement rateBenchmarked vs. peers→ 💰What they chargeFair sponsored-post price→

Known for

Qwen 3.8-Max Flagship 2.4T-parameter model (95B active) for autonomous coding, long-horizon research, and native multimodal agency. 🔴Autonomous coding: 10+ day self-evolving development loops with zero human oversight. 🔴Long-horizon mastery: 500+ turn iterative chip design (81% area reduction) and 365-day business simulations. 🔴Research agentic loops: Independent paper reproduction and algorithm

77 views Qwen 3.8-Max Flagship 2.4T-parameter model (95B active) for autonomous coding, long-horizon research, and native multimodal agency. 🔴Autonomous coding: 10+ day self-evolving development loops with zero human oversight. 🔴Long-horizon mastery: 500+ turn iterative chip design (81% area reduction) and 365-day business simulations. 🔴Research agentic loops: Independent paper reproduction and algorithm

74 views LeapTalk It’s for turning a single image and audio into a realistic, talking video avatar. - A virtual presenter reading a script - A customer-service or meeting avatar etc Main advantage is speed: it generates video in real time while keeping the person’s identity and lip movements stable. Based on Wan2.1-T2V-1.3B. 👉 Project

ComfyUI NVIDIA PiD Tiled Decoding Native ComfyUI nodes for seam-free latent-tiling. Decodes 2MP latents into 33MP in 60s on RTX 5090. 🔴Fixes color collapse and chroma noise. Verified up to 107MP with linear scaling at 1MP/s. 🔴Output resolution is fixed at 4x input latent size. No lossy VAE round-trips required. 👉 GitHub

68 views ComfyUI NVIDIA PiD Tiled Decoding Native ComfyUI nodes for seam-free latent-tiling. Decodes 2MP latents into 33MP in 60s on RTX 5090. 🔴Fixes color collapse and chroma noise. Verified up to 107MP with linear scaling at 1MP/s. 🔴Output resolution is fixed at 4x input latent size. No lossy VAE round-trips required. 👉 GitHub

ComfyUI-MiniMaxH3-Easy Unified workflow nodes for MiniMax H3 text, image, and reference-conditioned video gen. 🔴Streamlined media input port supporting mixed images, videos, and audio. 🔴Integrated @ reference editor for visual selection of media indices (<Picture N>, etc.). 🔴Simplified # dialogue blocks with automatic conversion to <d>...</d> markup. 👉 GitHub

68 views ComfyUI-MiniMaxH3-Easy Unified workflow nodes for MiniMax H3 text, image, and reference-conditioned video gen. 🔴Streamlined media input port supporting mixed images, videos, and audio. 🔴Integrated @ reference editor for visual selection of media indices (<Picture N>, etc.). 🔴Simplified # dialogue blocks with automatic conversion to <d>...</d> markup. 👉 GitHub

📣 Post reach

Avg views / post

14.9%

Reach vs followers

🔥 Top post: Qwen 3.8-Max Flagship 2.4T-parameter model (95B active) for aut · 77 views

Recent posts

View on Telegram ↗

🔥 Top post Qwen 3.8-Max Flagship 2.4T-parameter model (95B active) for autonomous coding, long-horizon research, and native multimodal agency. 🔴Autonomous coding: 10+ day self-evolving development loops with zero human oversight. 🔴Long-horizon mastery: 500+ turn iterative chip design (81… 👁 77

ComfyUI MiniMax H3 Prompt Engineer Yet another prompt enhancer node. 🔴 Semantic asset addressing via @ syntax for integrated image, video, and audio streams. 🔴 T2VA, I2VA, FL2VA, L2VA, and multi-shot timelines. 🔴 Flexible backend integr… 👁 54

ComfyUI MiniMaxH3 Image Editing Custom nodes for MiniMax H3 that generate a short compatible frame sequence and extract a single frame from it. 👉 GitHub 👁 57

▶ MiniMax-H3 T2VA Prompt Rewriter LoRA 27B LoRA adapter for structured text-to-audio-video expansion. Generates detailed shot descriptions, camera motions, and synchronized soundscapes. Compatible with LightX2V for optimized local inferen… 👁 58

ComfyUI NVIDIA PiD Tiled Decoding Native ComfyUI nodes for seam-free latent-tiling. Decodes 2MP latents into 33MP in 60s on RTX 5090. 🔴Fixes color collapse and chroma noise. Verified up to 107MP with linear scaling at 1MP/s. 🔴Output reso… 👁 68

ComfyUI-MiniMaxH3-Easy Unified workflow nodes for MiniMax H3 text, image, and reference-conditioned video gen. 🔴Streamlined media input port supporting mixed images, videos, and audio. 🔴Integrated @ reference editor for visual selection … 👁 68 ComfyUI Video Tiler It’s a lifesaver for upscaling high-res AI video without crashing your GPU. It slices giant frames into tiles, processes them individually, and blends them back together seamlessly. 🔴The disk-backed workflow saves one… 👁 68

VocalRender Score-native singing voice synthesis (SVS) generating audio directly from lyrics, MIDI, and tempo without time-aligned references. 🔴2.3B parameter architecture 🔴 Interleaved lyric-note tokens for automated duration and melism… 👁 67

▶ MiniMax-H3-TAE Quickly trained 2D tine VAE for MiniMax-H3. 👉 HF 👁 68

ComfyUI Sol-Attn Blackwell Memory-efficient sparse attention and MLP chunking for video diffusion, optimized for NVIDIA Blackwell (RTX 50-series) and MiniMax H3. 🔴 Fastest reported int8 attention at 65K tokens (108.95 ms) with 3.7x bette… 👁 68

▶ Hunyuan3D-Buffalo 1.0 Tool for creating and editing 3D objects using text instructions. It’s mainly useful for game assets, animation, VR/AR, 3D printing, product design, and rapid prototyping. It works like a text-based 3D design assist… 👁 68

▶ LeapTalk It’s for turning a single image and audio into a realistic, talking video avatar. - A virtual presenter reading a script - A customer-service or meeting avatar etc Main advantage is speed: it generates video in real time while … 👁 74

📊 Channel activity & format

Posting cadence

19 / week

A high-frequency channel — multiple posts a day, an always-on feed.

Content format

Mixed media

A mix of video and images — flexible for most campaign formats. Recent: 4 video · 7 image.

💡 Facts

👁️Averages 66 views per post.

🕵️ Fake follower check

Estimated

41/100

Low Credibility score

69%

Mixed Real audience

Medium Fake-follower risk

Low Data confidence

Est. 69% real, active audience · Medium fake-follower risk.
Low engagement (~0.1× of followers engage each post) — a sign of an inflated or inactive audience.

Low confidence — few signals are public for this listing, so treat it as a rough screen. Heuristic estimate from engagement, follower ratios, account age & growth — a screening signal, not a guarantee.

About

AI dev ⬥ vibe‑coding ⬥ creativity One author ⬦ one POV ⬦ zero noise ComfyUI, models, AI stuff

📸 Gallery

🔀 Audience overlap

Estimated

Estimated shared audience with similar creators — useful for avoiding overlap (or doubling down) when planning a campaign.

tain1angel рыжий гуль 17%

saifex // nx.mida 17%

Сэдкет 🐾 14%

⊹ ֶָ࣪ Isha and Jinx' lair ֶָ࣪ ⊹ 11%

астрайган 10%

Разговоры с Внутренней Женщиной | Алëна Ярчевская 10%

More like this

Find more →

tain1angel рыжий гуль

Разговоры с Внутренней Женщиной | Алëна Ярчевская

@alena_yarchevskaya

434 Subscribers

⊹ ֶָ࣪ Isha and Jinx' lair ֶָ࣪ ⊹

@isha_and_jinx

212 Subscribers

Auto-generated from Telegram's public data — no affiliation with or endorsement by SocialDB. Is this you? Claim it · Remove