Tag

ZONOS2

Multilingual, low-latency text-to-speech model for speech generation and zero-shot voice cloning. Uses an MoE backbone with ECAPA-TDNN speaker embeddings, supports audio prefixes, fine-grained prosody/emotion controls and 44.1kHz output; optimized for Linux + NVIDIA GPUs.

Qwopus-3.6-27B-Coder

A quantized 27B coder LLM fine-tuned for repository-level code generation, multi-turn tool calling, and agentic workflows — packaged for local GGUF/llama.cpp deployment with MTP speculative decoding and trace-inversion SFT. Optimized for developer tooling; experimental and not fully safety-validated.

MiniMax Sparse Attention

Implements a blockwise sparse attention (MiniMax Sparse Attention) that scores and Top-k selects key-value blocks per Grouped Query Attention group to enable attention over million-token contexts. Paired with an exp-free Top-k GPU kernel and KV-outer sparse execution, it reduces per-token attention compute and yields large prefill/decoding speedups.

Rio 3.5 Open 397B

A post-trained Mixture-of-Experts multimodal LLM with ~397B total (≈17B active) and a 1,010,000-token context for image-text-to-text and conversational tasks. Integrates SwiReasoning to switch between latent and explicit reasoning; MIT-licensed and optimized for Portuguese/English research and on-prem inference.

moonshotai/Kimi-K2.7-Code

An agentic multimodal coding model for long-horizon software tasks: MoE architecture (1T params, 32B activated), 256K context, image/video input, native int4 quantization and preserved chain-of-thought (thinking) mode. Tuned for multi-step coding workflows and vLLM/SGLang deployment.

unsloth/MiniMax-M3-GGUF

Provides experimental GGUF-format quantized weights for MiniMax-M3 to run local multimodal (image‑text‑video) inference via llama.cpp or Unsloth Studio. The model is very large (~428B params) and requires GPU offload or large CPU RAM; llama.cpp currently falls back from sparse to dense attention.

Kimi-K2.7-Code-GGUF

Provides a locally runnable GGUF quantized build of Kimi K2.7 Code for multimodal, coding-focused agentic workflows — a 1T-parameter MoE model with 256K context, native int4 support, preserved thinking-mode, and image/video input support.

Kimi K3

Provides an open-weight native multimodal agent that understands text and images within a 1,048,576-token context window for long-horizon coding, visual reasoning, and tool-driven workflows. Uses a 2.8T-parameter Mixture-of-Experts architecture (KDA + AttnRes) with MXFP4 quantization; best suited for research and large-scale inference setups.

Kairos: A Native World Model Stack for Physical AI

Learns, maintains, and runs unified world models for Physical AI using a cross-embodiment pretraining curriculum and a hybrid linear temporal-attention architecture. Emphasizes long-horizon state persistence, theoretical bounds on error accumulation, and deployment-aware low-latency inference for real-world embodied agents.

Qwable 27B

Fine-tuned full checkpoint of the Qwen3.6-27B base that produces structured, trace-style assistant outputs for code, technical reasoning, and instruction-following. Packaged for local GGUF conversion and local inference; not a LoRA adapter and not validated for production use.

GLM-5.2 (Unsloth GGUF)

Provides GGUF-quantized GLM-5.2 builds for local text-generation with a solid 1M-token context, dynamic 1-/2-bit quant options, and Unsloth runtime integrations — targeted at long-horizon coding, reasoning and agent workflows. MIT licensed.

TurboServe: Serving Streaming Video Generation Efficiently and Economically

Serves interactive, long-lived streaming video-generation sessions by jointly scheduling session placement and GPU autoscaling to meet tight per-chunk latency. Combines migration-aware placement, load-driven autoscaling, coalesced chunk processing, GPU–CPU offloading and NCCL GPU–GPU migration; reports ~37% reductions in worst-case per-chunk latency and GPU operating cost.