FP8-quantized checkpoint of the Qwen3.8 text-only causal LLM (2.4T params, 95B activated) for text-generation; preserves near-original performance, supports very long contexts (262k–1M), Mixture-of-Experts architecture, and is compatible with vLLM/SGLang/TokenSpeed. Thinking mode and preserve_thinking are enabled by default.
Multimodal Mixture-of-Experts text-generation model that accepts text, images, video and audio and returns text; preview open-weight release with 280B total params, 16B activated params, up to 512K token context and BF16/FP8 checkpoints under Apache-2.0.
Lightweight sparse-MoE LLM (7.9B params, ~1.3B activated per token) designed for hybrid multi-step reasoning and agentic tasks. Uses a KDA–MLA hybrid attention stack and a 128-expert sparse FFN; offered in BF16/FP8/INT4 for local and edge deployment.
Open-weight 30B-parameter Mixture-of-Experts LLM with 3B active params, NVFP4-quantized checkpoint, and speculative-decoding support for long-context (up to 1M tokens) agentic, chat, reasoning and tool-calling workloads optimized for NVIDIA GPUs.
Provides an FP8-post-trained 27B multimodal causal language model with a native vision encoder, large-context support (262,144 native, extensible to 1,000,000), controllable thinking-mode reasoning, and compatibility with common inference engines for deployment.
A 27B Qwen3.8 vision‑language causal transformer quantized to NVFP4 for lower‑memory inference. Provides 262K native context (extensible to 1M), Unsloth Dynamic V3.0 4‑bit quantization and MTP support so Qwen3.8‑class multimodal workloads can run on 24GB‑class GPUs.
Separates knowledge storage (a global Memory) from iterative reasoning operators (multiple Reasoners) to improve knowledge compression and inference efficiency; reports a 7B model matching baseline with 62.6% of training data and a 35B Intern-S2-Mobius achieving ~4x end-to-end speedup.
Provides a full GGUF quant ladder of an "abliterated" Qwen3.8-27B for local llama.cpp inference — includes every K-quant, embedded MTP speculative head, and optional vision projectors; refusal behavior was reduced at the weight level, so validate before production.
A Gated-DeltaNet-aware mixed-precision GGUF quantization of Qwen3.8-27B for efficient local inference; preserves the MTP draft head and offers an optional BF16 mmproj for images. Weights are ~11.73 GiB (3.69 bpw), sized for 16–24 GB GPUs at modest context.
Provides quantized GGUF variants of Qwen3.8-27B with an 'Aggressive' uncensoring profile and an optional HauhauCS FastMTP sidecar to accelerate MTP speculative decoding; includes a BF16 vision projector and K_P quant levels for VRAM/quality trade-offs.
Enables interactive serving of large Mixture-of-Experts (MoE) models on personal machines by adapting offload and execution to measured device bandwidth and agentic workload patterns. Key features include bandwidth-adaptive execution, semantic-aware caching of recurrent state, and an elastic GPU expert cache; supports 20+ MoE models and runs models from ~35B to 753B on consumer/workstation GPUs.
A 9B open-weight reasoning LLM that uses a self-improvement loop to auto-generate tasks, construct scaffolds, and optimize rollouts for stronger agentic coding and long-context reasoning. Single-GPU deployable, supports tool-calling and a 262,144-token context window.