Generates unified embeddings for text, images, video, visual documents and interleaved multimodal inputs with configurable output dimensions and Matryoshka truncation to trade accuracy for cost. Model weights and code are released under Apache-2.0; the 9B variant scores 80.6 on MMEB-v2.
AI Model2026
GGUF-quantized, refusal-removed build of Qwen3.8-Flash-Next for llama.cpp that provides multimodal (image+text), reasoning and tool-calling capabilities; released for security research and red-teaming under the Apache-2.0 license.
AI Model2026
GGUF-quantized build of Qwen3.8-Flash-Next for image-text-to-text inference and local deployment. Ships with Unsloth Dynamic 3.0 quantization, thinking-mode controls (preserve_thinking, reasoning_effort), and native long-context support (262k, extensible to 1M with YaRN).
AI Model2026
A 770B-parameter Mixture-of-Experts instruct model from Tencent that natively supports 1,048,576-token contexts, Gated DSA attention, and speculative MTP decoding; open-sourced under Apache-2.0 with BF16 and FP8 weights for deployable inference.
AI Model2026
Drop-in abliterated (refusal-removed) build of GLM-5.3-Flash that bakes refusal-direction removal into block-FP8 safetensors, yielding an uncensored 320B (18B active) multimodal MoE model with a 1M-token context. Intended for red-teaming, interpretability, and robustness research; MIT license; not for production without added guardrails.
AI Model2026
Provides a cybersecurity-focused CRACK variant of GLM-5.3 FP8 that reduces refusals for offensive-security, red-team, exploit-development and malware-analysis queries while retaining native FP8 speed on Hopper GPUs; MIT-licensed for authorized security work.
AI Model2026
Sparse MoE causal LLM that uses Mixture-of-Value Attention (MoVA) to store 36B parameters while activating ~4B per token; supports a native 524,288-token context and is released with final checkpoints, training data, and training code under open license.
AI Model2026
NVFP4-quantized checkpoint of Qwen3.8-Flash-Next for GPU-optimized multimodal autoregressive inference — routed MoE experts in W4A4 NVFP4 while attention/ancillary layers remain BF16; ~2.7× smaller than the BF16 source and supports very long contexts.
Introduces Declarative Attention (DA), a zero-shot protocol that has LMs declare which parts of long context to attend to during chain-of-thought, letting the runtime build dynamic attention masks and skip most KV-cache reads. Produces large token savings (up to ~52% on Gemma-4-31B) with modest accuracy loss.
Shows that fully quantizing all 496 linear layers—including the recurrent Gated DeltaNet—of a hybrid 27B LLM to 4-bit NVFP4 W4A4 preserves benchmark accuracy while reducing model size to 17.5 GiB and improving prefill speed; includes a calibrated NVFP4 checkpoint.
Compresses KV cache for long-chain reasoning by keeping prompt tokens and evicting remaining entries uniformly at random per attention head; across four models and six reasoning tasks it matches the strongest prior evictor while delivering 32–43% higher vLLM throughput. Relies on prompt protection and redundancy across heads/text to retain reasoning traces; suitable when static memory budgets and higher serving throughput are priorities.
AI Model2026
Multimodal foundation model for visual understanding, spatial reasoning and multi-step agent tool use — accepts text, multiple images and video at any resolution and supports long contexts (up to 128K tokens). Emphasizes fine-grained 2D/3D relations, affordance reasoning and embodied-AI planning.