从语言到影像，从开源到 API。读懂模型能力、比较真实成本，
为你的下一个想法，找到合适的智能引擎。
GGUF quantizations of apodex/Apodex-1.1-mini, a 35.95B-parameter Qwen3.5 MoE model for research, data, files, code, and tool-driven work. All model files retain the checkpoint's native MTP/NextN head. For image input, use the included F16 vision projector. *Results reported by Apodex AI from evaluations of the original Apodex-1.1 model using FrontierAgent, its agent harness and evaluation suite.*
译文排队中 · 当前显示来源原文
GGUF quantizations of Cloudflare/clef, a 27B multimodal decision model post-trained from Qwen3.8-27B. For image or video input, use the included multimodal projector (`clef-mmproj-BF16.gguf`). Tested with `torch` 2.11 and `transformers` 5.10.2 on a single H200. Image and video inputs also need `pillow`. `systemone` takes a Jev/SystemOne `POST /v1/systemone` request body and returns the same response body: `model`, `answers` keyed by question ID, and `usage`. A `choice` answer has `choice`, `confidence`, and `probabilities`; a `score` answer has the expected `score`, `confidence`, `legend`, and `probabilities`; a `noul` answer has the probability of true. `instructions` is optional, and `images` and `videos` may be added to the request.
译文排队中 · 当前显示来源原文
基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。
基础资料已收录 · 缺失资料后续补充
基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。
基础资料已收录 · 缺失资料后续补充
基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。
基础资料已收录 · 缺失资料后续补充
基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。
基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。
基础资料已收录 · 缺失资料后续补充
基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。
基础资料已收录 · 缺失资料后续补充
基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。
基础资料已收录 · 缺失资料后续补充
基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。
基础资料已收录 · 缺失资料后续补充
基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。
基础资料已收录 · 缺失资料后续补充
基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。
基础资料已收录 · 缺失资料后续补充
Mixed-precision **GSQ-RCO** GGUF quantizations of Nanbeige/Nanbeige4.3-33B, optimized for efficient, high-fidelity local inference in `llama.cpp`. **64k Long-Context (Low-VRAM)** Using 4-bit KV cache (`-ctk q4_0 -ctv q4_0`) allows 65k context to run within **< 12 GB total VRAM** on consumer 12 GB GPUs (e.g. RTX 3060 12GB, RTX 4070):
译文排队中 · 当前显示来源原文
Mixed-precision **GSQ-RCO** GGUF quantizations of nvidia/NVIDIA-Nemotron-3.5-Bolt-E4B-BF16, optimized for efficient, high-fidelity local inference in `llama.cpp`. The model supports a context length of up to 65,536 tokens (64K). Standard quantization severely degrades hybrid sliding-attention models with shared Key-Value caches, multimodal projections, and speculative draft targets. **GSQ-RCO** solves this through two targeted techniques: - **RCO:** Preserves all 54 shared `k_proj`, `v_proj`, and `norm` tensors across layers 24–41 in high precision (`Q5_K`/`Q6_K`), maintaining structural attention geometry, visual projection alignment, and DFlash draft compatibility. - **GSQ:** Optimizes per-layer precision guided by Hessian sensitivity calibration.
译文排队中 · 当前显示来源原文
Mixed-precision **GSQ-RCO** GGUF quantizations of TokenRhythm/NeoHorse-1-9B, optimized for efficient, high-fidelity local inference in `llama.cpp`. The model supports a context length of up to 262,144 tokens (262K). Standard quantization severely degrades hybrid SSM-Transformer models (e.g. stock low-bit quantizations cause recurrent state drift and degradation in reasoning coherence). **GSQ-RCO** solves this through two targeted techniques: - **RCO:** Keeps all 48 critical SSM parameters in unquantized **BF16**. - **GSQ:** Optimizes per-layer precision guided by Hessian sensitivity calibration.
译文排队中 · 当前显示来源原文
GGUF files for Nex-N2.5-mini, a long-context agentic model for coding, tool use, computer use, and multimodal workloads. The source checkpoint supports a native context length of **262,144 tokens (256K)**. *Benchmark results reported by Nex-AGI for the original Nex-N2.5 family. These figures are not measurements of this GGUF conversion.* The projector is optional for text-only use. Use it with a current llama.cpp build that supports the model's multimodal path.
译文排队中 · 当前显示来源原文
基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。
基础资料已收录 · 缺失资料后续补充
基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。
Mixed-precision **GSQ-RCO** GGUF quantizations of ornith-ai/Ornith-1.5-9B, optimized for efficient, high-fidelity local inference in `llama.cpp`. The model supports a context length of up to 262,144 tokens (262K). Standard quantization severely degrades hybrid SSM-Transformer models (e.g. stock IQ3_M perplexity drops to ~75.2). **GSQ-RCO** solves this through two targeted techniques: - **RCO:** Keeps all 48 critical SSM parameters in unquantized **BF16**. - **GSQ:** Optimizes per-layer precision guided by Hessian sensitivity calibration.
译文排队中 · 当前显示来源原文
GGUF, Safetensors, and MLX quantized formats of Qwen/Qwen-Image-2.1-Turbo for local image generation using the official turbo base weights. **Q4_K_M** is recommended for the best balance of size and quality for GGUF. **INT8 ConvRot** and **NVFP4** offer high performance on supported GPUs, while **MLX** formats are optimized for Apple Silicon. All required companion files (GGUF transformer, text encoder, and VAE) are hosted directly in this repository.
译文排队中 · 当前显示来源原文
GGUF, Safetensors, and MLX quantized formats of Qwen/Qwen-Image-2.1 for local image generation using the original upstream base weights. **Q4_K_M** is recommended for the best balance of size and quality for GGUF. **INT8 ConvRot** and **NVFP4** offer high performance on supported GPUs, while **MLX** formats are optimized for Apple Silicon. Use the model with ComfyUI and ComfyUI-GGUF.
译文排队中 · 当前显示来源原文
> [!IMPORTANT] > **Compatibility:** These GGUF files require `llama.cpp` **b10828** or later, which includes official support for the Spark-X2.5 (`spark2_5`) architecture. Applications with a bundled runtime must use an equivalent or newer build. llama.cpp support GGUF quantizations of XHToken/Spark-X2.5-4B, a 4B general-purpose language model for reasoning, coding, tool use, and agentic workflows. Native context: **1,048,576 tokens (1M)**. Includes the upstream chat_template.jinja. Checksums: SHA256SUMS.txt.
译文排队中 · 当前显示来源原文
基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。
基础资料已收录 · 缺失资料后续补充
基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。
可以调整厂商、分类或输入输出条件；新模型正持续收录。
调整月度 Token 用量，直观比较不同模型的预估 API 开支。
使用已采集且单位可比的 USD / 百万 Token 渠道报价。报价为采集时快照；不计缓存优惠、阶梯价、推理额外输出与税费，请以来源的当前报价为准。
先区分输入、输出和缓存价格，再使用同一任务和用量估算总费用。图像、音频和视频可能按张、时长或分辨率计费，不能直接与 Token 单价比较。
不一定。必须逐一核对具体模型版本的许可证、商用限制及使用政策。自己部署还需要考虑显存、推理资源和运维费用。
榜单取决于测试任务、版本与评测方法。建议用实际业务样本比较准确性、时延、稳定性和成本，再选择模型。
