找到适合你的 AI 模型。

输入文本→ 输出文本

价格待核验多模态与渠道费用须分别核对

加入对比来源 / 文档 ↗

Qwen3.8-Flash-Next-Gyro-abliterated-Huihui-GGUF

agentionai

> **Updated 2026-10-13: both files.** Same file names, sizes and tensor types, better weights from an improved encode. This is huihui-ai/Huihui-Qwen3.8-Flash-Next-abliterated encoded the same way as our Gyro files. Abliteration is huihui-ai's work: it removes the refusal direction from the model's weights. We only quantized it. **Which setup fits your hardware, NVIDIA and AMD instructions, context limits, vision and speculative decoding are all on the main Gyro card and apply unchanged here; only the file names differ.**

译文排队中 · 当前显示来源原文

输入文本 · 图像→ 输出文本

价格待核验多模态与渠道费用须分别核对

加入对比来源 / 文档 ↗

Qwen3.8-Flash-Next-Gyro-Coder-GGUF

agentionai

Qwen3.8-Flash-Next is a 125B mixture-of-experts model, the kind that normally needs a workstation full of GPUs. Gyro-Coder brings it to a single RTX 3090, 4090 or 7900 XTX: 19.9 GiB of GPU memory. **New: Gyro-Coder-L for 32 GB cards.** The same experts at about 3.2 bits per weight: 27.5 GiB of GPU memory, and closer to the full model than Gyro-Coder on every set we measure. On C++ it is closer than Gyro-M, which keeps all 512 experts. **Updated 2026-10-29: Gyro-Coder-L v1.2** (`Qwen3.8-Flash-Next-Gyro-Coder-L-TQ3_0.gguf`, same name and size). Thanks to a detailed community report, three experts that matter for tool calls and symbols are back in the expert set, in place of three that did less. Tool calls no longer break under sampling: across 100 `<parameter=` sites in real agent sessions, v1.0 wrote a broken tag at 4 and v1.2 at none. Roman numerals are fixed too, and KLD is unchanged on every set. v1.2 replaces both v1.0 and the v1.1 preview.

译文排队中 · 当前显示来源原文

输入文本→ 输出文本

价格待核验多模态与渠道费用须分别核对

加入对比来源 / 文档 ↗

Qwen3.8-Flash-Next-Gyro-GGUF

agentionai

**Qwen3.8-Flash-Next (125B MoE) on one GPU: 27.6 GiB for a 32 GB card, 35 GiB for 48 GB, 50 GiB for 64 GB.** The best quality per GiB at every size we ship. Gyro-S fits a 32 GB card: 27.64 GiB of weights, 28.8 GiB at 64k context, up to 128k context. Coding on a 24 GB card: agentionai/Qwen3.8-Flash-Next-Gyro-Coder-GGUF. Abliterated Gyro-S and Gyro-M: agentionai/Qwen3.8-Flash-Next-Gyro-abliterated-Huihui-GGUF. The routed experts are stored in a rotated basis with our own rotor code (Agention Precision Rotor, APR). The `TQ1_0` / `TQ2_0` / `TQ3_0` in the names is only a size class for the Hub (the experts use the trellis types TQK6/TQK7, TQ2_T and TQP5-7). **Upstream llama.cpp cannot load these files yet.** They run in:

译文排队中 · 当前显示来源原文

输入文本 · 图像→ 输出文本

价格待核验多模态与渠道费用须分别核对

加入对比来源 / 文档 ↗

Qwen3.8-Flash-Next-MTP-Q8_0-GGUF

agentionai

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。

输入文本→ 输出文本

价格待核验多模态与渠道费用须分别核对

加入对比来源 / 文档 ↗

Qwen3.8-Flash-Next-MTP-ROCmFP4-FAST-GGUF

agentionai

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

输入文本→ 输出文本

价格待核验多模态与渠道费用须分别核对

加入对比来源 / 文档 ↗

Qwen3.8-Flash-Next-ROCmFP4-FAST-imatrix-GGUF

agentionai

A 180 B model that runs **entirely on the GPU** of a 128 GB unified-memory box — 87.06 GiB at 4.23 bpw, within 2.5% perplexity of the unquantized model. Sized for the 96 GiB VRAM carve-out of a Strix Halo (Ryzen AI MAX+ 395 / Radeon 8060S). *Throughput against context depth, against our closest mainline-compatible quant run on both this fork and stock llama.cpp — numbers and discussion under Holding up at long context.* - **Everything stays on the GPU.** Not the experts, not the 51.2 B-parameter n-gram table. No host-RAM fallback, no CPU compute in the hot path. - **Fast prefill that holds at depth** — 423 t/s at 512 tokens, 357 t/s at 8k, still 245 t/s at 32k and 138 t/s at 128k. Prefill is compute-bound and the ROCmFP4 types are cheaper to dequantize. Matters most for long prompts, agentic loops, and anything that re-reads a large context. - **Better perplexity than that same mainline tier**, by about two standard errors, and better than AesSedai's IQ4_XS and IQ3_S at 20–30 GB smaller. - **Vision included** — the model's own tower, at f16, losslessly converted. - **Speculative decoding included** — the model's own MTP head. Up to **40 tok/s generation** observed on a Strix Halo with adaptive drafting; acceptance is content-dependent, so the rate moves with what you are generating. - **imatrix-calibrated** on 1540 chunks from two corpora, so the low-bit tensors are fitted to real activations rather than assumed.

译文排队中 · 当前显示来源原文

输入文本→ 输出文本

价格待核验多模态与渠道费用须分别核对

加入对比来源 / 文档 ↗

Signal-3.8-27B

agentionai

> [!IMPORTANT] > **Update 2026-09-13 — use these settings for stable, short reasoning** > Temperature 0.6, min-p 0.05 (top-p 0.95, top-k 20), KV cache q8_0 or f16, and the MTP draft head on. > Without min-p, long answers can run away into repetition on some draws; a 4-bit value cache makes that > deterministic. Both are fixed by the settings above, on Signal and on the base model alike. > [!WARNING] > The weights in this repository were updated on 2026-09-13 (the GGUF tiers were rebuilt from them). The first release's tiers could, rarely, > repeat an answer line when sampling without the draft head; the rebuilt tiers showed no loop or cap > in 55 traces under the same conditions. If you downloaded before that date, re-download. AgentionAI Signal is a minimally invasive fine-tune of Qwen3.8-27B designed for lower generation latency and better token efficiency. On our held-out general-prompt evaluation, Signal produces **57% fewer answer tokens** and uses **52% fewer thinking tokens**, while matching or improving the measured answer quality of the base model.

译文排队中 · 当前显示来源原文

输入文本 · 图像→ 输出文本

价格待核验多模态与渠道费用须分别核对

加入对比来源 / 文档 ↗

Signal-3.8-27B-GGUF

agentionai

> [!IMPORTANT] > **Recommended settings** > Temperature 0.6, top-p 0.95, top-k 20, min-p 0.05, **presence_penalty 1.0**, KV cache q8_0 or f16. > Keep min-p and presence_penalty on: without them, long generations can run into repetition on some > draws (a 4-bit value cache makes that deterministic). If you enable the MTP draft head on recent > mainline llama.cpp, also pass **`--no-spec-draft-backend-sampling`** — the default backend draft > sampling does not apply your sampler to accepted draft tokens, which can let repetition run away > regardless of the settings above. > [!WARNING] > The tiers in this repository were rebuilt on 2026-09-13. The first release's tiers could, rarely, > repeat an answer line when sampling without the draft head; the rebuilt tiers showed no loop or cap > in 55 traces under the same conditions. If you downloaded before that date, re-download. AgentionAI Signal is a minimally invasive fine-tune of Qwen3.8-27B designed for lower generation latency and better token efficiency. On our held-out general-prompt evaluation, Signal produces **25% fewer answer tokens** and uses **41% fewer thinking tokens**, while matching or improving the measured answer quality of the base model.

译文排队中 · 当前显示来源原文

输入文本 · 图像→ 输出文本

价格待核验多模态与渠道费用须分别核对

加入对比来源 / 文档 ↗

Signal-3.8-Flash-Next-GGUF

agentionai

AgentionAI Signal is a minimally invasive fine-tune of Qwen3.8-Flash-Next for lower latency and better token efficiency. On MATH-500 it spends **26% fewer reasoning tokens** than the base model with **half as many hesitation markers**, and scores higher (99.2% against 97.5%). As a coding agent on Terminal-Bench 2.0 it finishes the same tasks in **25% less wall time**. With speculative decoding it is faster still: the multi-token-prediction drafter agrees with Signal 90% of the time against 69% for the base model, so more tokens are accepted per forward pass. Signal gets there by being direct, not by cutting corners. Reasoning models spend a large share of their tokens on hesitation and re-checking rather than on new steps. Signal keeps every step that does work and drops the ones that only describe the process, and in answers it drops preambles, sign-offs, excess formatting and narration while keeping the substance. The 177B mixture-of-experts backbone, the vision encoder and the multi-token-prediction draft head all work exactly as in the base model, so it is a drop-in replacement. Signal is trained by self-distillation, on Qwen3.8-Flash-Next's own answers generated under an instruction to be direct that the released model no longer needs. No external data and no other model's outputs went into it, so it keeps the base model's knowledge and voice intact.

译文排队中 · 当前显示来源原文

输入文本 · 图像→ 输出文本

价格待核验多模态与渠道费用须分别核对