Gemma 4 26B-A4B, post-trained with GRPO against a reward model learned from 1.2 million double-blind votes cast by HiWaifu users inside their own role-play conversations. Put back into the same arena, blind, it met GLM-5.1 in 1,430 battles and won 49.6% of the decided votes; against a 13-model field including Gemini, DeepSeek-v4 and Qwen's character models it won 54.7%. Most open role-play models are tuned on preferences that come from an LLM judge, from a handful of annotators, or from synthetic pairs. We had something rarer: a live arena where, inside ordinary chats on our platform, a user is occasionally shown two candidate replies and asked which one they want to continue with. Those…
Open weights
gemma
25.8B parameters
262,144 tokens
transformers
This checkpoint is an AutoRound model-free MXFP8 RTN quantization of exported in llmcompressor / compressed-tensors format. Routed experts and the self-attention projections present in the source are stored as F8E4M3; sensitive/shared and multimodal weights remain BF16. Static FP8 KV scales were calibrated with AutoRound using the text dataset NeelNanda/pile-10k; the vision tower was not quantized. On the paired repository lmeval protocol, the four primary metrics were non-decreasing relative to the BF16 baseline using the same vLLM FP8 KV cache. The AQA gate was GO. This is a result for those tasks and settings only, not a claim of lossless quantization or general quality improvement. The…
Open weights
apache-2.0
25.8B parameters
262,144 tokens
transformers
This repository contains an MXFP8-weight checkpoint derived from exported in compressed-tensors format. The checkpoint retains the source multimodal components, but the evaluation reported here covers text tasks only. The exported checkpoint contains 11,635 F8E4M3 weight tensors, 838 BF16 weight tensors, 11,635 U8 block-scale tensors, and 171 FP32 scale/metadata tensors. Routed-expert weights and source-present text self-attention projections are MXFP8; the router, shared MLP, vision tower, and embeddings remain BF16. Measured on 2026-09-30 with lm-eval 0.4.13 and vLLM 0.29.0. The tested settings were TRITONATTN, tensor parallelism 2, pipeline parallelism 1, batch size 64, maxnumseqs=64…
Open weights
apache-2.0
25.8B parameters
262,144 tokens
transformers
Darwin-27B-RSI is Darwin-27B-Opus after Recursive Self-Improvement (RSI): the model was improved using only signal it produced itself. During self-improvement, the model itself (its weights) improves by learning only from its own solutions. No human-written solutions or reasoning traces are used; correctness is checked automatically (agreement across its own samples and code execution). Under an identical evaluation protocol, Darwin-27B-RSI improves over its parent on graduate-level science reasoning — +5.24 points on GPQA Diamond (single sample) and +3.79 points with majority voting — with every gain statistically significant in paired tests. As the reasoning engine of Darwin-27B-JEV on…
Open weights
apache-2.0
26.9B parameters
262,144 tokens
transformers
Prism ML's ternary Ternary-Bonsai-2-27B build of Qwen/Qwen3.8-27B, repacked for chad, a Claude-Code-style local coding agent for Apple Silicon, with its speculative decoder bundled in. This is chad's default model. Created using Bonsai by Prism ML. with, already quantized. Nothing is built on first run. Every projection of Qwen3.8-27B (a dense qwen35 hybrid: 64 layers, 48 GatedDeltaNet + 16 full attention) is stored in a Hadamard-rotated basis: multiplied by a fixed sign vector and put through a blockwise Walsh-Hadamard transform offline, then quantized to 2-bit affine group-128 whose three levels reproduce the ternary set {−s, 0, +s}. The rotation costs no extra bits and no extra weight…
Open weights
apache-2.0
26.9B parameters
262,144 tokens
mlx
Benefits high quality CPU inference TQ2 on Llama.cpp and Ollama via QAT - Robotcs, Routing, Coding, Multimedia, Advanced tool calling via JiRackDeltaNetTokenizer - JiRack DeltaNet understand video and images that best for Robotics also A fast and efficient 27B model optimized for CPU inference. Built on a Qwen3.8-style DeltaNet architecture (hybrid attention + SSM), with an updated tokenizer that includes Routing, Media, Vision, Sound, Tool call, and Robotics tags. Ready-to-run GGUF quantizations, and native Ollama support with reasoning disabled by default for fast, direct responses. - JiRack is a cloud-ready model that helps save money on cloud infrastructure. It can be used as an expert…
Open weights
mit
27.3B parameters
262,144 tokens
