Model Spend Arena2ND ED.
35B · dense · 2026-08-14

KAT-Coder V2.5 35B

Runs on ≤32 GB · Hugging Face ↗

llama.cppvLLMOllama

Not independently scored by Artificial Analysis — its base model Qwen3.6-35B-A3B is the closest proxy

Kwaipilot's agentic coder — a 35B-total Qwen3.5-MoE (the same architecture family as Ornith 35B), Apache-2.0, tagged code + agent. Trending on Hugging Face yet unscored by AA, so exactly the kind of capable open coder the index misses.

Sizes on disk

Real GGUF file sizes = weight VRAM.

QuantizationSize
Q3_K_M16.2 GB
IQ4_XS18.8 GB
Q4_019.9 GB
Q4_K_S20.6 GB
Q4_K_M21.4 GB
Q421.8 GB
Q4_122.0 GB
Q5_K_M25.0 GB
Q2_K25.7 GB
Q8_036.9 GB
Q6_K60.4 GB
BF1669.4 GB

Config tips

~20 GB at Q4 — spills past a 16 GB card, comfortable on 24 GB+ or the 32 GB tier. MoE, so only a few billion params compute per token; built for multi-turn agentic coding, so a single-shot BCB understates it. (Unmeasured — genuinely hard to benchmark. It's a vision-language model, so vLLM 0.26 won't load it (visual-tower weights missing) AND its text-only GGUF loads in Ollama but generates empty output. Needs a runtime that handles the full VL model.)