a11oy · Multi-LLM Ensemble Agent
14 voters
Λ-aggregator
Doctrine v11
Fan-out prompt to selected LLM voters → score across Yuyay-13 axes → Λ-aggregate → winner. Sovereign-default:
khipu-gguf
always participates. Cloud voters activate only when their env var is present.
Voters
14 total
khipu-gguf
SZL CPU inference lab (llama.cpp GGUF)
qwen-local
local-vLLM
hf-inference-llama
HuggingFace Inference API (Meta-Llama)
hf-inference-mistral
HuggingFace Inference API (Mistral)
hf-inference-qwen
HuggingFace Inference API (Qwen)
deepseek-r1
DeepSeek API / HuggingFace Inference
kimi-k2
Moonshot AI / OpenRouter
glm-4.6
Zhipu AI (z.ai) / OpenRouter
hermes-4-405b
Nous Research / OpenRouter / HuggingFace
minimax-m2
MiniMax / OpenRouter
step-3.7-flash
StepFun / OpenRouter
nomos-1
Nous Research / HuggingFace
nemotron-super-49b
NVIDIA NIM
xiaomi-mimo
Nous Research / Xiaomi direct
Prompt
Prompt
Hello — which model are you and what can you do?
System prompt (optional)
Ask the ensemble
Λ-aggregator result
Winner · Λ score
—
Voter responses
Full JSON response + Khipu receipt
Doctrine v11 LOCKED 749/14/163 · Λ = Conjecture 1 (NOT a theorem; 163 sorries) · khipu-gguf sovereign-default · cloud voters require env var · Co-Authored-By: Perplexity Computer Agent
⚖
Compliance / GRC
🌐
Spaces