Infrastructure●●●Banger
Ranvier – Prefix-aware routing for LLM inference
Routes LLM requests to GPUs with cached KV prefixes, skipping redundant prefill computation.
WizardryBig Brain
mindsaspire
103mo ago
Back to browse
GitHub Repository
Tier — tier-based tool routing for AI agents of any size
13 starsPython
Tier – Adaptive tool routing that makes small LLMs 10pt more accurate
by pranabsarkar·Apr 15, 2026·4 points·3 comments
AI Analysis
●●●BangerBig BrainNiche Gem
Makes 1.5B models 10% more accurate by hiding 90% of tool descriptions.
Strengths
- •Decomposes tool selection probability into family versus specific tool selection math.
- •Cuts token usage by 97% for tiny models without sacrificing accuracy.
- •Benchmarked on Ollama with 1,000+ calls across four different model sizes.
Weaknesses
- •Requires manual tool family tagging which adds setup overhead for developers.
- •Benefits diminish significantly once you move past 35B parameter models.
Category
Target Audience
Developers building local AI agents or running models on edge devices
Similar To
LangChain · LlamaIndex · LiteLLM
Similar Projects
AI/ML●●Solid
Lumisift – improves data retention in RAG from ~40% to 87%
Information density scoring beats semantic similarity for scientific RAG retrieval.
Big BrainNiche Gem
benmora
102mo ago
AI/ML●●Solid
Kronaxis Router – Don't pay frontier prices when a local LLM is enough
LLM cost routing with LoRA awareness when LiteLLM already handles basic proxying.
Big BrainSolve My Problem
JasonDuke
202mo ago
AI/ML○Pass
Boosted LightFace – A Hybrid DNN and GBM Model for Facial Recognition
Facial recognition ensemble paper, not a shipped product or reproducible codebase.
serengil
203mo ago
AI/ML●Mid
100% LLM accuracy–no fine-tuning, JSON only
Ancient Rome Q&A benchmark shows 81pp accuracy lift, but lacks adversarial defense evidence.
Big Brain
MysticBirdie
223mo ago
AI/ML●Mid
I reduced LLM inference GPU calls by 94% using semantic routing
94% GPU reduction claim needs verifiable benchmarks to stand out.
Bold BetShip It
kanacki
2114d ago