The Labs · O.S. Recommendation

Open-Source Model Registry

A filterable, JSON-driven registry of open-source models triaged for commercial specialty insurance. Each entry carries a Lab Rationale — why we track it, and how to deploy it responsibly.

Lab Rationale

EvoLLM-JP-v1-10B

Complex policy reasoning

Open source

Why we track it. Created via evolutionary merging of math and generalist models, it inherently possesses strong logical deduction skills. Fits comfortably on a single GPU for on-device, agentic orchestration of complex, multi-layered commercial insurance claims.

Lab notes. The merged architecture provides a unique blend of mathematical precision and linguistic fluency—ideal for reserve calculations.

Open-Source Model Registry

EvoLLM-JP-v1-10B

Llama-3-Karamaru-v1

Kokoro-82M

F5-TTS

SenseVoiceSmall

Whisper Large V3 Turbo

Moshiko Audio LLM

Fish Speech 1.5

Parler-TTS Mini

Kame (8B)

DeepSeek-OCR-2

GOT-OCR 2.0

Florence-2-large

Llama-3.2-11B-Vision

Pixtral-12B

Qwen2.5-VL-7B

PaliGemma 2 3B

Llama-3-EvoVLM-JP-v2

Bonsai-27B (1-Bit)

DeepSeek-R1-Distill-14B

Phi-4 (14B)

Qwen2.5-Coder-7B

Llama-3.2-3B-Instruct

Gemma 2 9B

Hermes 3 (8B)

Nemotron-Mini-4B

GLM-4-9B-Chat

SuperNova-Medius (14B)

BGE-M3

Qwen2.5-Math-7B

CTM-ImageNet (0.2B)

CTM-Maze-Large

Kame (8B)

Llama-3-EvoVLM-JP-v2

LADDER (TTRL Harness)

Duck-Harness ARC

SparseLM-2B

EvoLLM-JP-v1-10B