AI/ML●●Solid
Imagedojo.ai – Blind arena for Google, OpenAI, and xAI image generators
LMSYS Arena for images, but the leaderboard lacks volume—359 images doesn't drive statistical confidence.
Crowd PleaserEye Candy
vtail
153mo ago
Back to browse
GitHub Repository
7 starsTypeScript
A/B test your own VLMs for document parsing (Self-hosted Arena)
by matthew624·Feb 19, 2026·1 point·0 comments
AI Analysis
●●SolidSolve My ProblemSlickNiche Gem
Document parsing A/B test arena with ELO ranking—niche but real alternative to OCR Arena.
Strengths
- •Fair blind battle design with weighted matchmaking ensures underrepresented models get tested equally.
- •Real-time token streaming via SSE + Markdown/LaTeX rendering makes result comparison immediate and readable.
- •Docker one-click deploy + multi-provider support (Anthropic, OpenAI, Ollama, etc.) lowers friction significantly.
Weaknesses
- •Solves a real pain point but audience is narrow: teams with self-hosted VLMs and private document sensitivity.
- •90% Claude Code attribution raises questions about custom architecture vs. scaffolded boilerplate.
Category
Target Audience
ML engineers evaluating custom document parsing models, teams comparing VLMs privately
Similar To
OCR Arena · Hugging Face Model Arena
Similar Projects
AI/ML●●●Banger
ParseBench – Document parsing benchmark for AI agents
First benchmark measuring semantic correctness over text similarity for document parsing.
Big BrainDark Horse
pierre
952mo ago
AI/ML●●Solid
Local Document Parsing for Agents
LlamaIndex open-sources their parser core, but LlamaParse cloud still handles complex layouts.
Solve My ProblemShip It
cheesyFish
2012mo ago
Developer Tools●●Solid
Bbt – Black Box Testing Directly from Your Documentation
Tests live in README as plain English; clever partial parsing eliminates Gherkin boilerplate overhead.
Big BrainNiche Gem
LionelDraghi
303mo ago
AI/ML●●Solid
Extract (YC P25) – Fast, accurate document parsing
Per-span confidence scores let you review uncertain OCR before trusting 200k-page runs.
Solve My ProblemSlick
soamikapadia
3010d ago
AI/ML●Mid
I blind-tested 14 LLMs on a WP plugin task. Surprising Findings
Rigorous benchmark methodology, but it's research not a tool you can use.
Niche Gem
guilamu
321mo ago