# poetiq-ai/poetiq-arc-agi-solver

This repository allows reproduction of Poetiq's record-breaking submission to the ARC-AGI-1 and ARC-AGI-2 benchmarks.

Repository: https://github.com/poetiq-ai/poetiq-arc-agi-solver
Canonical: https://ross.abutalabs.com/products/poetiq-arc-agi-solver
Homepage: https://poetiq.ai/posts/arcagi_announcement/
Language: Python
License: MIT
License Family: permissive
Last push: 2025-12-16T17:23:38+00:00

## Health v2 (maintenance only)
Score: 42/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 57, release rhythm 35, longevity 22
- inputs: {"age_days": 308, "days_push": 260, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1282, forks 214 (observed 2026-08-28T04:04:14.057184+00:00)

## What it is
A Python research codebase that reproduces Poetiq's record-breaking, leaderboard-topping submissions to the ARC-AGI-1 and ARC-AGI-2 abstract reasoning benchmarks. It runs configurable reasoning setups (e.g., the Poetiq 3 config) on top of frontier LLM APIs such as Gemini and OpenAI to solve ARC puzzle tasks and measure cost versus accuracy.

## Use cases
- reproduce state-of-the-art ARC-AGI benchmark results
- solve ARC-AGI puzzle tasks using frontier LLMs
- evaluate Gemini and OpenAI models on abstract reasoning problems
- experiment with reasoning configs on the ARC-AGI public eval sets
- measure cost versus accuracy trade-offs on ARC-AGI-2
- research LLM abstraction and reasoning capabilities for ARC Prize

## When to choose
- You want to replicate or verify Poetiq's published ARC-AGI-1/2 results for research or benchmarking
- You need a runnable baseline for LLM-based solvers on the ARC-AGI public eval sets
- You are studying prompting and configuration strategies that achieve high reasoning scores at low cost

## When to avoid
- You need production-ready software - the code expects editing constants in main.py and uncommenting configs, and is research-grade
- You want an offline or self-hosted solver - it requires paid Gemini/OpenAI API keys
- You need a general-purpose reasoning engine beyond the ARC-AGI task format
- You want to train or fine-tune models - it only orchestrates inference over existing LLM APIs

## Facets
- artifact type: application
- maturity: active
- function: machine-learning, llm-inference, prompt-engineering, benchmarking
- domain: artificial-intelligence, large-language-models, machine-learning
- platform: python, cli, cross-platform
- tags: arc-agi, abstract-reasoning, benchmark-reproduction, reasoning, llm-orchestration, gemini, openai, state-of-the-art, research-code, puzzle-solving

## Member repositories
- poetiq-ai/poetiq-arc-agi-solver (main) score 42

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:14.057184+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T04:57:43.371657+00:00, confidence not recorded.
  - readme: https://github.com/poetiq-ai/poetiq-arc-agi-solver (fetched 2026-08-28T04:04:14.057184+00:00, sha 2f7a3e977920)
  - homepage: https://poetiq.ai/posts/arcagi_announcement/ (fetched 2026-08-29T12:12:59.034466+00:00, sha ef75c6619edd)
- Data as of 2026-08-30T08:39:29.467469+00:00.