Can AI Return a Copy-Ready Spreadsheet Formula? I Tested 3 Models

Anvi Pardhi — Wed, 07 Oct 2026 05:20:50 +0000

This is a submission for the Kaggle Benchmarking Challenge

What I Benchmarked

I tested whether language models can update Excel-style cell references when a formula is copied to a new cell.

The 20 cases covered relative, absolute, and mixed references; ranges; formulas with multiple references; and copying in different directions. Each model was asked to return only the updated formula. I scored an answer as correct when its first line matched the expected formula, ignoring whitespace and letter case.

I started with five simple cases, and all three models scored 1.00. That ceiling result pushed me to add longer copy moves and more complex formulas.

Models Tested

I tested Claude Sonnet 5.5, Claude Opus 5.5, and Gemini 3.7 Flash—the models available to me on Kaggle. The two Claude models let me compare results within one model family; Gemini gave me a comparison across providers.

Findings

Model

Score

Claude Sonnet 5.5

0.25

Claude Opus 5.5

1.00

Gemini 3.7 Flash

1.00

The biggest surprise was what the low score meant. Sonnet often explained how the references should move, but did not put a complete, copy-ready formula on the first line. One case returned the original formula unchanged. Because this task requires a usable formula, those responses failed—even when the explanation suggested the model understood part of the transformation.

That changed how I think about spreadsheet assistance: understanding a formula’s movement is only part of the job. The output also has to be usable in the cell.

Next, I’d separate formula correctness from output-format compliance, then test more Excel edge cases and repeat runs to see whether the model differences hold.

My Benchmark

kaggle.com/benchmarks/anvipardhi/fill-accuracy

Trail Quest: Private, On-Device AI for Nature Walks

Anvi Pardhi — Tue, 06 Oct 2026 09:48:24 +0000

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass

What I Built

Trail Quest is a mobile-first web app that turns a little time outdoors into a nature-spotting quest. Choose a setting—such as a park, garden, neighborhood, trail, or balcony—and a time limit. Trail Quest gives you three observation challenges, then encourages you to put your phone away and explore.

You can complete or skip challenges, keep private reflection notes, and save completed walks in a local nature journal. There’s also a high-contrast sunlight mode for using the app outdoors.

Demo

Try Trail Quest: https://trail-quest-l6yb.onrender.com

Code

Source code: https://github.com/Anvi-2006/trail-quest

How I Built It

Trail Quest uses WebLLM to run the open-weight SmolLM2-135M-Instruct-q4f16_1-MLC model in the browser with WebGPU. Quest generation happens on the user’s device, so the user’s setting and prompts aren’t sent to an AI API.

If the model can’t run on a device, Trail Quest offers clearly labeled handcrafted quests instead. Handcrafted mode works offline. AI mode needs an initial model download; after that, the browser can reuse its cached model for local generation, subject to browser storage and WebGPU support.

I built the interface around a short, accessible flow: choose a setting and duration, get three quests, and then switch to a “phone away” screen. The app also includes safety guidance, local journal storage, Markdown export, and synthesized chimes made with the Web Audio API.

Why Does Open Innovation Matter?

The open-weight model and browser-based inference make this experience possible without sending a person’s outdoor plans to a hosted AI service. The model can be inspected and swapped, and the handcrafted fallback keeps the app useful when AI inference isn’t available.

That matters for an app whose goal is to help people pay attention to the world around them: it can keep personal inputs on-device, reduce dependence on a paid API, and let the community adapt the experience for different models and devices.

Prize Categories

Best Use of Render — Render hosts the Trail Quest web app.

DEV Community: Anvi Pardhi