vishva.lol is a creative coding laboratory and playground for interactive web experiments, aesthetic productivity tools, and digital toys developed by Vishva Variya.

What kind of projects are on vishva.lol?

The site features a variety of experiments including an aesthetic Pomodoro timer, a startup pivot museum, a live flight tracker (Sky Spy), an AI model leaderboard (AI Battle), 3D character toys (Toodles), and engine sound simulators (Roar).

Is vishva.lol free to use?

Yes, all experiments and tools on vishva.lol are free to use and designed for browser-based interaction without requiring any downloads or signups.

AI Battle is a live weekly leaderboard game where users vote for their favorite AI models to see which one tops the rankings.

How do I vote in AI Battle?

Tap the boost button next to your favorite model. You can vote multiple times to push your preferred AI up the leaderboard.

Which AI models are included?

The default lineup features Kimi K2.6, Claude Sonnet 4.6, DeepSeek V3.2, Claude Opus 4.7, Gemini 3 Flash, Hunyuan HY3, Step 3.5 Flash, MiniMax M2.7, Grok 4.1 Fast, Nemotron 3 Super 120B A12B. Search the catalog to add others from makers like Moonshot AI, Anthropic, DeepSeek, Google, Tencent Hunyuan, StepFun AI, MiniMax, xAI, NVIDIA.

How often does the leaderboard reset?

AI Battle runs on weekly cycles. A new battle starts each week so the leaderboard always reflects current model momentum.

←./vishva.lol

ai_battle·weekly_

SYNC

weekly coronation

Crownthemodel

vote weekly · 10 models in pool · one tap, one vote

resets --d --:--:--your votes 0

top company

Moonshot AI

0 votes

AI Battle — a weekly popularity arena for AI models

Tap to boost your favourite large language model toward the top of a live, weekly leaderboard.

AI Battle is a casual popularity contest for AI models. Every week the leaderboard resets, the default lineup of frontier models starts at zero, and visitors tap a boost button to push their favourite up the rankings. The live activity feed shows recent votes streaming in from other tabs around the world.

It is not a benchmark, not a capability test, and not affiliated with any of the model makers. It is a public mood board for which AI brand people felt strongest about in any given week.

How voting works

Each tap on a model card sends an anonymous vote to a Firestore-backed counter. There is no account, no email, no identifier tied to your votes — only an aggregate per-model count and a few rolling activity samples. Votes are rate-limited per device to keep one enthusiastic visitor from drowning the leaderboard.

The leaderboard ranks models by total boosts received during the current weekly window. When the week rolls over, the counters reset and a fresh battle starts.

The default lineup

The arena ships with a curated set of frontier models: Kimi K2.6, Claude Sonnet 4.6, DeepSeek V3.2, Claude Opus 4.7, Gemini 3 Flash, Hunyuan HY3, Step 3.5 Flash, MiniMax M2.7, Grok 4.1 Fast, Nemotron 3 Super 120B A12B. The catalog search lets you add others from makers like Moonshot AI, Anthropic, DeepSeek, Google, Tencent Hunyuan, StepFun AI, MiniMax, xAI, NVIDIA so you can champion an underdog or a niche favourite.

Why this exists

Most public AI rankings are either capability benchmarks or paid market reports. AI Battle is the opposite: a low-stakes, free-to-play snapshot of public enthusiasm. Use it as a conversation piece, not a procurement recommendation.

Frequently asked questions

Are the votes real?: Yes. Every tap increments a real Firestore counter shared by every visitor. The feed shows live samples of recent boosts.
Is there a prize?: No. Nothing is staked or paid out. It is a popularity leaderboard, not a sweepstakes or contest of skill with rewards.
How often does the leaderboard reset?: Weekly. A fresh battle starts each cycle, so the rankings reflect current sentiment instead of all-time totals.
Can my company / model be added?: The catalog grows on its own. Search for the model you want to add, and if it is missing, reach out through the contact page.