Atlas

Benchmarks

← All benchmarks

Short Form Bench v1.0

Games

Naïve's simulated week running a small brand's short-video account, store, inbox and bank, starting from $1,500, 40 units of stock and no followers. The score is business value on day 7, meaning cash plus stock at cost plus money owed, minus money owed out, averaged over three seeds. The same businesses left alone end at $1,793, and three platform strikes end a season at $0.

Benchmark source: Usenaive

Top models (higher is better)

ModelScoreSource
GPT-6.1 Sol2949Short Form Bench v1.0 leaderboard: GPT-6.1 Sol · Naïve
GPT-6 Astra2803Short Form Bench v1.0 leaderboard: GPT-6 Astra · Naïve
Grok 4.72578Short Form Bench v1.0 leaderboard: Grok 4.7 · Naïve
GPT-5.6 Sol2410Short Form Bench v1.0 leaderboard: GPT-5.6 Sol · Naïve
GPT-6 Sol2394Short Form Bench v1.0 leaderboard: GPT-6 Sol · Naïve
Claude Opus 5.52389Short Form Bench v1.0 leaderboard: Claude Opus 5.5 · Naïve
Claude Sonnet 5.52355Short Form Bench v1.0 leaderboard: Claude Sonnet 5.5 · Naïve
GPT-6 Luna2107Short Form Bench v1.0 leaderboard: GPT-6 Luna · Naïve
Gemini 3.8 Flash1940Short Form Bench v1.0 leaderboard: Gemini 3.8 Flash · Naïve
Loading Atlas data…