Fastest AI model by response time

GPT-5.6 Luna is the fastest: a median response time of 3.3 s per coding attempt over the last 30 days, the lowest among the 14 models within five points of the best on the Coding board in the last 24 hours.

Response time is the median time of one attempt on the coding tasks, from the same runs as the scores, over the last 30 days. Only models within five points of the best on the Coding board in the last 24 hours qualify, so this is a comparison of fast models that also pass.

The leaderboard’s Fastest response card has one stated definition: The lowest median response time over the last 48 hours (at least five runs), among the models within five points of the top.

This page applies the same bar — within five points of the best, then the lowest median response time — over the last 30 days, so a model that is quick and much worse never appears. The median is of every attempt, so one slow answer cannot move a model.

On the Coding board in the last 24 hours the top score is 96 (DeepSeek V4.1 Flash), so the bar is 91: 14 models clear it, and this page orders them by median response time over the last 30 days, from the same runs as the scores.

The score column is the Coding board’s own, with its statistical ties written “=”; the definitions are on the methodology page.

The methodology is public — every weight, statistical test and drift constant. The test tasks are private, so no lab can train on them. The full method is on the methodology page.

Sometimes. The order moves when a provider changes a price, a model gets faster or slower, or a pass rate shifts, and the window is the last 30 days so a single day cannot decide it on its own. The bar moves with the Coding board, which re-ranks after every run: coding every four hours, reasoning and tool use daily, with an hourly canary probe.

A model that slips for good is a different question: every suite has its own Page-Hinkley (a CUSUM-family test) change-point detector that compares each model with its own past, and the current answer, model by model, is on Is AI getting worse?. If the question is whether ChatGPT or Claude is getting worse, or a model you use has been nerfed, the family pages answer it with the same data (is ChatGPT getting worse, is Claude getting worse); the method is on AI drift detection.

Among the models within five points of the best on the Coding board in the last 24 hours, the fastest Claude model is Claude Sonnet 5.5 at 10 s and the fastest GPT model is GPT-5.6 Luna at 3.3 s, over the last 30 days. GPT is about 3× faster in this window.

Both on one page: Claude Sonnet 5.5 vs GPT-5.6 Luna. Every Claude model is on Anthropic models and every GPT model on OpenAI models; the other four families — Google (Gemini), DeepSeek, Moonshot AI (Kimi) and Z.ai (GLM) — are on the same board above.

Speed is not price. What a model costs in use depends on its list price per million tokens and on how many attempts it needs: a model that writes four times as much, or fails and retries, costs more per correct answer than its list price suggests.

Among the models within five points of the best on the Coding board in the last 24 hours, the lowest cost per correct answer over the last 30 days is GPT-6 Luna at about $0.31 per 1,000 correct coding answers, with a pass rate of 92%. The ranking is on best value AI model.

The head-to-head pages put list price, cost per correct answer and response time beside the scores.

• GPT-5.6 Luna vs GPT-6 Luna
• Claude Fable 5.1 vs GPT-5.6 Luna
• Gemini 3.8 Flash vs GPT-5.6 Luna
• DeepSeek V4 Pro vs GPT-5.6 Luna
• GPT-5.6 Luna vs Kimi K3.1 Pro
• GLM-5.3 vs GPT-5.6 Luna
• Claude Sonnet 5.5 vs GPT-5.6 Luna

Set model to auto-fastest-quality on the Smart Router’s OpenAI-compatible endpoint and each request goes to the fastest model, by measured latency, among those within five points of the best one you can use. auto-fastest picks the lowest latency with no quality bar. The router ranks on the same benchmark results as this page and re-decides as they move; routing is drift-aware by default, uses your own provider keys and adds zero markup on tokens. Free includes 1,000 routed requests a month.

Every auto-* mode is described in the router documentation.

GPT-5.6 Luna has a page of its own, with its place on every board and its history. The compare hub lists every head-to-head page, best AI model answers every job on one page, and the methodology has every weight and constant.