Updated 4 November 2026 · Artificial Analysis
The fastest AI model
GPT-6.1 Terra from OpenAI gives the fastest complete answer of any capable model: a 500-token reply in 8.7 s (3.1 s to start, then 88 tok/s).
| # | Model | Full answer | First token | Speed | Intelligence | Price / 1M tokens |
|---|---|---|---|---|---|---|
| 1 | GPT-6.1 Terra OpenAI | 8.7 s | 3.1 s | 88 tok/s | 38 | $2.00 |
| 2 | Claude Fable 5.1 Anthropic | 11 s | 0.7 s | 48 tok/s | 39 | $20 |
| 3 | Gemini 4 Carbon Google | 12 s | 5.5 s | 81 tok/s | 54 | $8.00 |
| 4 | GPT-6 Astra OpenAI | 13 s | 1.8 s | 43 tok/s | 43 | $20 |
| 5 | Claude Sonnet 5.5 Anthropic | 14 s | 8.6 s | 92 tok/s | 42 | $4.00 |
| 6 | GPT-6.1 Sol OpenAI | 14 s | 4.7 s | 52 tok/s | 46 | $4.00 |
| 7 | Claude Opus 5.5 Anthropic | 19 s | 12 s | 72 tok/s | 48 | $8.00 |
| 8 | Kimi K3.1 Pro Kimi · open weights | 20 s | 2.4 s | 28 tok/s | 47 | $4.00 |
| 9 | Muse Wonder Meta | 31 s | 27 s | 130 tok/s | 44 | $8.00 |
| 10 | Gemini 4 Argon Google | 42 s | 36 s | 81 tok/s | 50 | $8.00 |
Ranked by how long a 500-token answer takes: the median wait for the first token plus 500 tokens at the median output speed, both measured by Artificial Analysis. Models that think before answering can write fast and still finish last. Only models scoring at least 36 on the Intelligence Index are listed.
Data: Artificial Analysis (intelligence and coding indexes, prices, speed) and OpenRouter (open weights, context windows). Updated 4 November 2026; refreshed every 6 hours.
More rankings
All rankings · Best overall · Best open model · Best value · Cheapest good model · Best for coding · Longest context · Price table