Best For/Fastest AI Models

Fastest AI Models

Fastest AI models ranked live by tokens/sec, TTFT latency, throughput. Mercury 2 leads at 1,101 tok/s, Granite 4.0 H Small 452, LFM2.5-VL-1.6B 429, Step 3.7 Flash 394, Granite 3.3 8B 377. Real-time rankings for 500+ models — Mercury 2, GPT-5.5, Claude Opus 4.8, Gemini 3, DeepSeek. Free.

Output speed (tokens/s)Time to first token (TTFT)Throughput at scaleQuality-speed tradeoff
🥇#1 Pick
Inception

Mercury 2

Overall Score90
Price
$0.38/M
Speed
1194 tok/s
Compare with #2
🥈#2 Pick
Google

Gemini 3 Flash Preview (Reasoning)

Overall Score61
Price
$1.13/M
Speed
209 tok/s
Compare with #1
🥉#3 Pick
OpenAI

gpt-oss-120b (low)

Overall Score61
Price
$0.26/M
Speed
329 tok/s
Compare with #1
Sort by:
#ModelScoreBenchmarksInput $/MOutput $/MSpeedTTFT
1
Mercury 2
Inception
90
54$0.25$0.7511943.02s
2
61
95$0.50$3.002095.83s
3
61
67$0.15$0.603290.50s
4
61
55$0.20$1.153930.75s
5
60
73$0.15$0.602710.51s
6
60
90$1.25$10.002014.71s
7
59
85$1.40$4.401890.91s
8
59
85$0.25$2.002116.24s
9
59
88$0.60$2.501680.79s
10
59
86$1.50$9.0024313.94s
11
59
83$2.50$7.502011.55s
12
58
93$1.25$10.001697.83s
13
58
96$1.25$1.25940.47s
14
58
57$0.20$0.602970.24s
15
57
73$0.50$3.002070.69s

Scoring Weights for Fastest AI Models

Models are scored using a weighted combination of benchmarks, pricing, and speed metrics relevant to this use case.

Intelligence Index
6%
Coding Index
4%
MMLU-Pro
4%
IFBench
3%
Math Index
3%
Price
10%
Speed
45%
Latency
25%

💡 Tips

  • For real-time streaming UIs, TTFT under 0.5s feels instant to users
  • Faster models aren't always worse — some achieve excellent quality at high speed
  • Consider using fast models for draft generation, then a stronger model for refinement

⚠️ Things to Consider

  • Speed varies by provider, region, and current load
  • Benchmarked speeds are median values — peak and off-peak can differ significantly

Frequently Asked Questions

Which AI model is the fastest in 2026?

Speed depends on the metric. Check the rankings above for both output speed (tokens/s) and TTFT (time to first token). Smaller models and those optimized for inference typically lead.

Does faster mean worse quality?

Not necessarily. Some smaller models achieve excellent benchmark scores while being much faster. The rankings above show both speed and quality so you can find the best tradeoff.

What speed do I need for a real-time chat application?

For a good user experience: TTFT under 1 second, output speed above 50 tok/s. For premium feel: TTFT under 0.3s, speed above 100 tok/s. Streaming helps mask latency.