Usage rankings

Ranked from input and output tokens that actually passed through everyais—not vendor claims or estimates.

Method

The denominator is all input + output tokens in the period; trends and generation speed require at least 100 relevant measurements.

Speed and trend metrics require at least 100 measured requests in the relevant sample. An em dash means unavailable—not zero.

Composition within the listed top models (totals 100%).

Based on input + output tokens. Share uses all catalog-visible token traffic, including models beyond the displayed limit.

Model token usage ranking
#	Model	Usage share
1	GPT-5.6 Luna everyais/gpt-5-6-luna	28.6%
2	Gemini 3.5 Flash Lite everyais/gemini-3-5-flash-lite	26.7%
3	Claude Haiku 4.5 everyais/claude-haiku-4-5	15.9%
4	Claude Opus 4.8 everyais/claude-opus-4-8	7.1%
5	Claude Sonnet 5 everyais/claude-sonnet-5	4.0%
6	Gemini 3.7 Flash everyais/gemini-3-7-flash	3.4%
7	DeepSeek-V4-Flash-0731 everyais/deepseek-v4-flash-0731	3.1%
8	Kimi-K3 everyais/kimi-k3	2.7%
9	GPT-5.6 Sol everyais/gpt-5-6-sol	2.7%
10	Gemini 3.1 PRO Preview Customtools everyais/gemini-3-1-pro-preview-customtools	2.2%
11	GPT-5.6 Terra everyais/gpt-5-6-terra	1.0%
12	Gemini 3.1 PRO Preview everyais/gemini-3-1-pro-preview	0.8%
13	Claude Opus 5 everyais/claude-opus-5	0.8%
14	gemini-3.8-flash everyais/gemini-3-8-flash	0.6%
15	Gemini 2.5 Flash everyais/gemini-2-5-flash	0.5%
16	GLM-5.3-Flash everyais/glm-5-3-flash	<0.1%
17	google.gemma-4-26b-a4b everyais/google-gemma-4-26b-a4b	<0.1%
18	Claude Fable 5 everyais/claude-fable-5	<0.1%

Average streaming speed among the 18 token-ranked models above, not the entire catalog. Fewer than 100 measurements are excluded.

Not enough data — generation speed requires at least 100 streaming measurements per model.