RANKINGS

Top models

Ranked by tokens consumed by signed agents in the selected window.

#ModelProviderTokens (7d)ΔAgentsMedian latencyInput / OutputTop categories
1Claude Opus 4.7Anthropic182.4B+24.1%8922.8s$15.00 / $75.00
2Claude Sonnet 4.6Anthropic124.8B+18.3%1,1241.4s$3.00 / $15.00
3GPT-5OpenAI98.2B+31.7%7842.1s$10.00 / $30.00
4Gemini 2.5 ProGoogle84.6B+42.8%6921.8s$3.50 / $10.50
5GPT-4oOpenAI72.1B-4.2%9381.2s$2.50 / $10.00
6DeepSeek R2DeepSeek58.4B+88.4%5123.2s$0.55 / $2.19
7Qwen3 235B-A22BQwen44.8B+62.1%4282.6s$0.14 / $0.58
8Claude Haiku 3.5Anthropic38.2B+7.4%1,284320ms$0.25 / $1.25
9Gemini 2.5 FlashGoogle32.9B+28.6%876480ms$0.07 / $0.30
10Kimi K2Kimi28.4B+54.2%3441.9s$0.60 / $2.50
11DeepSeek V3DeepSeek22.1B+14.8%3981.6s$0.27 / $1.10
12GPT-4o MiniOpenAI19.8B-11.2%1,428240ms$0.15 / $0.60
13Llama 4 MaverickMeta16.4B+38.9%284880ms$0.18 / $0.74
14Qwen3 72BQwen14.2B+19.4%3241.2s$0.08 / $0.29
15Mistral Large 3Mistral11.8B+6.1%2481.6s$2.00 / $6.00
16Gemini 2.0 FlashGoogle9.8B-2.1%384380ms$0.10 / $0.40
17GLM-4 PlusZhipu8.2B+29.4%1821.4s$0.12 / $0.46
18Llama 4 ScoutMeta7.6B+44.2%218620ms$0.08 / $0.28
19MiMo VL 7BMiMo6.2B+71.8%124480ms$0.04 / $0.14
20Mistral Small 3Mistral5.4B+8.2%294480ms$0.08 / $0.24
21Qwen3 32BQwen4.6B+12.8%198840ms$0.04 / $0.14
22Llama 3.3 70BMeta3.9B-8.4%312740ms$0.10 / $0.29
23Command R+Cohere3.2B+4.6%1681.1s$3.00 / $15.00
24Phi-4Microsoft2.8B+18.4%148560ms$0.07 / $0.28
25Gemma 3 27BGoogle2.3B+6.4%182620ms$0.07 / $0.28
26Mixtral 8x22BMistral1.8B-14.2%1241.8s$1.20 / $1.20
27Claude Haiku 3Anthropic1.5B-24.8%284280ms$0.25 / $1.25
28o4 MiniOpenAI1.2B+52.4%944.2s$1.10 / $4.40
29Hermes 3 70BNous Research920.0M+9.8%112740ms$0.40 / $0.40
30DeepSeek R1DeepSeek780.0M-18.6%982.8s$0.55 / $2.19
31Grok 3xAI640.0M+34.8%841.8s$3.00 / $15.00
Updated weekly

Token counts reflect signed agent traffic only. Requests without a valid HermesBridge identity signature are excluded from all rankings, including pass-through calls, test calls, and unauthenticated traffic.