OpenRouter · 2026-07-29

OpenRouter rankings

Every model on the OpenRouter gateway, ranked by the tokens it actually processed. Not a benchmark — this is where developer traffic and developer money went.

Models tracked

50

with measurable traffic

Tokens / day

7.5T

latest archived day

Est. $ / day

$22.8M

tokens × list price

Leader

14.77%

DeepSeek: DeepSeek V4 Flash

What the OpenRouter rankings show in July 2026

DeepSeek: DeepSeek V4 Flash holds 14.77% of tracked OpenRouter tokens, moving +4.50pp over the last 30 days. At the lab level, DeepSeek accounts for 22.3% of all traffic across 3 models.

Share is computed from the daily token counts OpenRouter publishes, archived once a day so that 7-day and 30-day changes compare like for like. Estimated dollars apply each model's list price (30% prompt / 70% completion) to its token volume, which is why a model can carry a large share of tokens and a small share of spend.

Top 50 models on OpenRouter

Ranked by share of gateway tokens on the latest archived day.

Tidelines — LLM usage and market share
#ModelLabShare %Tokens/d7d Δ30d Δ
1DeepSeek: DeepSeek V4 FlashDeepSeek
14.77
1.2T+4.64pp+4.50pp
2Xiaomi: MiMo-V2.5Xiaomi
8.97
736B-8.35pp+0.31pp
3Tencent: Hy3Tencent
7.49
614B-0.42pp+0.36pp
4DeepSeek: DeepSeek V4 ProDeepSeek
6.72
551B+1.27pp+2.33pp
5Z.ai: GLM 5.2Z-AI
5.70
467B-1.13pp+0.86pp
6NVIDIA: Nemotron 3 UltraNvidia
4.37
358B+1.52pp+2.62pp
7StepFun: Step 3.7 FlashStepFun
3.96
325B-1.10pp+0.75pp
8MiniMax: MiniMax M3MiniMax
3.72
305B-0.10pp-4.76pp
9MoonshotAI: Kimi K3Moonshot
2.95
242B+0.90pp+2.20pp
10Ling-3.0-flash (free)Other
2.77
227B+1.25pp+1.25pp
11Claude Opus 5Anthropic
2.16
177B+1.16pp+1.16pp
12Anthropic: Claude Sonnet 4.6Anthropic
2.13
174B+0.69pp-1.54pp
13Anthropic: Claude Sonnet 5Anthropic
2.02
166B+0.17pp+1.80pp
14Google: Gemini 3 Flash PreviewGoogle
1.92
158B+0.18pp-0.11pp
15Google: Gemini 2.5 Flash LiteGoogle
1.43
117B+0.47pp+0.25pp
16Anthropic: Claude Opus 4.8Anthropic
1.23
101B-2.29pp-2.77pp
17Xiaomi: MiMo-V2.5-ProXiaomi
1.16
95B+0.19pp+0.03pp
18Google: Gemini 3.1 Flash LiteGoogle
1.12
92B+0.06pp+0.21pp
19Google: Gemini 2.5 FlashGoogle
1.00
82B-0.04pp-0.29pp
20OpenAI: GPT-5.6 LunaOpenAI
0.94
77B+0.27pp+0.61pp
21OpenAI: gpt-oss-120bOpenAI
0.93
76B+0.13pp-0.12pp
22OpenAI: GPT-5.6 SolOpenAI
0.86
71B-0.15pp+0.21pp
23OpenAI: GPT-5.6 TerraOpenAI
0.83
68B+0.52pp+0.61pp
24DeepSeek: DeepSeek V3.2DeepSeek
0.80
65B+0.23pp-0.14pp
25OpenAI: GPT-5.4OpenAI
0.79
65B+0.02pp+0.04pp
26xAI: Grok 4.5xAI
0.75
61B-0.42pp+0.14pp
27Poolside: Laguna S 2.1Other
0.71
58B+0.10pp+0.38pp
28Google: Gemma 4 26B A4B Google
0.69
57B+0.16pp+0.08pp
29Google: Gemma 4 31BGoogle
0.68
56B+0.02pp+0.09pp
30Cohere: North Mini Code (free)Other
0.68
56B+0.34pp+0.38pp
31NVIDIA: Nemotron 3 SuperNvidia
0.62
51B-0.02pp-0.17pp
32Pplx Embed V1 0.6BOther
0.57
46B+0.20pp+0.20pp
33Google: Gemini 3.5 FlashGoogle
0.54
44B-0.15pp-0.33pp
34Anthropic: Claude Fable 5Anthropic
0.45
37B-0.10pp-0.40pp
35Google: Gemini 3.6 FlashGoogle
0.44
36B+0.01pp+0.01pp
36OpenAI: GPT-4o-miniOpenAI
0.42
35B+0.03pp-0.14pp
37Anthropic: Claude Haiku 4.5Anthropic
0.41
34B-0.01pp-0.12pp
38inclusionAI: Ling-2.6-flashOther
0.41
34B+0.13pp+0.05pp
39OpenAI: GPT-5.5OpenAI
0.40
33B-0.12pp-1.94pp
40MoonshotAI: Kimi K2.5Moonshot
0.37
30B-0.19pp+0.06pp
41Anthropic: Claude Opus 4.7Anthropic
0.36
29B-2.06pp-3.46pp
42OpenAI: GPT-5.4 MiniOpenAI
0.35
28B+0.05pp-0.05pp
43Anthropic: Claude Opus 4.6Anthropic
0.34
28B+0.00pp-0.19pp
44OpenAI: GPT-5.4 NanoOpenAI
0.34
28B+0.04pp+0.08pp
45Google: Gemini 3.1 Pro PreviewGoogle
0.32
26B-0.08pp-0.12pp
46MoonshotAI: Kimi K2.6Moonshot
0.32
26B-0.61pp-0.37pp
47Poolside: Laguna XS 2.1Other
0.32
26B+0.13pp+0.03pp
48Qwen: Qwen3.7 FlashAlibaba
0.31
26B+0.00pp+0.00pp
49MiniMax: MiniMax M2.7MiniMax
0.28
23B+0.07pp-0.02pp
50OpenAI: GPT-5 MiniOpenAI
0.28
23B+0.02pp-0.02pp
Data: OpenRouter as of 2026-07-29 · Source: tidelines.ai/openrouter-rankings

OpenRouter rankings FAQ

What are the OpenRouter rankings?
A daily ranking of every model routed through the OpenRouter gateway, ordered by the share of tokens it processes. Tidelines archives the public OpenRouter token counts each day and computes share, momentum and estimated spend from them.
Which model has the highest OpenRouter usage?
DeepSeek: DeepSeek V4 Flash (DeepSeek) leads with 14.77% of tracked OpenRouter tokens as of 2026-07-29.
How often do the rankings update?
Once a day. Each day's snapshot is archived, so the 7-day and 30-day changes are measured against real historical data rather than recomputed estimates.
Is this the same as an LLM benchmark leaderboard?
No. Benchmark leaderboards score models on test sets. These rankings measure what developers actually send traffic to and pay for.