SmophyAI

Published daily · Next update 02:00 UTC

Daily snapshot · immutable permalink

AI model usage rankings - September 21, 2026

As published on September 21, 2026, Z.ai: GLM 5.3 Flash (batch) led real OpenRouter usage share at 20.9%. Anthropic: Claude Sonnet 4.6 (batch), Meta: Muse Spark 1.3, OpenAI: GPT-5.6 Luna Pro (batch), Anthropic: Claude Opus 4.8 (batch) entered the leaderboard for the first time. Qwen: Qwen3.7 Flash moved 9 places down since September 20, 2026.

  1. #1 Z.ai: GLM 5.3 Flash (batch) - 20.9% usage share
  2. #2 DeepSeek: DeepSeek V4.1 Flash (batch) - 11.7% usage share
  3. #3 Tencent: Hy4 preview - 9.0% usage share
  4. #4 DeepSeek: DeepSeek V4 Flash 0423 - 8.3% usage share
  5. #5 OpenAI: GPT-5.6 Luna (batch) - 6.3% usage share
  6. #6 Xiaomi: MiMo-V2.5 - 4.7% usage share
  7. #7 NVIDIA: Nemotron 3 Ultra (free) - 3.6% usage share
  8. #8 Z.ai: GLM 5.3 (batch) - 2.5% usage share
  9. #9 Tencent: Hy3 - 2.4% usage share
  10. #10 Google: Gemini 3.8 Flash (batch) - 1.7% usage share
Change since September 20, 2026
#ModelUsage shareRank change
01Z.ai: GLM 5.3 Flash (batch)20.9%no change
02DeepSeek: DeepSeek V4.1 Flash (batch)11.7%no change
03Tencent: Hy4 preview9.0%no change
04DeepSeek: DeepSeek V4 Flash 04238.3%no change
05OpenAI: GPT-5.6 Luna (batch)6.3%no change
06Xiaomi: MiMo-V2.54.7%no change
07NVIDIA: Nemotron 3 Ultra (free)3.6%no change
08Z.ai: GLM 5.3 (batch)2.5%+1 (+0.7pts)
09Tencent: Hy32.4%-1 (-0.4pts)
10Google: Gemini 3.8 Flash (batch)1.7%+2 (+0.2pts)
11Meta: Muse Spark 1.3 Contributor1.4%-1 (-0.3pts)
12DeepSeek: DeepSeek V4 Pro 04231.3%+2 (0pts)
13Upstage: Solar Pro 41.3%-2 (-0.2pts)
14Z.ai: GLM 5.21.2%-1 (-0.2pts)
15OpenAI: GPT-5.6 Sol (batch)1.1%no change
16Anthropic: Claude Sonnet 5 (batch)1.1%+4 (+0.4pts)
17MiniMax: MiniMax M3 (batch)1.1%-1 (0pts)
18MoonshotAI: Kimi K3 (batch)1.0%-1 (0pts)
19inclusionAI: Ling 3.0 Flash Fin0.9%no change
20Anthropic: Claude Opus 5 (batch)0.8%+1 (+0.2pts)
21Poolside: Laguna S 2.1 (free)0.8%-3 (-0.2pts)
22OpenAI: GPT-6 Astra (batch)0.8%no change
23Google: Gemini 3 Flash Preview (batch)0.6%+1 (0pts)
24OpenAI: GPT-5.6 Terra (batch)0.5%+4 (+0.2pts)
25Google: Gemini 2.5 Flash Lite0.5%no change
26inclusionAI: Ling 3.0 Flash VL0.4%no change
27OpenAI: gpt-oss-120b (batch)0.4%+5 (+0.1pts)
28Nex AGI: Nex-N2.5-Pro0.4%-5 (-0.2pts)
29Google: Gemini 3.7 Flash (batch)0.4%no change
30Google: Gemini 2.5 Flash0.4%+8 (+0.1pts)
31Google: Gemini 3.1 Flash Lite (batch)0.4%+5 (+0.1pts)
32Anthropic: Claude Fable 5.1 (batch)0.4%+2 (0pts)
33Qwen: Qwen3.8 Flash0.3%-3 (0pts)
34NVIDIA: Nemotron 3.5 Lightning (free)0.3%-7 (-0.1pts)
35Google: Gemma 4 31B (free)0.3%+4 (0pts)
36Anthropic: Claude Sonnet 4.6 (batch) New0.3%-
37Meta: Muse Spark 1.3 New0.3%-
38DeepSeek: DeepSeek V3.20.3%-3 (0pts)
39Qwen: Qwen3.8 27B (batch)0.3%+3 (0pts)
40Qwen: Qwen3.7 Flash0.3%-9 (-0.1pts)
41Thinking Machines: Inkling0.3%-8 (-0.1pts)
42Google: Gemma 4 26B A4B (free)0.2%+2 (0pts)
43SpaceXAI: Grok 4.6 (batch)0.2%+3 (0pts)
44OpenAI: GPT-5.6 Luna Pro (batch) New0.2%-
45Anthropic: Claude Opus 4.8 (batch) New0.2%-
46NVIDIA: Nemotron 3 Super0.2%-6 (-0.1pts)

This page is a permanent, unchanging record of real OpenRouter usage share as it stood on September 21, 2026. Later corrections to the data pipeline appear as new dated snapshots - this page's figures never change.

Source: OpenRouter (openrouter.ai/rankings), as of September 21, 2026.

Benchmark scores: Artificial Analysis (artificialanalysis.ai) via OpenRouter (openrouter.ai/rankings).

Methodology: www.smophy.ai/benchmark/methodology

Token counts originate from each provider's own tokenizer and are not directly comparable across providers.