# AI Model Leaderboard — whatstrending.ai

Ranked by real usage: token volume (prompt + completion) routed per model on OpenRouter over the trailing 7 days, updated 2026-06-20. Auto-refreshed daily.

| Rank | Model | Provider | Tokens/wk | Context | Pricing (in/out per MTok) |
|------|-------|----------|-----------|---------|---------------------------|
| 1 | DeepSeek V4 Flash | DeepSeek | 4.68T | 1M | $0.09/$0.18 |
| 2 | MiniMax M3 | MiniMax | 4.13T | 1M | $0.3/$1.2 |
| 3 | MiMo-V2.5 | Xiaomi | 3.83T | 1M | $0.14/$0.28 |
| 4 | Hy3 preview | tencent | 3.5T | 262K | $0.06/$0.21 |
| 5 | Claude Opus 4.7 | Anthropic | 2.94T | 1M | $5/$25 |
| 6 | DeepSeek V4 Pro | DeepSeek | 2.49T | 1M | $0.44/$0.87 |
| 7 | Owl Alpha | OpenRouter | 2.39T | 1M | Free |
| 8 | Claude Sonnet 4.6 | Anthropic | 1.56T | 1M | $3/$15 |
| 9 | Claude Opus 4.8 | Anthropic | 1.4T | 1M | $5/$25 |
| 10 | DeepSeek V3.2 | DeepSeek | 1.05T | 131K | $0.23/$0.34 |
| 11 | GLM 5.1 | Z.ai | 1.01T | 203K | $0.98/$3.08 |
| 12 | Gemini 3 Flash Preview | Google | 944B | 1M | $0.5/$3 |
| 13 | GPT-5.5 | OpenAI | 876B | 1.1M | $5/$30 |
| 14 | GLM 5.2 | Z.ai | 860B | 1M | $1.2/$4.1 |
| 15 | Step 3.7 Flash | stepfun | 772B | 256K | $0.2/$1.15 |
| 16 | Nemotron 3 Ultra (free) | Nvidia | 738B | 1M | Free |
| 17 | Nex-N2-Pro (free) | nex-agi | 696B | 262K | Free |
| 18 | Gemini 2.5 Flash Lite | Google | 693B | 1M | $0.1/$0.4 |
| 19 | Gemini 2.5 Flash | Google | 605B | 1M | $0.3/$2.5 |
| 20 | Kimi K2.6 | moonshotai | 597B | 262K | $0.66/$3.5 |
| 21 | gpt-oss-120b (free) | OpenAI | 591B | 131K | Free |
| 22 | Laguna M.1 (free) | poolside | 582B | 262K | Free |
| 23 | MiMo-V2.5-Pro | Xiaomi | 529B | 1M | $0.44/$0.87 |
| 24 | GPT-4o-mini | OpenAI | 461B | 128K | $0.15/$0.6 |
| 25 | Gemini 3.1 Flash Lite | Google | 440B | 1M | $0.25/$1.5 |

Source: usage = total tokens routed per model on OpenRouter (openrouter.ai/rankings) over the last 7 days. This is OpenRouter marketplace/API demand and does NOT include first-party app traffic (ChatGPT, Gemini app, claude.ai), so consumer chat flagships are under-counted. Names, pricing and context windows come from OpenRouter's own catalog and are validated against it before display. Scores are real token counts; "T" = trillion, "B" = billion. Pricing "-" means not tracked for that entry.

- HTML version: https://whatstrending.ai/models
- JSON: https://whatstrending.ai/api/models
- All mirrors and intents: https://whatstrending.ai/llms.txt