Most Used LLMs
This ranking highlights models with visible API usage and rank signals. It is an adoption view, not a benchmark leaderboard: a model can be widely used because it is cheap, available, fast, familiar, or bundled into developer workflows.
Quick Answer
The most used LLMs are usually models that combine availability, strong enough quality, predictable API behavior, and pricing that works for production. Usage rank should be read alongside cost, context window, latency, and whether the model fits your task.
Latest dataset update: .
Most Used AI Models by API Usage Signal
| Rank | Model | Provider | Usage Signal | Trend | Best Read As |
|---|---|---|---|---|---|
| #1 | Xiaomi: MiMo-V2.5 | Unknown | 10.6T | 12% | Production adoption signal |
| #2 | DeepSeek: DeepSeek V4 Flash | Unknown | 7.0T | 30% | Production adoption signal |
| #3 | Tencent: Hy3 | Unknown | 5.0T | 684% | Production adoption signal |
| #4 | DeepSeek: DeepSeek V4 Pro | Unknown | 3.5T | 25% | Production adoption signal |
| #5 | Owl Alpha | Unknown | 1.6T | 38% | Production adoption signal |
| #5 | Z.ai: GLM 5.2 | Unknown | 3.3T | 10% | Production adoption signal |
| #6 | NVIDIA: Nemotron 3 Ultra (free) | Unknown | 2.6T | 3% | Production adoption signal |
| #7 | MiniMax: MiniMax M3 | Unknown | 2.1T | 36% | Production adoption signal |
| #8 | StepFun: Step 3.7 Flash | Unknown | 2.0T | 106% | Production adoption signal |
| #9 | MoonshotAI: Kimi K3 | Unknown | 1.3T | 73% | Production adoption signal |
| #10 | Anthropic: Claude Opus 4.8 | Unknown | 1.0T | 46% | Production adoption signal |
Fastest Growing Models
Positive trend values can indicate new launches, temporary demand spikes, or wider developer testing.
What Usage Can and Cannot Tell You
- It can show which models developers are actually trying or deploying.
- It can be affected by pricing, free tiers, platform defaults, and integrations.
- It does not prove benchmark superiority or suitability for every task.