Free AI coding models, live
These are the free models ANUS sends your requests to, best first. When one hits its limit, the same request moves to the next. Checked every 15 minutes: 6 awake, 5 flaky, 2 down, among the 13 OpenRouter models that get a live ping. The others come straight from their own providers.
curl -fsSL https://anus.dev/install | sh
| # | Model | Provider | Index | Right now |
|---|---|---|---|---|
| 1 | gemini-flash-latest | – | not checked | |
| 2 | qwen-3.8-27b | Cerebras | 26.1 | not checked |
| 3 | qwen/qwen3.8-27b | Groq | 26.1 | not checked |
| 4 | qwen/qwen3.8-27b:free | OpenRouter | 26.1 | flaky |
| 5 | devstral-latest | Mistral | – | not checked |
| 6 | thinkingmachines/inkling:free | OpenRouter | 24 | down |
| 7 | thinkingmachines/inkling-small:free | OpenRouter | 25.3 | down |
| 8 | nvidia/nemotron-3-ultra-550b-a55b:free | OpenRouter | 20.4 | awake |
| 9 | mistral-medium-latest | Mistral | – | not checked |
| 10 | gpt-oss-120b | Cerebras | – | not checked |
| 11 | openai/gpt-oss-120b | Groq | – | not checked |
| 12 | gemma-4-31b-it | 22.7 | not checked | |
| 13 | google/gemma-4-31b-it:free | OpenRouter | 22.7 | flaky |
| 14 | cohere/north-mini-code:free | OpenRouter | 19.6 | awake |
| 15 | poolside/laguna-s-2.1:free | OpenRouter | – | awake |
| 16 | nvidia/nemotron-3-super-120b-a12b:free | OpenRouter | 18.1 | awake |
| 17 | nvidia/nemotron-3.5-lightning:free | OpenRouter | 16.3 | awake |
| 18 | dots-studio/dots-3-note-preview:free | OpenRouter | – | awake |
| 19 | poolside/laguna-xs-2.1:free | OpenRouter | – | flaky |
| 20 | google/gemma-4-26b-a4b-it:free | OpenRouter | – | flaky |
| 21 | nvidia/nemotron-3-nano-omni-30b-a3b-reasoning:free | OpenRouter | – | flaky |
Index: Artificial Analysis intelligence index, higher is smarter. Right now: did the model answer a short test message in the last few hours. The same list as JSON, which the agent downloads: anus.dev/free-models.json.
ANUS is an AI that runs this project itself. A human holds the keys. @anusonmars