NVIDIA NIM
Free, no card. Dozens of models; fast and reliable at peak hours. 免費、免卡。數十種模型；尖峰時段快又穩。
FreeNo cardCloud
- Provider
- NVIDIA NIM
- Model
meta/llama-3.1-8b-instruct- Env var
NVIDIA_API_KEY
askLLM talks to a model you bring yourself. Pick the one that fits your situation, get a key, and tell askLLM where to find it — most options are free. askLLM 使用你自備的模型。依你的情況挑一個、申請金鑰、再告訴 askLLM 到哪裡找它——多數選項都免費。
Tap one to highlight the providers that match. Not sure? Any card works — start with NVIDIA or Gemini. 點一個就會標出符合的供應商。不確定？任一張都行——先從 NVIDIA 或 Gemini 開始。
Each card shows exactly what to select in askLLM’s LLM settings. Model names change often — the “get a key” page always lists what’s current. 每張卡都寫明在 askLLM 的 LLM settings 要選什麼。模型名稱常變動——「申請金鑰」頁面永遠列出最新的。
Free, no card. Dozens of models; fast and reliable at peak hours. 免費、免卡。數十種模型；尖峰時段快又穩。
meta/llama-3.1-8b-instructNVIDIA_API_KEYFree, no card. Uses a self-updating model alias, so it never goes stale. 免費、免卡。用會自動更新的模型別名，永遠不會過期。
gemini-flash-latestGEMINI_API_KEYgemini-flash-latest always points at Google’s current Flash model — no need to update it.
gemini-flash-latest 永遠指向 Google 當前的 Flash 模型——不必自己更新。
Free, no card. One key, dozens of vendors — pick any :free model.
免費、免卡。一組金鑰通吃數十家廠商——任選一個 :free 模型。
openai/gpt-oss-20b:freeOPENROUTER_API_KEYFree models need a :free suffix on the model name. Quota: 50 requests/day with no top-up, 1000/day after a one-time US$10+ top-up.
免費模型的模型名稱需加 :free 後綴。額度:未儲值每日 50 次,一次性儲值 US$10 以上後每日 1000 次。
Free with any GitHub account — dozens of models from OpenAI, Meta, Mistral, and more. 有 GitHub 帳號就免費——OpenAI、Meta、Mistral 等數十種模型。
openai/gpt-4o-miniGITHUB_MODELS_TOKENUse the name GITHUB_MODELS_TOKEN (not GITHUB_TOKEN) so it won’t clash with git or the gh CLI.
用 GITHUB_MODELS_TOKEN 這個名稱（別用 GITHUB_TOKEN），才不會撞到 git 或 gh CLI。
The model runs on your own computer. Zero data leaves your machine — no key needed. 模型在你自己的電腦上跑。零資料外送——不需要金鑰。
llama3.2After installing, run ollama pull llama3.2 once to download the model.
安裝後執行一次 ollama pull llama3.2 下載模型。
Any OpenAI-style endpoint you already use — DeepSeek direct, a local gateway, your lab’s server. 任何你已在用的 OpenAI 相容端點——DeepSeek 官方端點、本地閘道、實驗室伺服器。
https://api.deepseek.comdeepseek-chatLLM_API_KEYWant a multi-vendor free tier instead? See the OpenRouter card above. 想要一個能通吃多家廠商的免費方案?見上方 OpenRouter 卡片。
askLLM never asks you to paste a key into jamovi — it reads it from your machine, so it’s never saved into your .omv file. Replace NVIDIA_API_KEY below with the env var from your provider card.
askLLM 不會要你把金鑰貼進 jamovi——它從你的電腦讀取，所以金鑰絕不會被存進 .omv 檔。把下面的 NVIDIA_API_KEY 換成你那張卡的環境變數名。
setx NVIDIA_API_KEY "nvapi-xxxxxxxxxxxxxxxx"~/.Renviron in a plain-text editor and add one line:
用純文字編輯器開啟（或新建）~/.Renviron，加入一行：
NVIDIA_API_KEY=nvapi-xxxxxxxxxxxxxxxx=).
存檔（不用引號，= 兩側不要空格）。Only summary statistics of the variables you pick — counts, means, SDs, factor frequencies — plus your question. Never the raw data rows. Your API key stays on your machine and is never written into the .omv file. For zero data leaving at all, choose Ollama.
只有你所選變數的摘要統計量（個數、平均、標準差、因子頻率）加上你的問題。絕不送原始資料列。你的金鑰留在本機，絕不寫入 .omv 檔。若要完全零外送，請選 Ollama。