Verity Mod AI Setup

Check your installed JAR. Stable 5.7.4 and test release 6.0.0-beta.9 both target Minecraft 1.20.1 with Forge. Install the dependencies required by that file, including matching GeckoLib. If Minecraft reports another missing dependency, resolve that before AI setup. A world that cannot load needs installation troubleshooting, not a new key.

CurseForge links Varmite's Full Verity JE Mod Setup | 6.0.0+ tutorial. Beta.9 reworked AI configuration. The older linked tutorials cover releases before 5.7.4, so their fields are not automatically a match. Stable 5.7.4's notes record a missing Groq model fix.

Official Java project and setup link ↗ · 5.7.4 release notes ↗ · 6.0.0-beta.9 release notes ↗ · Install Verity first →

Groq and OpenRouter need a connection, your own provider key, and an available model. Their limits and catalogs can change. Local Ollama runs a downloaded model on your computer; leave enough memory and processing capacity for Minecraft too.

The 6.0.0+ tutorial covers all three paths. Test one first. Addresses below are reference values: a base URL and a complete chat endpoint differ. Keep the address format expected by your release's tutorial.

• Cloud: create a key with the chosen provider, not a shared key from a download site.
• Local: install Ollama, download a model, and prove it can answer outside Minecraft.
• Do not enter an API key on VerityMod.link. This site neither needs nor validates it.

Varmite's official 6.0.0+ walkthrough ↗

Create a key in Groq's console and use it in your installed branch's Groq configuration. It is not an OpenRouter key. The API base address below is for comparison; a longer address in a working preset is not necessarily wrong.

Use a model available to your account. Copy its API identifier, not its marketing name. For model-not-found errors, compare Groq's model documentation with Verity's release notes. A valid key cannot restore a retired model.

1. Open Groq's console and create your own API key.
2. Check the current model list and your account's access to the selected model.
3. Follow the Groq portion of the matching Verity tutorial; enter the key privately and save the configuration.
4. Send one short text message in a test world, then wait for a response or record the exact error.

https://api.groq.com/openai/v1

Groq key and first-request documentation ↗ · Groq's current model identifiers ↗ · Groq API base address ↗

Use the OpenRouter chapter of Varmite's 6.0.0+ video; do not assume stable 5.x has the same controls. Create your key on OpenRouter and copy the full model slug from its catalog, including the provider prefix and any variant suffix.

Set a key credit limit and check whether the model is paid. Free models still have limits. For insufficient-credit errors, check both your account balance and the key's limit before changing Minecraft.

1. Create an OpenRouter key and choose a spending limit you are comfortable with.
2. Select an available text model and copy its exact model slug.
3. Use the OpenRouter settings demonstrated for your 6.x release; keep the key and model with that provider.
4. Save the configuration and perform the same short text test used for Groq.

https://openrouter.ai/api/v1

OpenRouter model and request guide ↗ · OpenRouter keys, limits, and authentication ↗ · OpenRouter model catalog ↗

Download official Ollama and open it. Download a local model that fits your machine and test it in Ollama first. The local API normally uses http://localhost:11434 without authentication. Ollama's cloud service is a different path; local inference does not need a cloud key.

Run these terminal commands. Replace YOUR_MODEL_TAG with the exact installed name from ollama ls, including its tag. YOUR_MODEL_TAG itself is not a model name. Wait for an answer, leave chat with /bye, and keep Ollama running while Minecraft uses it.

Follow the 6.x Ollama chapter with that model. Do not combine it with old proxy instructions; stable 5.x users should keep their branch's documented connection method. On a remote server, localhost means that server, not your PC.

ollama ls

ollama run YOUR_MODEL_TAG

ollama ps

Download official Ollama ↗ · Ollama command reference ↗ · Local versus cloud authentication ↗ · Ollama model download and first reply ↗

Use a disposable single-player world and your release's demonstrated interaction method. Ask Hello, Verity. Wait instead of repeatedly sending messages; cloud limits or local model loading can make rapid retries misleading.

Success means a relevant text reply, not just a visible character. Note the working version, provider, and model without recording your key. If the provider works independently but Verity does not, recheck the saved settings and version boundary.

Speech-to-text handles your microphone; text-to-speech speaks the answer. They are separate from the text model. Follow your release's voice settings, check input and output devices, and test one direction at a time. Leave optional voice add-ons out initially.

The project says AI supports multiple languages, but speech-to-text and text-to-speech have only been tested in English. Try a simple English phrase. When text works but sound does not, preserve the working AI settings and troubleshoot audio.

Project language and speech notes ↗

These are starting checks, not guaranteed diagnoses. Save the exact error and the action that triggered it. Redact API keys, tokens, private endpoints, and personal details before sharing a screenshot or log. If a key has been exposed, revoke it with the provider and create a replacement privately.

Full Verity Mod troubleshooting checklist → · Groq error explanations ↗ · OpenRouter error explanations ↗