rmerk adds NVIDIA's model catalog and makes Kimi the default

A fourth AI engine slot opens up in this fork, and the out-of-the-box brain is no longer Google's.

infrastructurechat-ui

rmerk has wired in NVIDIA's hosted model catalog as a new option alongside the providers Mike already supported, which means a single API key now unlocks Kimi K2.6, Llama 3.3, and DeepSeek R1 - a mix of open-weight and Chinese-lab models that tend to come in cheaper than the big-name US options. Streaming replies and multi-step tool use both work.

The more opinionated move is the default swap: a fresh install now boots with Kimi K2.6 selected rather than the Google model upstream picks. That makes the fork genuinely usable without a Google account, but it also means anyone who pulls these changes in and runs without their own keys could suddenly find their users' queries flowing to NVIDIA's endpoints. Worth a beat of thought before importing. There's also a half-finished edge - users can't yet bring their own NVIDIA key through the UI; it's environment-level only for now.

So what Legal-tech teams who want a non-Google default or cheaper inference economics should look here before rolling their own provider integration.

View this fork on GitHub →

Spotted something wrong? Or know the PR text has fresher detail than the writeup above?

Commits in this thread

1 commit from rmerk/mike, oldest first. Source extracted verbatim from the harvested git log.

SHA Subject Author Date

8535292c feat: add NVIDIA API Catalog provider with Kimi K2.6 as default Ryan Choi 2026-05-11 ↗ GitHub

SHA	Subject	Author	Date
`8535292c`	feat: add NVIDIA API Catalog provider with Kimi K2.6 as default	Ryan Choi	2026-05-11	↗ GitHub
commit body Adds a fourth LLM provider that targets build.nvidia.com's OpenAI-compatible Chat Completions endpoint, enabling Kimi K2.6, Llama 3.3 70B, and DeepSeek R1 through a single env key. Backend: new lib/llm/nvidia.ts handles streaming + multi-turn tool calls against integrate.api.nvidia.com/v1. ApiKeyProvider extended with "nvidia" (env-only - DB CHECK constraint still blocks per-user storage). Frontend: model picker now shows an NVIDIA group at the top with Kimi K2.6 selected by default. ApiKeyState and modelAvailability extended to recognise nvidia so the picker doesn't grey out available models. All gemini-3-flash-preview fallbacks swapped to moonshotai/kimi-k2.6 so the app boots usefully without a Gemini key.

commit body

Adds a fourth LLM provider that targets build.nvidia.com's OpenAI-compatible
Chat Completions endpoint, enabling Kimi K2.6, Llama 3.3 70B, and DeepSeek R1
through a single env key.

Backend: new lib/llm/nvidia.ts handles streaming + multi-turn tool calls
against integrate.api.nvidia.com/v1. ApiKeyProvider extended with "nvidia"
(env-only - DB CHECK constraint still blocks per-user storage).

Frontend: model picker now shows an NVIDIA group at the top with Kimi K2.6
selected by default. ApiKeyState and modelAvailability extended to recognise
nvidia so the picker doesn't grey out available models.

All gemini-3-flash-preview fallbacks swapped to moonshotai/kimi-k2.6 so the
app boots usefully without a Gemini key.

Capture this thread into my fork

Download a single Markdown prompt that tells Claude how to port every commit above into your working tree — adapting paths and structure to match your repo. Run it via claude -p < capture-thread-417.md from inside the repo you want the changes in.

⬇ Download capture-thread-417.md