Garry's Opinionated OpenClaw/Hermes Agent Brain
TypeScript
import type { Recipe } from '../types.ts';
/**
* NVIDIA NIM / API Catalog exposes OpenAI-compatible /v1/chat/completions
* and /v1/embeddings APIs.
*
* Retrieval models use asymmetric encoding. The gateway maps gbrain's
* document/query distinction to NVIDIA's wire values:
* document -> input_type: passage
* query -> input_type: query
*
* The model ids below intentionally keep NVIDIA's full catalog ids because
* the hosted endpoint expects values like `nvidia/nv-embedqa-e5-v5` in the
* request body. Short aliases are provided for CLI ergonomics.
*/
export const nvidia: Recipe = {
id: 'nvidia',
name: 'NVIDIA NIM',
tier: 'openai-compat',
implementation: 'openai-compatible',
// No shipped rate for these yet: they warn and run under a default cap; a
// user cap refuses them until `gbrain pricing set` registers one (X1 guard).
unpriced_models: [
'nvidia/nemotron-3-super-120b-a12b',
'nvidia/nv-embedqa-e5-v5',
'nvidia/llama-nemotron-embed-1b-v2',
'nvidia/nemotron-3-embed-8b',
'nvidia/nv-embed-v1',
'nvidia/nv-embedcode-7b-v1',
],
base_url_default: 'https://integrate.api.nvidia.com/v1',
auth_env: {
required: ['NVIDIA_API_KEY'],
setup_url: 'https://build.nvidia.com',
},
aliases: {
'nv-embedqa-e5-v5': 'nvidia/nv-embedqa-e5-v5',
'llama-nemotron-embed-1b-v2': 'nvidia/llama-nemotron-embed-1b-v2',
'nemotron-3-embed-8b': 'nvidia/nemotron-3-embed-8b',
'nemotron-3-super': 'nvidia/nemotron-3-super-120b-a12b',
'nemotron-3-super-120b-a12b': 'nvidia/nemotron-3-super-120b-a12b',
'nv-embed-v1': 'nvidia/nv-embed-v1',
'nv-embedcode-7b-v1': 'nvidia/nv-embedcode-7b-v1',
},
// No resolveAuth override: NVIDIA is plain `Authorization: Bearer <key>`,
// which defaultResolveAuth derives from auth_env.required. IRON RULE
// (test/ai/recipes-existing-regression.test.ts): only Azure overrides
// resolveAuth.
touchpoints: {
chat: {
models: [
'nvidia/nemotron-3-super-120b-a12b',
],
supports_tools: false,
supports_subagent_loop: false,
// Do not treat Nemotron as a Minions subagent driver until tool-calling
// and replay stability are proven through a separate adapter test.
max_context_tokens: 128000,
price_last_verified: '2026-05-24',
},
embedding: {
models: [
'nvidia/nv-embedqa-e5-v5',
'nvidia/llama-nemotron-embed-1b-v2',
'nvidia/nemotron-3-embed-8b',
'nvidia/nv-embed-v1',
'nvidia/nv-embedcode-7b-v1',
],
// PR #5921 (@abudhi19): Nemotron-3-Embed-8B answers 4096 wide whatever `dimensions` asks; its card documents
// 2048/1024 as "keep the first coordinates, then L2-normalize", which the gateway applies (fitMatryoshkaPrefix).
// Keys are folded ids; the Hugging Face / vLLM served name carries a -bf16 suffix.
matryoshka: {
'nvidia/nemotron-3-embed-8b': { native: 4096, prefixes: [1024, 2048] },
'nvidia/nemotron-3-embed-8b-bf16': { native: 4096, prefixes: [1024, 2048] },
},
// Default to the lightest tested hosted model. Larger NVIDIA models are
// supported via explicit embedding_dimensions (2048 or 4096).
default_dims: 1024,
dims_options: [1024, 2048, 4096],
// #4530: hard per-INPUT token caps (distinct from the batch budget).
// nv-embedqa-e5-v5 is an e5-family encoder with the original 512-token
// pre-training context; the hosted endpoint rejects any single input
// over it with a non-transient 400 ("Input length N exceeds maximum
// allowed token size 512"). Declaring it here makes the chunker split
// chunks to fit (resolveMaxChunkTokens). Only confirmed limits are
// declared — models absent from the map keep the default cap.
max_input_tokens: {
'nvidia/nv-embedqa-e5-v5': 512,
},
// Conservative split; hosted NVIDIA embedding endpoints require
// input_type and may reject large payloads before tokenizing.
max_batch_tokens: 8192,
chars_per_token: 4,
safety_factor: 0.75,
cost_per_1m_tokens_usd: undefined,
price_last_verified: '2026-05-24',
},
},
setup_hint: 'Get an API key at https://build.nvidia.com, then `export NVIDIA_API_KEY=...`.',
};