feat: add NVIDIA NIM to CLI provider selection wizards

Add NVIDIA NIM as a selectable provider in both the setup wizard
and init wizard CLI screens, using NVIDIA_API_KEY env var and
meta/llama-3.3-70b-instruct as the default model.

Closes #787

Co-authored-by: ilteoood <6383527+ilteoood@users.noreply.github.com>
Agent-Logs-Url: https://github.com/ilteoood/openfang/sessions/541bf2ad-f8d8-488c-84a2-4ca71e66730f

.env.example

@@ -32,9 +32,6 @@
 # Fireworks AI
 # FIREWORKS_API_KEY=...
 
-# NVIDIA NIM
-# NVIDIA_API_KEY=nvapi-...
-
 # ─── Local LLM Providers (no API key needed) ─────────────────────────
 
 # Ollama (default: http://localhost:11434)

.gitignore

@@ -45,4 +45,3 @@ Thumbs.db
 *.swo
 *~
 .serena/
-crates/openfang-desktop/gen/schemas/linux-schema.json

Cargo.lock

@@ -3812,7 +3812,7 @@ dependencies = [
 
 [[package]]
 name = "openfang-api"
-version = "0.5.1"
+version = "0.4.9"
 dependencies = [
  "async-trait",
  "axum",
@@ -3852,7 +3852,7 @@ dependencies = [
 
 [[package]]
 name = "openfang-channels"
-version = "0.5.1"
+version = "0.4.9"
 dependencies = [
  "aes",
  "async-trait",
@@ -3889,7 +3889,7 @@ dependencies = [
 
 [[package]]
 name = "openfang-cli"
-version = "0.5.1"
+version = "0.4.9"
 dependencies = [
  "clap",
  "clap_complete",
@@ -3916,7 +3916,7 @@ dependencies = [
 
 [[package]]
 name = "openfang-desktop"
-version = "0.5.1"
+version = "0.4.9"
 dependencies = [
  "axum",
  "open",
@@ -3942,7 +3942,7 @@ dependencies = [
 
 [[package]]
 name = "openfang-extensions"
-version = "0.5.1"
+version = "0.4.9"
 dependencies = [
  "aes-gcm",
  "argon2",
@@ -3970,7 +3970,7 @@ dependencies = [
 
 [[package]]
 name = "openfang-hands"
-version = "0.5.1"
+version = "0.4.9"
 dependencies = [
  "chrono",
  "dashmap",
@@ -3987,7 +3987,7 @@ dependencies = [
 
 [[package]]
 name = "openfang-kernel"
-version = "0.5.1"
+version = "0.4.9"
 dependencies = [
  "async-trait",
  "chrono",
@@ -4025,7 +4025,7 @@ dependencies = [
 
 [[package]]
 name = "openfang-memory"
-version = "0.5.1"
+version = "0.4.9"
 dependencies = [
  "async-trait",
  "chrono",
@@ -4044,7 +4044,7 @@ dependencies = [
 
 [[package]]
 name = "openfang-migrate"
-version = "0.5.1"
+version = "0.4.9"
 dependencies = [
  "chrono",
  "dirs 6.0.0",
@@ -4063,7 +4063,7 @@ dependencies = [
 
 [[package]]
 name = "openfang-runtime"
-version = "0.5.1"
+version = "0.4.9"
 dependencies = [
  "anyhow",
  "async-trait",
@@ -4097,7 +4097,7 @@ dependencies = [
 
 [[package]]
 name = "openfang-skills"
-version = "0.5.1"
+version = "0.4.9"
 dependencies = [
  "chrono",
  "hex",
@@ -4120,7 +4120,7 @@ dependencies = [
 
 [[package]]
 name = "openfang-types"
-version = "0.5.1"
+version = "0.4.9"
 dependencies = [
  "async-trait",
  "chrono",
@@ -4139,7 +4139,7 @@ dependencies = [
 
 [[package]]
 name = "openfang-wire"
-version = "0.5.1"
+version = "0.4.9"
 dependencies = [
  "async-trait",
  "chrono",
@@ -8818,7 +8818,7 @@ checksum = "b9cc00251562a284751c9973bace760d86c0276c471b4be569fe6b068ee97a56"
 
 [[package]]
 name = "xtask"
-version = "0.5.1"
+version = "0.4.9"
 
 [[package]]
 name = "yoke"

crates/openfang-cli/src/tui/screens/init_wizard.rs

@@ -188,6 +188,14 @@ const PROVIDERS: &[ProviderInfo] = &[
         needs_key: true,
         hint: "",
     },
+    ProviderInfo {
+        name: "nvidia",
+        display: "NVIDIA NIM",
+        env_var: "NVIDIA_API_KEY",
+        default_model: "meta/llama-3.3-70b-instruct",
+        needs_key: true,
+        hint: "",
+    },
     ProviderInfo {
         name: "claude-code",
         display: "Claude Code",

crates/openfang-cli/src/tui/screens/wizard.rs

@@ -127,6 +127,12 @@ const PROVIDERS: &[ProviderInfo] = &[
         default_model: "codegeex-4",
         needs_key: true,
     },
+    ProviderInfo {
+        name: "nvidia",
+        env_var: "NVIDIA_API_KEY",
+        default_model: "meta/llama-3.3-70b-instruct",
+        needs_key: true,
+    },
     ProviderInfo {
         name: "claude-code",
         env_var: "",

crates/openfang-runtime/src/model_catalog.rs

@@ -869,9 +869,7 @@ fn builtin_aliases() -> HashMap<String, String> {
         ("codex-4.1", "codex/gpt-4.1"),
         ("codex-o4", "codex/o4-mini"),
         // NVIDIA NIM aliases
-        ("nemotron", "meta/llama-3.3-70b-instruct"),
-        ("nemotron-super", "nvidia/llama-3.3-nemotron-super-49b-v1.5"),
-        ("nemotron-ultra", "nvidia/llama-3.1-nemotron-ultra-253b-v1"),
+        ("nemotron", "nvidia/llama-3.1-nemotron-70b-instruct"),
         // Venice aliases
         ("venice", "venice-uncensored"),
         // Claude Code aliases
@@ -2152,50 +2150,8 @@ fn builtin_models() -> Vec<ModelCatalogEntry> {
             aliases: vec![],
         },
         // ══════════════════════════════════════════════════════════════
-        // NVIDIA NIM (8)
+        // NVIDIA NIM (5)
         // ══════════════════════════════════════════════════════════════
-        ModelCatalogEntry {
-            id: "meta/llama-3.3-70b-instruct".into(),
-            display_name: "Llama 3.3 70B Instruct (NVIDIA NIM)".into(),
-            provider: "nvidia".into(),
-            tier: ModelTier::Balanced,
-            context_window: 128_000,
-            max_output_tokens: 4_096,
-            input_cost_per_m: 0.88,
-            output_cost_per_m: 0.88,
-            supports_tools: true,
-            supports_vision: false,
-            supports_streaming: true,
-            aliases: vec![],
-        },
-        ModelCatalogEntry {
-            id: "nvidia/llama-3.3-nemotron-super-49b-v1.5".into(),
-            display_name: "Nemotron Super 49B v1.5 (NVIDIA NIM)".into(),
-            provider: "nvidia".into(),
-            tier: ModelTier::Smart,
-            context_window: 128_000,
-            max_output_tokens: 4_096,
-            input_cost_per_m: 0.88,
-            output_cost_per_m: 0.88,
-            supports_tools: true,
-            supports_vision: false,
-            supports_streaming: true,
-            aliases: vec!["nemotron-super".into()],
-        },
-        ModelCatalogEntry {
-            id: "nvidia/llama-3.1-nemotron-ultra-253b-v1".into(),
-            display_name: "Nemotron Ultra 253B (NVIDIA NIM)".into(),
-            provider: "nvidia".into(),
-            tier: ModelTier::Frontier,
-            context_window: 128_000,
-            max_output_tokens: 4_096,
-            input_cost_per_m: 4.20,
-            output_cost_per_m: 4.20,
-            supports_tools: true,
-            supports_vision: false,
-            supports_streaming: true,
-            aliases: vec!["nemotron-ultra".into()],
-        },
         ModelCatalogEntry {
             id: "nvidia/llama-3.1-nemotron-70b-instruct".into(),
             display_name: "Nemotron 70B Instruct (NVIDIA NIM)".into(),

docs/providers.md

@@ -1,6 +1,6 @@
 # LLM Providers Guide
 
-OpenFang ships with a comprehensive model catalog covering **3 native LLM drivers**, **21 providers**, **56 builtin models**, and **26 aliases**. Every provider uses one of three battle-tested drivers: the native **Anthropic** driver, the native **Gemini** driver, or the universal **OpenAI-compatible** driver. This guide is the single source of truth for configuring, selecting, and managing LLM providers in OpenFang.
+OpenFang ships with a comprehensive model catalog covering **3 native LLM drivers**, **20 providers**, **51 builtin models**, and **23 aliases**. Every provider uses one of three battle-tested drivers: the native **Anthropic** driver, the native **Gemini** driver, or the universal **OpenAI-compatible** driver. This guide is the single source of truth for configuring, selecting, and managing LLM providers in OpenFang.
 
 ---
 
@@ -549,41 +549,9 @@ For Gemini specifically, either `GEMINI_API_KEY` or `GOOGLE_API_KEY` will work.
 
 ---
 
-### 21. NVIDIA NIM
-
-| | |
-|---|---|
-| **Display Name** | NVIDIA NIM |
-| **Driver** | OpenAI-compatible |
-| **Env Var** | `NVIDIA_API_KEY` |
-| **Base URL** | `https://integrate.api.nvidia.com/v1` |
-| **Key Required** | Yes |
-| **Free Tier** | Yes (rate-limited) |
-| **Auth** | `Authorization: Bearer` header |
-| **Models** | 8 |
-
-**Available Models:**
-- `meta/llama-3.3-70b-instruct` (Balanced) -- recommended default
-- `nvidia/llama-3.3-nemotron-super-49b-v1.5` (Smart) -- NVIDIA-tuned, 49B
-- `nvidia/llama-3.1-nemotron-ultra-253b-v1` (Frontier) -- max quality, 253B
-- `nvidia/llama-3.1-nemotron-70b-instruct` (Smart)
-- `meta/llama-3.1-405b-instruct` (Frontier)
-- `meta/llama-3.1-70b-instruct` (Balanced)
-- `mistralai/mistral-large-latest` (Smart)
-- `nvidia/nemotron-4-340b-instruct` (Frontier)
-
-**Setup:**
-1. Sign up at [build.nvidia.com](https://build.nvidia.com)
-2. Generate an API key (starts with `nvapi-`)
-3. `export NVIDIA_API_KEY="nvapi-..."`
-
-**Notes:** NVIDIA NIM provides hosted inference for open-source and NVIDIA-tuned models via an OpenAI-compatible API. Supports tool calling and streaming. Also known as `nvidia-nim`.
-
----
-
 ## Model Catalog
 
-The complete catalog of all 56 builtin models, sorted by provider. Pricing is per million tokens.
+The complete catalog of all 51 builtin models, sorted by provider. Pricing is per million tokens.
 
 | # | Model ID | Display Name | Provider | Tier | Context Window | Max Output | Input $/M | Output $/M | Tools | Vision |
 |---|----------|-------------|----------|------|---------------|------------|-----------|------------|-------|--------|
@@ -640,14 +608,6 @@ The complete catalog of all 56 builtin models, sorted by provider. Pricing is pe
 | 51 | `grok-2-mini` | Grok 2 Mini | xai | Fast | 131,072 | 32,768 | $0.30 | $0.50 | Yes | No |
 | 52 | `hf/meta-llama/Llama-3.3-70B-Instruct` | Llama 3.3 70B (HF) | huggingface | Balanced | 128,000 | 4,096 | $0.30 | $0.30 | No | No |
 | 53 | `replicate/meta-llama-3.3-70b-instruct` | Llama 3.3 70B (Replicate) | replicate | Balanced | 128,000 | 4,096 | $0.40 | $0.40 | No | No |
-| 54 | `meta/llama-3.3-70b-instruct` | Llama 3.3 70B Instruct (NVIDIA NIM) | nvidia | Balanced | 128,000 | 4,096 | $0.88 | $0.88 | Yes | No |
-| 55 | `nvidia/llama-3.3-nemotron-super-49b-v1.5` | Nemotron Super 49B v1.5 (NVIDIA NIM) | nvidia | Smart | 128,000 | 4,096 | $0.88 | $0.88 | Yes | No |
-| 56 | `nvidia/llama-3.1-nemotron-ultra-253b-v1` | Nemotron Ultra 253B (NVIDIA NIM) | nvidia | Frontier | 128,000 | 4,096 | $4.20 | $4.20 | Yes | No |
-| 57 | `nvidia/llama-3.1-nemotron-70b-instruct` | Nemotron 70B Instruct (NVIDIA NIM) | nvidia | Smart | 128,000 | 4,096 | $0.88 | $0.88 | Yes | No |
-| 58 | `meta/llama-3.1-405b-instruct` | Llama 3.1 405B Instruct (NVIDIA NIM) | nvidia | Frontier | 128,000 | 4,096 | $5.00 | $16.00 | Yes | No |
-| 59 | `meta/llama-3.1-70b-instruct` | Llama 3.1 70B Instruct (NVIDIA NIM) | nvidia | Balanced | 128,000 | 4,096 | $0.88 | $0.88 | Yes | No |
-| 60 | `mistralai/mistral-large-latest` | Mistral Large (NVIDIA NIM) | nvidia | Smart | 128,000 | 4,096 | $2.00 | $6.00 | Yes | No |
-| 61 | `nvidia/nemotron-4-340b-instruct` | Nemotron 4 340B Instruct (NVIDIA NIM) | nvidia | Frontier | 4,096 | 4,096 | $4.20 | $4.20 | Yes | No |
 
 **Model Tiers:**
 
@@ -667,7 +627,7 @@ The complete catalog of all 56 builtin models, sorted by provider. Pricing is pe
 
 ## Model Aliases
 
-All 26 aliases resolve to canonical model IDs. Aliases are case-insensitive.
+All 23 aliases resolve to canonical model IDs. Aliases are case-insensitive.
 
 | Alias | Resolves To |
 |-------|------------|
@@ -694,9 +654,6 @@ All 26 aliases resolve to canonical model IDs. Aliases are case-insensitive.
 | `sonar` | `sonar-pro` |
 | `jamba` | `jamba-1.5-large` |
 | `command-r` | `command-r-plus` |
-| `nemotron` | `meta/llama-3.3-70b-instruct` |
-| `nemotron-super` | `nvidia/llama-3.3-nemotron-super-49b-v1.5` |
-| `nemotron-ultra` | `nvidia/llama-3.1-nemotron-ultra-253b-v1` |
 
 You can use aliases anywhere a model ID is accepted: in config files, REST API calls, chat commands, and the model routing configuration.
 
@@ -1090,7 +1047,6 @@ Quick reference for all provider environment variables:
 | Hugging Face | `HF_API_KEY` | Yes |
 | xAI | `XAI_API_KEY` | Yes |
 | Replicate | `REPLICATE_API_TOKEN` | Yes |
-| NVIDIA NIM | `NVIDIA_API_KEY` | Yes |
 
 ---