@agxnte/reidx 2.0.9 → 2.1.0

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@agxnte/reidx",
-  "version": "2.0.9",
+  "version": "2.1.0",
   "description": "Terminal-native personal intelligence and coding CLI (agent-first runtime)",
   "bin": {
     "reid": "bin/reidx.js"

package/pyproject.toml

@@ -1,7 +1,7 @@
 [project]
 name = "reidx"
-version = "2.0.9"
-description = "Terminal-native personal intelligence and coding CLI (agent-first runtime)"
+version = "2.1.0"
+description = "Terminal-native personal intelligence and coding CLI (agent-first runtime)"
 readme = "README.md"
 requires-python = ">=3.12"
 license = { text = "MIT" }

package/src/reidx/__init__.py

@@ -1,3 +1,3 @@
 """ReidX: terminal-native agent-first CLI runtime."""
 
-__version__ = "2.0.9"
+__version__ = "2.1.0"

package/src/reidx/provider/openai.py

@@ -2,7 +2,7 @@
 
 Speaks the OpenAI Chat Completions API (POST /v1/chat/completions). The same
 wire format is spoken by llama.cpp's server (`/v1/chat/completions`) and by
-LM Studio's local server — see `OpenAICompatibleProvider` below for the
+LM Studio's local server - see `OpenAICompatibleProvider` below for the
 local-endpoint variant that skips the auth header when no key is set.
 
 Tool schemas ReidX produces (`ToolBase.schema()`) are already OpenAI-style,
@@ -24,6 +24,7 @@ class OpenAIProvider(BaseProvider):
     name = "openai"
     DEFAULT_BASE_URL = "https://api.openai.com"
     DEFAULT_MODEL = "gpt-4o-mini"
+    MODELS_PATH = "/v1/models"
 
     def __init__(
         self,
@@ -126,7 +127,7 @@ class OpenAIProvider(BaseProvider):
         return self._parse(body, model)
 
     def fetch_models(self) -> list[str]:
-        url = f"{self.base_url}/v1/models"
+        url = f"{self.base_url}{self.MODELS_PATH}"
         body = get_json(url, self._headers())
         models: list[str] = []
         for item in body.get("data", []):
@@ -148,6 +149,7 @@ class OpenAICompatibleProvider(OpenAIProvider):
     name = "openai-compatible"
     DEFAULT_BASE_URL = "http://localhost:8080"
     DEFAULT_MODEL = "local"
+    MODELS_PATH = "/v1/models"
 
     def __init__(
         self,
@@ -174,4 +176,4 @@ class OpenAICompatibleProvider(OpenAIProvider):
 
 def _json_dump(obj) -> str:  # type: ignore[no-untyped-def]
     import json
-    return json.dumps(obj, ensure_ascii=False)
+    return json.dumps(obj, ensure_ascii=False)

package/src/reidx/provider/store.py

@@ -153,6 +153,8 @@ def validate_provider(
         msg = str(exc)
         if msg.startswith("HTTP 401") or msg.startswith("HTTP 403"):
             return False, f"authentication failed ({msg})"
+        if msg.startswith("HTTP 404"):
+            return False, f"endpoint not found - check base URL ({msg})"
         if msg.startswith("HTTP "):
             return False, f"provider error: {msg}"
         return False, f"{msg} (use --skip-verify to save anyway)"

package/src/reidx/provider_manager/catalog.py

@@ -1,3 +1,10 @@
+"""Provider catalog: canonical metadata for known providers.
+
+All OpenAI-compatible providers use `base_url` without a `/v1` suffix -
+the client appends `/v1` when building request URLs. For Anthropic, the
+client uses the base URL as-is with `/v1/messages` and `/v1/models`.
+Ollama uses its native `/api` endpoints.
+"""
 from __future__ import annotations
 
 from dataclasses import dataclass, field
@@ -22,276 +29,277 @@ _BUILTIN: list[ProviderDefinition] = [
     ProviderDefinition(
         id="openai", name="OpenAI", description="GPT-4o, o-series, and DALL-E models",
         kind="openai", base_url="https://api.openai.com", default_model="gpt-4o-mini",
-        aliases=["gpt", "chatgpt"], popular=True, icon="🟢",
+        aliases=["gpt", "chatgpt"], popular=True, icon="OAI",
     ),
     ProviderDefinition(
         id="anthropic", name="Anthropic", description="Claude Sonnet, Opus, and Haiku models",
         kind="anthropic", base_url="https://api.anthropic.com", default_model="claude-sonnet-4-20250514",
-        aliases=["claude"], auth_method="x-api-key", popular=True, icon="🟠",
+        aliases=["claude"], auth_method="x-api-key", popular=True, icon="ANT",
     ),
     ProviderDefinition(
         id="google-ai", name="Google AI", description="Gemini Pro, Flash, and Flash-Lite models",
         kind="openai-compatible", base_url="https://generativelanguage.googleapis.com/v1beta/openai",
         default_model="gemini-2.0-flash", aliases=["gemini", "google", "google-gemini"],
-        popular=True, icon="🔵",
+        popular=True, icon="GAI",
     ),
-    ProviderDefinition(
+ProviderDefinition(
         id="openrouter", name="OpenRouter", description="Unified API for 300+ models from all providers",
-        kind="openai-compatible", base_url="https://openrouter.ai/api/v1",
-        default_model="auto", aliases=["or"], popular=True, icon="🔀",
+        kind="openai-compatible", base_url="https://openrouter.ai/api",
+        default_model="auto", aliases=["or"], popular=True, icon="ORT",
     ),
     ProviderDefinition(
         id="zai", name="Z.ai", description="GLM-4 and GLM-4.5 series models",
         kind="openai-compatible", base_url="https://open.bigmodel.cn/api/paas/v4",
-        default_model="glm-4", aliases=["z-ai", "zhipu", "glm", "bigmodel"], popular=True, icon="🟣",
+        default_model="glm-4", aliases=["z-ai", "zhipu", "glm", "bigmodel"], popular=True, icon="ZAI",
     ),
     ProviderDefinition(
         id="agentrouter", name="AgentRouter", description="Multi-model routing for AI agents",
         kind="openai-compatible", base_url="https://api.agentrouter.io/v1",
-        default_model="auto", aliases=["agent-router"], icon="🤖",
+        default_model="auto", aliases=["agent-router"], icon="AGT",
     ),
     ProviderDefinition(
         id="together", name="Together AI", description="Open-source Llama, Mixtral, and Qwen models",
         kind="openai-compatible", base_url="https://api.together.xyz/v1",
         default_model="meta-llama/Llama-3.3-70B-Instruct-Turbo", aliases=["together-ai"],
-        popular=True, icon="🤝",
+        popular=True, icon="TGT",
     ),
     ProviderDefinition(
         id="fireworks", name="Fireworks AI", description="Fast inference for open-source models",
         kind="openai-compatible", base_url="https://api.fireworks.ai/inference/v1",
         default_model="accounts/fireworks/models/llama-v3p1-70b-instruct",
-        aliases=["fireworks-ai"], popular=True, icon="🎆",
+        aliases=["fireworks-ai"], popular=True, icon="FRW",
     ),
     ProviderDefinition(
         id="groq", name="Groq", description="Ultra-fast LPU inference for Llama and Mixtral",
         kind="openai-compatible", base_url="https://api.groq.com/openai/v1",
-        default_model="llama-3.3-70b-versatile", aliases=[], popular=True, icon="⚡",
+        default_model="llama-3.3-70b-versatile", aliases=[], popular=True, icon="GRQ",
     ),
     ProviderDefinition(
         id="deepinfra", name="DeepInfra", description="Cost-effective inference for open models",
         kind="openai-compatible", base_url="https://api.deepinfra.com/v1/openai",
-        default_model="meta-llama/Meta-Llama-3.1-70B-Instruct", aliases=[], icon="🔧",
+        default_model="meta-llama/Meta-Llama-3.1-70B-Instruct", aliases=[], icon="DPI",
     ),
     ProviderDefinition(
         id="cerebras", name="Cerebras", description="Fastest AI inference with Wafer-Scale Engines",
         kind="openai-compatible", base_url="https://api.cerebras.ai/v1",
-        default_model="llama3.1-70b", aliases=[], icon="🧠",
+        default_model="llama3.1-70b", aliases=[], icon="CER",
     ),
     ProviderDefinition(
         id="cohere", name="Cohere", description="Command R+ and Aya models for enterprise",
         kind="openai-compatible", base_url="https://api.cohere.ai/v1",
-        default_model="command-r-plus-08-2024", aliases=[], icon="🔗",
+        default_model="command-r-plus-08-2024", aliases=[], icon="COH",
     ),
     ProviderDefinition(
         id="xai", name="xAI", description="Grok models from xAI",
         kind="openai-compatible", base_url="https://api.x.ai/v1",
-        default_model="grok-2-latest", aliases=["grok", "x-ai"], popular=True, icon="✖️",
+        default_model="grok-2-latest", aliases=["grok", "x-ai"], popular=True, icon="XAI",
     ),
     ProviderDefinition(
         id="mistral", name="Mistral", description="Mistral Large, Codestral, and Pixtral models",
         kind="openai-compatible", base_url="https://api.mistral.ai/v1",
-        default_model="mistral-large-latest", aliases=["mistral-ai"], popular=True, icon="🌬️",
+        default_model="mistral-large-latest", aliases=["mistral-ai"], popular=True, icon="MIS",
     ),
     ProviderDefinition(
         id="perplexity", name="Perplexity", description="Sonar models with built-in web search",
         kind="openai-compatible", base_url="https://api.perplexity.ai",
-        default_model="llama-3.1-sonar-large-128k-online", aliases=[], icon="🔮",
+        default_model="llama-3.1-sonar-large-128k-online", aliases=[], icon="PPL",
     ),
     ProviderDefinition(
-        id="ollama", name="Ollama", description="Run models locally — Llama, Qwen, Phi, and more",
+        id="ollama", name="Ollama", description="Run models locally Llama, Qwen, Phi, and more",
         kind="ollama", base_url="http://localhost:11434",
-        default_model="llama3.2", aliases=[], auth_method="none", popular=True, icon="🐪",
+        default_model="llama3.2", aliases=[], auth_method="none", popular=True, icon="OLL",
     ),
     ProviderDefinition(
         id="lm-studio", name="LM Studio", description="Local model server with OpenAI-compatible API",
         kind="openai-compatible", base_url="http://localhost:1234/v1",
-        default_model="local", aliases=["lmstudio"], auth_method="none", icon="🏠",
+        default_model="local", aliases=["lmstudio"], auth_method="none", icon="LMS",
     ),
     ProviderDefinition(
         id="vllm", name="vLLM", description="High-throughput local inference server",
         kind="openai-compatible", base_url="http://localhost:8000/v1",
-        default_model="auto", aliases=[], auth_method="none", icon="🚀",
+        default_model="auto", aliases=[], auth_method="none", icon="VLL",
     ),
     ProviderDefinition(
         id="azure-openai", name="Azure OpenAI", description="OpenAI models via Azure cloud (set base URL to your endpoint)",
         kind="openai", base_url="https://api.openai.com", default_model="gpt-4o-mini",
-        aliases=["azure"], icon="☁️",
+        aliases=["azure"], icon="AZO",
     ),
     ProviderDefinition(
         id="huggingface", name="HuggingFace", description="Inference API for 100k+ models",
         kind="openai-compatible", base_url="https://api-inference.huggingface.co/v1",
-        default_model="meta-llama/Meta-Llama-3.1-70B-Instruct", aliases=["hf"], icon="🤗",
+        default_model="meta-llama/Meta-Llama-3.1-70B-Instruct", aliases=["hf"], icon="HF",
     ),
     ProviderDefinition(
         id="sambanova", name="SambaNova", description="Fast RDU-based inference for open models",
         kind="openai-compatible", base_url="https://api.sambanova.ai/v1",
-        default_model="Meta-Llama-3.1-70B-Instruct", aliases=[], icon="🔴",
+        default_model="Meta-Llama-3.1-70B-Instruct", aliases=[], icon="SMB",
     ),
     ProviderDefinition(
         id="novita", name="Novita", description="Affordable API for 100+ open models",
         kind="openai-compatible", base_url="https://api.novita.ai/v3/openai",
-        default_model="meta-llama/llama-3.1-70b-instruct", aliases=[], icon="💎",
+        default_model="meta-llama/llama-3.1-70b-instruct", aliases=[], icon="NVT",
     ),
     ProviderDefinition(
         id="featherless", name="Featherless", description="Serverless inference for open models",
         kind="openai-compatible", base_url="https://api.featherless.ai/v1/openai",
-        default_model="meta-llama/Meta-Llama-3.1-70B-Instruct", aliases=[], icon="🪶",
+        default_model="meta-llama/Meta-Llama-3.1-70B-Instruct", aliases=[], icon="FTL",
     ),
     ProviderDefinition(
         id="chutes", name="Chutes", description="GPU-accelerated inference for AI workloads",
         kind="openai-compatible", base_url="https://api.chutes.ai/v1",
-        default_model="deepseek-ai/DeepSeek-V3", aliases=[], icon="🎯",
+        default_model="deepseek-ai/DeepSeek-V3", aliases=[], icon="CHT",
     ),
     ProviderDefinition(
         id="aimlapi", name="AI/ML API", description="Unified API for 200+ AI models",
         kind="openai-compatible", base_url="https://api.aimlapi.com/v1",
-        default_model="gpt-4o", aliases=["ai-ml-api"], icon="📊",
+        default_model="gpt-4o", aliases=["ai-ml-api"], icon="AML",
     ),
     ProviderDefinition(
         id="hyperbolic", name="Hyperbolic", description="GPU marketplace for AI inference",
         kind="openai-compatible", base_url="https://api.hyperbolic.xyz/v1",
-        default_model="meta-llama/Meta-Llama-3.1-70B-Instruct", aliases=[], icon="📈",
+        default_model="meta-llama/Meta-Llama-3.1-70B-Instruct", aliases=[], icon="HYP",
     ),
     ProviderDefinition(
         id="deepseek", name="DeepSeek", description="DeepSeek-V3 and DeepSeek-R1 reasoning models",
         kind="openai-compatible", base_url="https://api.deepseek.com/v1",
-        default_model="deepseek-chat", aliases=[], popular=True, icon="🐳",
+        default_model="deepseek-chat", aliases=[], popular=True, icon="DSK",
     ),
     ProviderDefinition(
         id="moonshot", name="Moonshot", description="Kimi K1.5 and Moonshot models",
         kind="openai-compatible", base_url="https://api.moonshot.cn/v1",
-        default_model="moonshot-v1-8k", aliases=["kimi"], icon="🌙",
+        default_model="moonshot-v1-8k", aliases=["kimi"], icon="MSH",
     ),
     ProviderDefinition(
         id="voyage-ai", name="Voyage AI", description="Embeddings and reranking models",
         kind="openai-compatible", base_url="https://api.voyageai.com/v1",
-        default_model="voyage-large-2", aliases=[], icon="🧭",
+        default_model="voyage-large-2", aliases=[], icon="VOY",
     ),
     ProviderDefinition(
         id="nvidia-nim", name="NVIDIA NIM", description="NVIDIA inference microservices for open models",
         kind="openai-compatible", base_url="https://integrate.api.nvidia.com/v1",
-        default_model="meta/llama-3.1-70b-instruct", aliases=["nvidia"], icon="💚",
+        default_model="meta/llama-3.1-70b-instruct", aliases=["nvidia"], icon="NVD",
     ),
     ProviderDefinition(
         id="ai21", name="AI21 Labs", description="Jamba models with long context windows",
         kind="openai-compatible", base_url="https://api.ai21.com/v1",
-        default_model="jamba-1.5-large", aliases=[], icon="🃏",
+        default_model="jamba-1.5-large", aliases=[], icon="A21",
     ),
     ProviderDefinition(
         id="databricks", name="Databricks", description="Serving endpoints for fine-tuned models (set base URL to your workspace)",
         kind="openai-compatible", base_url="https://dbc-xxxxxxxx.cloud.databricks.com/serving-endpoints",
         default_model="databricks-dbrx-instruct",
-        aliases=[], icon="🧱",
+        aliases=[], icon="DBS",
     ),
     ProviderDefinition(
         id="replicate", name="Replicate", description="Run open-source models in the cloud",
         kind="openai-compatible", base_url="https://api.replicate.com/v1",
-        default_model="meta/llama-2-70b-chat", aliases=[], icon="🔁",
+        default_model="meta/llama-2-70b-chat", aliases=[], icon="RPL",
     ),
     ProviderDefinition(
         id="siliconflow", name="SiliconFlow", description="Cost-effective inference for 200+ models",
         kind="openai-compatible", base_url="https://api.siliconflow.cn/v1",
-        default_model="deepseek-ai/DeepSeek-V3", aliases=[], icon="🌊",
+        default_model="deepseek-ai/DeepSeek-V3", aliases=[], icon="SFL",
     ),
     ProviderDefinition(
         id="nebius", name="Nebius", description="Cloud GPU platform for AI inference",
         kind="openai-compatible", base_url="https://api.studio.nebius.ai/v1",
-        default_model="meta-llama/Meta-Llama-3.1-70B-Instruct", aliases=[], icon="☁️",
+        default_model="meta-llama/Meta-Llama-3.1-70B-Instruct", aliases=[], icon="NEB",
     ),
     ProviderDefinition(
         id="lepton", name="Lepton AI", description="Serverless inference with auto-scaling",
         kind="openai-compatible", base_url="https://api.lepton.ai/v1",
-        default_model="meta-llama/Llama-3.2-70B-Instruct", aliases=[], icon="⚛️",
+        default_model="meta-llama/Llama-3.2-70B-Instruct", aliases=[], icon="LPT",
     ),
     ProviderDefinition(
         id="lambda", name="Lambda Labs", description="GPU cloud for AI training and inference",
         kind="openai-compatible", base_url="https://api.lambdalabs.com/v1",
-        default_model="meta-llama/Meta-Llama-3.1-70B-Instruct", aliases=["lambda-labs"], icon="λ",
+        default_model="meta-llama/Meta-Llama-3.1-70B-Instruct", aliases=["lambda-labs"], icon="LDA",
     ),
     ProviderDefinition(
         id="kluster", name="Kluster AI", description="Distributed GPU network for inference",
         kind="openai-compatible", base_url="https://api.kluster.ai/v1",
-        default_model="meta-llama/Meta-Llama-3.1-70B-Instruct", aliases=[], icon="🔗",
+        default_model="meta-llama/Meta-Llama-3.1-70B-Instruct", aliases=[], icon="KLU",
     ),
     ProviderDefinition(
         id="upstage", name="Upstage", description="Solar Mini and Pro models for Korean and English",
         kind="openai-compatible", base_url="https://api.upstage.ai/v1",
-        default_model="solar-pro", aliases=[], icon="☀️",
+        default_model="solar-pro", aliases=[], icon="UPS",
     ),
     ProviderDefinition(
         id="yi", name="Yi (01.AI)", description="Yi-Large and Yi-34B models",
         kind="openai-compatible", base_url="https://api.01.ai/v1",
-        default_model="yi-large", aliases=["01-ai", "yi-ai"], icon="🎯",
+        default_model="yi-large", aliases=["01-ai", "yi-ai"], icon="YI",
     ),
     ProviderDefinition(
         id="volcengine", name="Volcengine", description="ByteDance's Ark platform for Doubao models",
         kind="openai-compatible", base_url="https://ark.cn-beijing.volces.com/api/v3",
-        default_model="doubao-pro-32k", aliases=["ark", "bytedance"], icon="🌋",
+        default_model="doubao-pro-32k", aliases=["ark", "bytedance"], icon="VOL",
     ),
     ProviderDefinition(
         id="hunyuan", name="Tencent Hunyuan", description="Tencent's Hunyuan large language models",
         kind="openai-compatible", base_url="https://api.hunyuan.cloud.tencent.com/v1",
-        default_model="hunyuan-turbos-latest", aliases=["tencent"], icon="🐧",
+        default_model="hunyuan-turbos-latest", aliases=["tencent"], icon="HUN",
     ),
     ProviderDefinition(
         id="cloudflare-ai", name="Cloudflare Workers AI", description="Serverless AI inference at the edge",
         kind="openai-compatible", base_url="https://api.cloudflare.com/client/v4",
         default_model="@cf/meta/llama-3.1-70b-instruct",
-        aliases=["cf-ai"], icon="orange",
+        aliases=["cf-ai"], icon="CFL",
     ),
     ProviderDefinition(
         id="monsterapi", name="Monster API", description="Affordable GPU inference for open models",
         kind="openai-compatible", base_url="https://api.monsterapi.ai/v1",
-        default_model="meta-llama/Meta-Llama-3.1-70B-Instruct", aliases=[], icon="👾",
+        default_model="meta-llama/Meta-Llama-3.1-70B-Instruct", aliases=[], icon="MST",
     ),
     ProviderDefinition(
         id="baseten", name="Baseten", description="Model deployment and serving platform",
         kind="openai-compatible", base_url="https://inference.baseten.co/v1",
-        default_model="meta-llama/Meta-Llama-3.1-70B-Instruct", aliases=[], icon="🏗️",
+        default_model="meta-llama/Meta-Llama-3.1-70B-Instruct", aliases=[], icon="BST",
     ),
     ProviderDefinition(
         id="anyscale", name="Anyscale", description="Ray-based scalable model serving",
         kind="openai-compatible", base_url="https://api.endpoints.anyscale.com/v1",
-        default_model="meta-llama/Llama-3.1-70B-Instruct", aliases=[], icon="📐",
+        default_model="meta-llama/Llama-3.1-70B-Instruct", aliases=[], icon="ANY",
     ),
     ProviderDefinition(
         id="jan", name="Jan", description="Local AI model server with web UI",
         kind="openai-compatible", base_url="http://localhost:1337/v1",
-        default_model="local", aliases=[], auth_method="none", icon="🖥️",
+        default_model="local", aliases=[], auth_method="none", icon="JAN",
     ),
     ProviderDefinition(
         id="llamacpp", name="Llama.cpp Server", description="Local C++ inference server",
         kind="openai-compatible", base_url="http://localhost:8080",
-        default_model="local", aliases=["llama-cpp"], auth_method="none", icon="🦙",
+        default_model="local", aliases=["llama-cpp"], auth_method="none", icon="LPP",
     ),
     ProviderDefinition(
         id="watsonx", name="IBM Watsonx", description="Enterprise AI platform with Granite models",
         kind="openai-compatible", base_url="https://us-south.ml.cloud.ibm.com",
         default_model="ibm/granite-3-8b-instruct",
-        aliases=["ibm"], icon="🔷",
+        aliases=["ibm"], icon="IBM",
     ),
     ProviderDefinition(
-        id="aleph-alpha", name="Aleph Alpha", description="European sovereign AI — Luminous models",
+        id="aleph-alpha", name="Aleph Alpha", description="European sovereign AI Luminous models",
         kind="openai-compatible", base_url="https://api.aleph-alpha.eu/v1",
-        default_model="luminous-supreme-control", aliases=[], icon="α",
+        default_model="luminous-supreme-control", aliases=[], icon="ALP",
     ),
     ProviderDefinition(
         id="predibase", name="Predibase", description="Fine-tune and serve open-source LLMs",
         kind="openai-compatible", base_url="https://serving.app.predibase.com/v1",
-        default_model="meta-llama/Meta-Llama-3.1-8B-Instruct", aliases=[], icon="🎯",
+        default_model="meta-llama/Meta-Llama-3.1-8B-Instruct", aliases=[], icon="PDB",
     ),
     ProviderDefinition(
         id="gravity", name="Gravity API", description="Multi-model API gateway",
         kind="openai-compatible", base_url="https://api.gravityapi.com/v1",
-        default_model="auto", aliases=[], icon="🌌",
+        default_model="auto", aliases=[], icon="GRV",
     ),
     ProviderDefinition(
         id="infermatic", name="Infermatic", description="Fast inference for popular open models",
         kind="openai-compatible", base_url="https://api.infermatic.ai/v1",
-        default_model="meta-llama/Meta-Llama-3.1-70B-Instruct", aliases=[], icon="💡",
+        default_model="meta-llama/Meta-Llama-3.1-70B-Instruct", aliases=[], icon="INF",
     ),
 ]
 
+
 _BY_ID: dict[str, ProviderDefinition] = {p.id: p for p in _BUILTIN}
 
 
@@ -330,4 +338,4 @@ def search(query: str) -> list[ProviderDefinition]:
         if score > 0:
             results.append((score, p))
     results.sort(key=lambda x: (-x[0], x[1].name))
-    return [p for _, p in results]
+    return [p for _, p in results]

package/src/reidx/provider_manager/palette.py

@@ -96,6 +96,7 @@ class ProviderPalette:
     CONFIRM = "confirm"
     WIZARD = "wizard"
     MESSAGE = "message"
+    MODELS = "models"
 
     def __init__(
         self,
@@ -163,7 +164,7 @@ class ProviderPalette:
     def is_list_screen(self) -> bool:
         return self._active and self.screen in (
             self.LIST, self.KEYS, self.MANAGE, self.RENAME_SEL,
-            self.DELETE_SEL, self.CONFIRM, self.WIZARD,
+            self.DELETE_SEL, self.CONFIRM, self.WIZARD, self.MODELS,
         ) and not self.is_input_screen()
 
     def activate(self) -> None:
@@ -221,6 +222,10 @@ class ProviderPalette:
             self.screen = self._message_next
             self.selected_index = 0
             self._scroll_offset = 0
+        elif self.screen == self.MODELS:
+            self.screen = self.MANAGE
+            self.selected_index = 0
+            self._scroll_offset = 0
         elif self.screen == self.WIZARD:
             if self.wizard_step_idx > 0:
                 self.wizard_step_idx -= 1
@@ -311,6 +316,8 @@ class ProviderPalette:
                 if step.field == "auth_method":
                     return AUTH_OPTIONS
             return []
+        if self.screen == self.MODELS:
+            return self._models_items()
         return []
 
     def _list_items(self) -> list[PaletteItem]:
@@ -327,14 +334,14 @@ class ProviderPalette:
             if d.name.lower() in stored:
                 sp = stored[d.name.lower()]
                 active_key = sp.active_key()
-                icon = "●" if active_key else "○"
+                icon = "?" if active_key else "	"
                 items.append(PaletteItem(
                     label=d.name, description=d.description, icon=icon,
                     kind="stored", data=sp,
                 ))
             else:
                 items.append(PaletteItem(
-                    label=d.name, description=d.description, icon="◆",
+                    label=d.name, description=d.description, icon="?",
                     kind="catalog", data=d,
                 ))
 
@@ -342,7 +349,7 @@ class ProviderPalette:
             if sp.name.lower() not in {d.name.lower() for d in defs}:
                 if not q or q in sp.name.lower() or q in sp.kind.lower() or q in sp.base_url.lower():
                     active_key = sp.active_key()
-                    icon = "●" if active_key else "○"
+                    icon = "?" if active_key else "	"
                     items.append(PaletteItem(
                         label=sp.name, description=f"{sp.kind}  {sp.base_url}", icon=icon,
                         kind="stored", data=sp,
@@ -359,7 +366,7 @@ class ProviderPalette:
         if self.current_provider:
             for k in self.current_provider.keys:
                 is_active = k.id == self.current_provider.active_key_id
-                icon = "●" if is_active else "○"
+                icon = "?" if is_active else "	"
                 desc = "active" if is_active else ""
                 items.append(PaletteItem(label=k.label, description=desc, icon=icon, kind="action", data=k))
         items.append(PaletteItem(label="Add API Key", icon="+", kind="action"))
@@ -369,13 +376,14 @@ class ProviderPalette:
         items: list[PaletteItem] = []
         sp = self.current_provider
         if sp and len(sp.keys) > 1:
-            items.append(PaletteItem("Switch Key", "Change the active API key", icon="→"))
+            items.append(PaletteItem("Switch Key", "Change the active API key", icon=""))
         items.append(PaletteItem("Add Key", "Add another API key", icon="+"))
         if sp and sp.keys:
-            items.append(PaletteItem("Rename Key", "Rename a key label", icon="✎"))
-            items.append(PaletteItem("Delete Key", "Remove a key", icon="✕"))
-        items.append(PaletteItem("Edit Provider", "Change base URL, model, etc.", icon="⚙"))
-        items.append(PaletteItem("Remove Provider", "Delete provider and all keys", icon="✕", kind="action"))
+            items.append(PaletteItem("Rename Key", "Rename a key label", icon="?"))
+            items.append(PaletteItem("Delete Key", "Remove a key", icon="?"))
+        items.append(PaletteItem("View Models", "Browse and set default model", icon="M"))
+        items.append(PaletteItem("Edit Provider", "Change base URL, model, etc.", icon="?"))
+        items.append(PaletteItem("Remove Provider", "Delete provider and all keys", icon="?", kind="action"))
         return items
 
     def _key_select_items(self, _mode: str) -> list[PaletteItem]:
@@ -383,10 +391,43 @@ class ProviderPalette:
         if self.current_provider:
             for k in self.current_provider.keys:
                 is_active = k.id == self.current_provider.active_key_id
-                icon = "●" if is_active else "○"
+                icon = "?" if is_active else "	"
                 items.append(PaletteItem(label=k.label, icon=icon, kind="action", data=k))
         return items
 
+    def _models_items(self) -> list[PaletteItem]:
+        items: list[PaletteItem] = []
+        if not self.current_provider:
+            return items
+        
+        sp = self.current_provider
+        try:
+            provider = build_provider(ProviderRecord(
+                name=sp.name, kind=sp.kind, base_url=sp.base_url,
+                api_key=sp.decrypted_api_key(), default_model=sp.default_model,
+                auth_method=sp.auth_method,
+            ))
+            models = provider.fetch_models()
+        except Exception as exc:
+            log.warning("failed to fetch models for %s: %s", sp.name, exc)
+            items.append(PaletteItem("Error loading models", str(exc), icon="!", kind="action"))
+            return items
+        
+        if not models:
+            items.append(PaletteItem("No models available", "Provider returned empty model list", icon="!", kind="action"))
+            return items
+        
+        current_model = sp.default_model or ""
+        for model in models:
+            is_current = model == current_model
+            icon = "?" if is_current else "	"
+            desc = "current" if is_current else ""
+            items.append(PaletteItem(
+                label=model, description=desc, icon=icon, kind="model", data=model
+            ))
+        
+        return items
+
     def _handle_list_enter(self, item: PaletteItem) -> None:
         if self.screen == self.LIST:
             self._on_list_select(item)
@@ -402,6 +443,8 @@ class ProviderPalette:
             self._on_confirm_select(item)
         elif self.screen == self.WIZARD:
             self._on_wizard_select(item)
+        elif self.screen == self.MODELS:
+            self._on_models_select(item)
 
     def _on_list_select(self, item: PaletteItem) -> None:
         if item.kind == "action":
@@ -472,6 +515,11 @@ class ProviderPalette:
             self.selected_index = 0
             self._scroll_offset = 0
             self._invalidate()
+        elif label == "View Models":
+            self.screen = self.MODELS
+            self.selected_index = 0
+            self._scroll_offset = 0
+            self._invalidate()
         elif label == "Edit Provider":
             self._start_wizard(edit=True)
         elif label == "Remove Provider":
@@ -503,6 +551,46 @@ class ProviderPalette:
             self._scroll_offset = 0
             self._invalidate()
 
+    def _models_items(self) -> list[PaletteItem]:
+        items: list[PaletteItem] = []
+        if not self.current_provider:
+            return items
+        
+        sp = self.current_provider
+        try:
+            provider = build_provider(ProviderRecord(
+                name=sp.name, kind=sp.kind, base_url=sp.base_url,
+                api_key=sp.decrypted_api_key(), default_model=sp.default_model,
+                auth_method=sp.auth_method,
+            ))
+            models = provider.fetch_models()
+        except Exception as exc:
+            log.warning("failed to fetch models for %s: %s", sp.name, exc)
+            items.append(PaletteItem("Error loading models", str(exc), icon="!", kind="action"))
+            return items
+        
+        if not models:
+            items.append(PaletteItem("No models available", "Provider returned empty model list", icon="!", kind="action"))
+            return items
+        
+        current_model = sp.default_model or ""
+        for model in models:
+            is_current = model == current_model
+            icon = "?" if is_current else "	"
+            desc = "current" if is_current else ""
+            items.append(PaletteItem(
+                label=model, description=desc, icon=icon, kind="model", data=model
+            ))
+        
+        return items
+
+    def _on_models_select(self, item: PaletteItem) -> None:
+        if item.kind == "model" and item.data:
+            self.current_provider.default_model = item.data
+            self.db.save_provider(self.current_provider)
+            self._register_current()
+            self._show_message(f"Set default model to '{item.data}'", self.MANAGE)
+
     def _on_confirm_select(self, item: PaletteItem) -> None:
         if item.label == "Confirm":
             if self._pending_action:
@@ -524,6 +612,13 @@ class ProviderPalette:
         self.wizard_data[step.field] = item.label
         self._wizard_advance()
 
+    def _on_models_select(self, item: PaletteItem) -> None:
+        if item.kind == "model" and item.data:
+            self.current_provider.default_model = item.data
+            self.db.save_provider(self.current_provider)
+            self._register_current()
+            self._show_message(f"Set default model to '{item.data}'", self.MANAGE)
+
     def _handle_input_enter(self) -> None:
         text = self.input_buf.text
         if self.screen == self.KEY_LABEL:
@@ -820,6 +915,9 @@ class ProviderPalette:
             return [(f"bold {ACCENT}", "  + Custom Provider"), ("", "  "), (DIM, f"Step {self.wizard_step_idx + 1}/{total}: {step.prompt}")]
         if self.screen == self.MESSAGE:
             return [(f"bold {OK}", "  ✓ Done")]
+        if self.screen == self.MODELS:
+            name = self.current_provider.name if self.current_provider else ""
+            return [(f"bold {ACCENT}", f"  {name}"), ("", "  "), (DIM, "Models")]
         return [(f"bold {ACCENT}", "  Provider")]
 
     def content_fragments(self) -> list[tuple[str, str]]:
@@ -829,7 +927,7 @@ class ProviderPalette:
             return [(f"{WARN}", f"  {self._confirm_text}")]
         if self.screen == self.WIZARD and self.is_input_screen():
             step = WIZARD_STEPS[self.wizard_step_idx]
-            hint = " (optional — press Enter to skip)" if step.optional else ""
+            hint = " (optional - press Enter to skip)" if step.optional else ""
             return [(DIM, f"  {step.prompt}{hint}")]
         if self.is_input_screen():
             return [(DIM, f"  {self.input_prompt_text}")]
@@ -850,7 +948,7 @@ class ProviderPalette:
             desc = item.description
             remaining = max(1, inner - len(label_part) - 1)
             if len(desc) > remaining:
-                desc = desc[: remaining - 1] + "…"
+                desc = desc[: remaining - 1] + "."
             desc_part = f" {desc}" if desc else ""
             pad = max(0, inner - len(label_part) - len(desc_part))
             if is_sel:
@@ -878,12 +976,14 @@ class ProviderPalette:
         if self.is_input_screen():
             return [(DIM, "  Enter to continue  Esc to go back")]
         if self.screen == self.CONFIRM:
-            return [(DIM, "  ↑↓ Select  Enter Confirm  Esc Cancel")]
+            return [(DIM, "   Select  Enter Confirm  Esc Cancel")]
         if self.screen == self.MESSAGE:
             return [(DIM, "  Enter to continue")]
         if self.screen == self.LIST:
-            return [(DIM, "  ↑↓ Navigate  Enter Select  Esc Close  Type to search")]
-        return [(DIM, "  ↑↓ Navigate  Enter Select  Esc Back")]
+            return [(DIM, "   Navigate  Enter Select  Esc Close  Type to search")]
+        if self.screen == self.MODELS:
+            return [(DIM, "   Navigate  Enter Select  Esc Back")]
+        return [(DIM, "   Navigate  Enter Select  Esc Back")]
 
     def input_label(self) -> str:
         if self.screen == self.KEY_LABEL:
@@ -899,8 +999,8 @@ class ProviderPalette:
 
     def search_label(self) -> str:
         if self.screen == self.LIST:
-            return " ⌕ "
-        return " › "
+            return " ? "
+        return " > "
 
     def content_height(self) -> int:
         if self.screen == self.MESSAGE:
@@ -913,4 +1013,4 @@ class ProviderPalette:
         return max(1, min(len(items), self.max_content_lines()))
 
     def total_height(self) -> int:
-        return self.content_height() + 6
+        return self.content_height() + 6