muonroi-cli 1.4.1 → 1.5.0

package/dist/src/models/catalog.json

@@ -1,349 +1,349 @@
-{
-  "version": "2.2",
-  "updated_at": "2026-06-04",
-  "description": "Local fallback catalog: deepseek + siliconflow + openai (ChatGPT OAuth) + google (Gemini, via AI Studio key or Google OAuth). SiliconFlow pricing verified 2026-05-21; Gemini 2.5 pricing per ai.google.dev/pricing 2026-06-04 (3.x flash/pro entries APPROXIMATE — see per-model descriptions). Model IDs verified against ai.google.dev/gemini-api/docs/models 2026-06-04. See ./catalog.README.md for curation rationale.",
-  "models": [
-    {
-      "id": "deepseek-v4-flash",
-      "name": "DeepSeek V4 Flash (native)",
-      "provider": "deepseek",
-      "tier": "fast",
-      "context_window": 128000,
-      "max_output_tokens": 8000,
-      "input_price_per_million": 0.27,
-      "output_price_per_million": 1.1,
-      "cached_input_price_per_million": 0.027,
-      "reasoning": true,
-      "thinking_type": "enabled",
-      "supports_effort": false,
-      "default_reasoning_effort": null,
-      "description": "Native DeepSeek V4 Flash. Tool-call capable. More expensive than SiliconFlow's hosted V4-Flash ($0.14/$0.28) — prefer SF unless rate-limited.",
-      "aliases": ["deepseek-flash-native"],
-      "supports_vision": false
-    },
-    {
-      "id": "deepseek-v4-pro",
-      "name": "DeepSeek V4 Pro (native)",
-      "provider": "deepseek",
-      "tier": "premium",
-      "context_window": 128000,
-      "max_output_tokens": 16000,
-      "input_price_per_million": 0.55,
-      "output_price_per_million": 2.19,
-      "cached_input_price_per_million": 0.055,
-      "reasoning": true,
-      "thinking_type": "enabled",
-      "supports_effort": false,
-      "default_reasoning_effort": null,
-      "description": "Native DeepSeek V4 Pro. Tool-call capable. Cheaper than SF-hosted V4-Pro ($1.74/$3.48) — this is the premium-tier default.",
-      "aliases": ["deepseek-pro-native"],
-      "supports_vision": false
-    },
-    {
-      "id": "deepseek-ai/DeepSeek-V4-Flash",
-      "name": "DeepSeek V4 Flash (via SiliconFlow)",
-      "provider": "siliconflow",
-      "tier": "fast",
-      "context_window": 1049000,
-      "max_output_tokens": 8000,
-      "input_price_per_million": 0.14,
-      "output_price_per_million": 0.28,
-      "reasoning": true,
-      "thinking_type": "enabled",
-      "supports_effort": false,
-      "default_reasoning_effort": null,
-      "description": "DeepSeek V4 Flash on SiliconFlow — cheaper than native. 1M context. Reasoning_tokens share output budget.",
-      "aliases": ["deepseek-v4-flash-sf", "deepseek-flash-sf"],
-      "supports_vision": false
-    },
-    {
-      "id": "deepseek-ai/DeepSeek-V4-Pro",
-      "name": "DeepSeek V4 Pro (via SiliconFlow)",
-      "provider": "siliconflow",
-      "tier": "premium",
-      "context_window": 1049000,
-      "max_output_tokens": 16000,
-      "input_price_per_million": 1.74,
-      "output_price_per_million": 3.48,
-      "reasoning": true,
-      "thinking_type": "enabled",
-      "supports_effort": false,
-      "default_reasoning_effort": null,
-      "description": "DeepSeek V4 Pro on SiliconFlow. Prefer native deepseek-v4-pro for cost ($0.55/$2.19) when key available.",
-      "aliases": ["deepseek-v4-pro-sf", "deepseek-pro-sf"],
-      "supports_vision": false
-    },
-    {
-      "id": "Qwen/Qwen3-8B",
-      "name": "Qwen3 8B (SiliconFlow)",
-      "provider": "siliconflow",
-      "tier": "fast",
-      "context_window": 131072,
-      "max_output_tokens": 8192,
-      "input_price_per_million": 0.06,
-      "output_price_per_million": 0.06,
-      "reasoning": false,
-      "thinking_type": "hybrid",
-      "supports_effort": false,
-      "default_reasoning_effort": null,
-      "description": "Cheapest tool-capable SiliconFlow model. Symmetric pricing — ideal for mechanical tool-execution loops. Supports thinking/non-thinking hybrid mode.",
-      "aliases": ["qwen3-8b-sf", "alibaba/Qwen3-8B"],
-      "supports_vision": false
-    },
-    {
-      "id": "Qwen/Qwen3-30B-A3B-Instruct-2507",
-      "name": "Qwen3 30B A3B Instruct (SiliconFlow)",
-      "provider": "siliconflow",
-      "tier": "balanced",
-      "context_window": 262144,
-      "max_output_tokens": 8192,
-      "input_price_per_million": 0.09,
-      "output_price_per_million": 0.3,
-      "reasoning": false,
-      "thinking_type": null,
-      "supports_effort": false,
-      "default_reasoning_effort": null,
-      "description": "MoE balanced default. 3B active params keeps cost low while 30B total gives strong instruction following + tool use. Recommended SF balanced pick.",
-      "aliases": ["qwen3-30b-instruct-sf", "alibaba/Qwen3-30B-A3B-Instruct-2507"],
-      "supports_vision": false
-    },
-    {
-      "id": "Qwen/Qwen3-Coder-30B-A3B-Instruct",
-      "name": "Qwen3 Coder 30B A3B (SiliconFlow)",
-      "provider": "siliconflow",
-      "tier": "balanced",
-      "context_window": 262144,
-      "max_output_tokens": 8192,
-      "input_price_per_million": 0.07,
-      "output_price_per_million": 0.28,
-      "reasoning": false,
-      "thinking_type": null,
-      "supports_effort": false,
-      "default_reasoning_effort": null,
-      "description": "Coding-specialized Qwen3 MoE. Cheaper than the generic Qwen3-30B variant. Use when pil.taskType is coding (router override pending).",
-      "aliases": ["qwen3-coder-sf", "qwen-coder-sf", "alibaba/Qwen3-Coder-30B-A3B-Instruct"],
-      "supports_vision": false
-    },
-    {
-      "id": "deepseek-ai/DeepSeek-V3.2",
-      "name": "DeepSeek V3.2 (SiliconFlow)",
-      "provider": "siliconflow",
-      "tier": "balanced",
-      "context_window": 163840,
-      "max_output_tokens": 8192,
-      "input_price_per_million": 0.27,
-      "output_price_per_million": 0.42,
-      "reasoning": true,
-      "thinking_type": "enabled",
-      "supports_effort": false,
-      "default_reasoning_effort": null,
-      "description": "Cheap DeepSeek reasoning option for balanced-tier turns that benefit from CoT. Tool-capable. Use when Qwen3-30B is not strong enough on reasoning.",
-      "aliases": ["deepseek-v3.2-sf"],
-      "supports_vision": false
-    },
-    {
-      "id": "deepseek-ai/DeepSeek-R1",
-      "name": "DeepSeek R1 (SiliconFlow)",
-      "provider": "siliconflow",
-      "tier": "premium",
-      "context_window": 163840,
-      "max_output_tokens": 16384,
-      "input_price_per_million": 0.5,
-      "output_price_per_million": 2.18,
-      "reasoning": true,
-      "thinking_type": "enabled",
-      "supports_effort": false,
-      "default_reasoning_effort": null,
-      "description": "Cheaper premium reasoning alternative to V4-Pro. Tool-capable. Use when V4-Pro is rate-limited or for benchmark parity.",
-      "aliases": ["deepseek-r1-sf"],
-      "supports_vision": false
-    },
-    {
-      "id": "z-ai/GLM-4.6V",
-      "name": "GLM 4.6V Vision (SiliconFlow)",
-      "provider": "siliconflow",
-      "tier": "balanced",
-      "context_window": 131072,
-      "max_output_tokens": 4096,
-      "input_price_per_million": 0.3,
-      "output_price_per_million": 0.9,
-      "reasoning": false,
-      "thinking_type": null,
-      "supports_effort": false,
-      "default_reasoning_effort": null,
-      "description": "Vision-language model with confirmed function-calling support. Use when the turn contains image attachments (router vision-override pending).",
-      "aliases": ["glm-4.6v-sf"],
-      "supports_vision": true
-    },
-    {
-      "id": "gpt-5.4-mini",
-      "name": "GPT-5.4 mini (OpenAI ChatGPT OAuth)",
-      "provider": "openai",
-      "tier": "fast",
-      "context_window": 256000,
-      "max_output_tokens": 32000,
-      "input_price_per_million": 0,
-      "output_price_per_million": 0,
-      "reasoning": true,
-      "thinking_type": "enabled",
-      "supports_effort": true,
-      "default_reasoning_effort": "low",
-      "description": "OpenAI GPT-5.4 mini via ChatGPT subscription OAuth (chatgpt.com/backend-api/codex, Responses API). Subscription-billed — per-token price N/A (0 placeholder). Added to local fallback catalog so OAuth-granted OpenAI models resolve when the CP catalog endpoint is unreachable.",
-      "aliases": ["gpt-5.4-mini-oauth"],
-      "supports_vision": false
-    },
-    {
-      "id": "gpt-5.4",
-      "name": "GPT-5.4 (OpenAI ChatGPT OAuth)",
-      "provider": "openai",
-      "tier": "premium",
-      "context_window": 256000,
-      "max_output_tokens": 32000,
-      "input_price_per_million": 0,
-      "output_price_per_million": 0,
-      "reasoning": true,
-      "thinking_type": "enabled",
-      "supports_effort": true,
-      "default_reasoning_effort": "medium",
-      "description": "OpenAI GPT-5.4 via ChatGPT subscription OAuth (chatgpt.com/backend-api/codex, Responses API). Subscription-billed — per-token price N/A (0 placeholder). Added to local fallback catalog so OAuth-granted OpenAI models resolve when the CP catalog endpoint is unreachable.",
-      "aliases": ["gpt-5.4-oauth"],
-      "supports_vision": false
-    },
-    {
-      "id": "gpt-5.3-codex",
-      "name": "GPT-5.3 Codex (OpenAI ChatGPT OAuth)",
-      "provider": "openai",
-      "tier": "balanced",
-      "context_window": 256000,
-      "max_output_tokens": 32000,
-      "input_price_per_million": 0,
-      "output_price_per_million": 0,
-      "reasoning": true,
-      "thinking_type": "enabled",
-      "supports_effort": true,
-      "default_reasoning_effort": "medium",
-      "description": "OpenAI GPT-5.3 Codex (coding-specialized) via ChatGPT subscription OAuth (chatgpt.com/backend-api/codex, Responses API). Subscription-billed — per-token price N/A (0 placeholder). Added to local fallback catalog so OAuth-granted OpenAI models resolve when the CP catalog endpoint is unreachable.",
-      "aliases": ["gpt-5.3-codex-oauth", "codex"],
-      "supports_vision": false
-    },
-    {
-      "id": "gemini-2.5-flash",
-      "name": "Gemini 2.5 Flash",
-      "provider": "google",
-      "tier": "balanced",
-      "context_window": 1048576,
-      "max_output_tokens": 65536,
-      "input_price_per_million": 0.3,
-      "output_price_per_million": 2.5,
-      "reasoning": true,
-      "thinking_type": "enabled",
-      "supports_effort": false,
-      "default_reasoning_effort": null,
-      "description": "Google Gemini 2.5 Flash via Generative Language API (AI Studio key — console aistudio.google.com/app/apikey, or stored Google OAuth). Multimodal, 1M context, thinking-capable. Pricing per ai.google.dev/pricing (2026-06-04).",
-      "aliases": ["gemini-flash", "gemini-2.5-flash-latest"],
-      "supports_vision": true
-    },
-    {
-      "id": "gemini-2.5-pro",
-      "name": "Gemini 2.5 Pro",
-      "provider": "google",
-      "tier": "premium",
-      "context_window": 1048576,
-      "max_output_tokens": 65536,
-      "input_price_per_million": 1.25,
-      "output_price_per_million": 10.0,
-      "reasoning": true,
-      "thinking_type": "enabled",
-      "supports_effort": false,
-      "default_reasoning_effort": null,
-      "description": "Google Gemini 2.5 Pro via Generative Language API (AI Studio key or stored Google OAuth). Most advanced 2.5 reasoning model, multimodal, 1M context. Pricing per ai.google.dev/pricing (2026-06-04).",
-      "aliases": ["gemini-pro", "gemini-2.5-pro-latest"],
-      "supports_vision": true
-    },
-    {
-      "id": "gemini-3.5-flash",
-      "name": "Gemini 3.5 Flash",
-      "provider": "google",
-      "tier": "fast",
-      "context_window": 1048576,
-      "max_output_tokens": 65536,
-      "input_price_per_million": 0.5,
-      "output_price_per_million": 3.0,
-      "reasoning": true,
-      "thinking_type": "enabled",
-      "supports_effort": false,
-      "default_reasoning_effort": null,
-      "description": "Google Gemini 3.5 Flash via Generative Language API. Frontier agentic/coding flash model, multimodal, 1M context, thinking-capable. Pricing APPROXIMATE (verify at ai.google.dev/pricing before cost-sensitive use).",
-      "aliases": ["gemini-flash-3.5"],
-      "supports_vision": true
-    },
-    {
-      "id": "gemini-3.1-flash-lite",
-      "name": "Gemini 3.1 Flash-Lite",
-      "provider": "google",
-      "tier": "fast",
-      "context_window": 1048576,
-      "max_output_tokens": 65536,
-      "input_price_per_million": 0.15,
-      "output_price_per_million": 0.6,
-      "reasoning": true,
-      "thinking_type": "enabled",
-      "supports_effort": false,
-      "default_reasoning_effort": null,
-      "description": "Google Gemini 3.1 Flash-Lite via Generative Language API. Cheapest multimodal Gemini, 1M context, limited thinking. Pricing APPROXIMATE (verify at ai.google.dev/pricing before cost-sensitive use).",
-      "aliases": ["gemini-flash-lite"],
-      "supports_vision": true
-    },
-    {
-      "id": "gemini-3.1-pro-preview",
-      "name": "Gemini 3.1 Pro Preview",
-      "provider": "google",
-      "tier": "premium",
-      "context_window": 1048576,
-      "max_output_tokens": 65536,
-      "input_price_per_million": 2.0,
-      "output_price_per_million": 12.0,
-      "reasoning": true,
-      "thinking_type": "enabled",
-      "supports_effort": false,
-      "default_reasoning_effort": null,
-      "description": "Google Gemini 3.1 Pro Preview via Generative Language API. Strongest agentic/reasoning Gemini, multimodal, 1M context. Preview model — id/pricing may change; pricing APPROXIMATE (verify at ai.google.dev/pricing).",
-      "aliases": ["gemini-pro-preview"],
-      "supports_vision": true
-    },
-    {
-      "id": "grok-4.3",
-      "name": "Grok 4.3 (xAI)",
-      "provider": "xai",
-      "tier": "premium",
-      "context_window": 1048576,
-      "max_output_tokens": 32768,
-      "input_price_per_million": 1.25,
-      "output_price_per_million": 2.5,
-      "reasoning": true,
-      "thinking_type": "enabled",
-      "supports_effort": true,
-      "default_reasoning_effort": "medium",
-      "description": "Flagship model mạnh nhất hiện tại của xAI. Agentic tool-calling xuất sắc, hallucination thấp, hỗ trợ reasoning + non-reasoning mode. Context 1M tokens. Model mặc định khuyến nghị cho Grok CLI.",
-      "aliases": ["grok-4.3", "grok-latest", "grok-4"],
-      "supports_vision": true
-    },
-    {
-      "id": "grok-build-0.1",
-      "name": "Grok Build 0.1 (xAI)",
-      "provider": "xai",
-      "tier": "balanced",
-      "context_window": 262144,
-      "max_output_tokens": 65536,
-      "input_price_per_million": 1.0,
-      "output_price_per_million": 2.0,
-      "reasoning": true,
-      "thinking_type": "enabled",
-      "supports_effort": false,
-      "default_reasoning_effort": null,
-      "description": "Model coding chuyên biệt, nhanh, tối ưu cho agentic coding workflows. Chính là model powering Grok Build CLI. Rất phù hợp cho lập trình, debug, multi-step tasks.",
-      "aliases": ["grok-build", "grok-code-fast", "grok-build-0.1"],
-      "supports_vision": true
-    }
-  ]
-}
+{
+  "version": "2.2",
+  "updated_at": "2026-06-04",
+  "description": "Local fallback catalog: deepseek + siliconflow + openai (ChatGPT OAuth) + google (Gemini, via AI Studio key or Google OAuth). SiliconFlow pricing verified 2026-05-21; Gemini 2.5 pricing per ai.google.dev/pricing 2026-06-04 (3.x flash/pro entries APPROXIMATE — see per-model descriptions). Model IDs verified against ai.google.dev/gemini-api/docs/models 2026-06-04. See ./catalog.README.md for curation rationale.",
+  "models": [
+    {
+      "id": "deepseek-v4-flash",
+      "name": "DeepSeek V4 Flash (native)",
+      "provider": "deepseek",
+      "tier": "fast",
+      "context_window": 128000,
+      "max_output_tokens": 8000,
+      "input_price_per_million": 0.27,
+      "output_price_per_million": 1.1,
+      "cached_input_price_per_million": 0.027,
+      "reasoning": true,
+      "thinking_type": "enabled",
+      "supports_effort": false,
+      "default_reasoning_effort": null,
+      "description": "Native DeepSeek V4 Flash. Tool-call capable. More expensive than SiliconFlow's hosted V4-Flash ($0.14/$0.28) — prefer SF unless rate-limited.",
+      "aliases": ["deepseek-flash-native"],
+      "supports_vision": false
+    },
+    {
+      "id": "deepseek-v4-pro",
+      "name": "DeepSeek V4 Pro (native)",
+      "provider": "deepseek",
+      "tier": "premium",
+      "context_window": 128000,
+      "max_output_tokens": 16000,
+      "input_price_per_million": 0.55,
+      "output_price_per_million": 2.19,
+      "cached_input_price_per_million": 0.055,
+      "reasoning": true,
+      "thinking_type": "enabled",
+      "supports_effort": false,
+      "default_reasoning_effort": null,
+      "description": "Native DeepSeek V4 Pro. Tool-call capable. Cheaper than SF-hosted V4-Pro ($1.74/$3.48) — this is the premium-tier default.",
+      "aliases": ["deepseek-pro-native"],
+      "supports_vision": false
+    },
+    {
+      "id": "deepseek-ai/DeepSeek-V4-Flash",
+      "name": "DeepSeek V4 Flash (via SiliconFlow)",
+      "provider": "siliconflow",
+      "tier": "fast",
+      "context_window": 1049000,
+      "max_output_tokens": 8000,
+      "input_price_per_million": 0.14,
+      "output_price_per_million": 0.28,
+      "reasoning": true,
+      "thinking_type": "enabled",
+      "supports_effort": false,
+      "default_reasoning_effort": null,
+      "description": "DeepSeek V4 Flash on SiliconFlow — cheaper than native. 1M context. Reasoning_tokens share output budget.",
+      "aliases": ["deepseek-v4-flash-sf", "deepseek-flash-sf"],
+      "supports_vision": false
+    },
+    {
+      "id": "deepseek-ai/DeepSeek-V4-Pro",
+      "name": "DeepSeek V4 Pro (via SiliconFlow)",
+      "provider": "siliconflow",
+      "tier": "premium",
+      "context_window": 1049000,
+      "max_output_tokens": 16000,
+      "input_price_per_million": 1.74,
+      "output_price_per_million": 3.48,
+      "reasoning": true,
+      "thinking_type": "enabled",
+      "supports_effort": false,
+      "default_reasoning_effort": null,
+      "description": "DeepSeek V4 Pro on SiliconFlow. Prefer native deepseek-v4-pro for cost ($0.55/$2.19) when key available.",
+      "aliases": ["deepseek-v4-pro-sf", "deepseek-pro-sf"],
+      "supports_vision": false
+    },
+    {
+      "id": "Qwen/Qwen3-8B",
+      "name": "Qwen3 8B (SiliconFlow)",
+      "provider": "siliconflow",
+      "tier": "fast",
+      "context_window": 131072,
+      "max_output_tokens": 8192,
+      "input_price_per_million": 0.06,
+      "output_price_per_million": 0.06,
+      "reasoning": false,
+      "thinking_type": "hybrid",
+      "supports_effort": false,
+      "default_reasoning_effort": null,
+      "description": "Cheapest tool-capable SiliconFlow model. Symmetric pricing — ideal for mechanical tool-execution loops. Supports thinking/non-thinking hybrid mode.",
+      "aliases": ["qwen3-8b-sf", "alibaba/Qwen3-8B"],
+      "supports_vision": false
+    },
+    {
+      "id": "Qwen/Qwen3-30B-A3B-Instruct-2507",
+      "name": "Qwen3 30B A3B Instruct (SiliconFlow)",
+      "provider": "siliconflow",
+      "tier": "balanced",
+      "context_window": 262144,
+      "max_output_tokens": 8192,
+      "input_price_per_million": 0.09,
+      "output_price_per_million": 0.3,
+      "reasoning": false,
+      "thinking_type": null,
+      "supports_effort": false,
+      "default_reasoning_effort": null,
+      "description": "MoE balanced default. 3B active params keeps cost low while 30B total gives strong instruction following + tool use. Recommended SF balanced pick.",
+      "aliases": ["qwen3-30b-instruct-sf", "alibaba/Qwen3-30B-A3B-Instruct-2507"],
+      "supports_vision": false
+    },
+    {
+      "id": "Qwen/Qwen3-Coder-30B-A3B-Instruct",
+      "name": "Qwen3 Coder 30B A3B (SiliconFlow)",
+      "provider": "siliconflow",
+      "tier": "balanced",
+      "context_window": 262144,
+      "max_output_tokens": 8192,
+      "input_price_per_million": 0.07,
+      "output_price_per_million": 0.28,
+      "reasoning": false,
+      "thinking_type": null,
+      "supports_effort": false,
+      "default_reasoning_effort": null,
+      "description": "Coding-specialized Qwen3 MoE. Cheaper than the generic Qwen3-30B variant. Use when pil.taskType is coding (router override pending).",
+      "aliases": ["qwen3-coder-sf", "qwen-coder-sf", "alibaba/Qwen3-Coder-30B-A3B-Instruct"],
+      "supports_vision": false
+    },
+    {
+      "id": "deepseek-ai/DeepSeek-V3.2",
+      "name": "DeepSeek V3.2 (SiliconFlow)",
+      "provider": "siliconflow",
+      "tier": "balanced",
+      "context_window": 163840,
+      "max_output_tokens": 8192,
+      "input_price_per_million": 0.27,
+      "output_price_per_million": 0.42,
+      "reasoning": true,
+      "thinking_type": "enabled",
+      "supports_effort": false,
+      "default_reasoning_effort": null,
+      "description": "Cheap DeepSeek reasoning option for balanced-tier turns that benefit from CoT. Tool-capable. Use when Qwen3-30B is not strong enough on reasoning.",
+      "aliases": ["deepseek-v3.2-sf"],
+      "supports_vision": false
+    },
+    {
+      "id": "deepseek-ai/DeepSeek-R1",
+      "name": "DeepSeek R1 (SiliconFlow)",
+      "provider": "siliconflow",
+      "tier": "premium",
+      "context_window": 163840,
+      "max_output_tokens": 16384,
+      "input_price_per_million": 0.5,
+      "output_price_per_million": 2.18,
+      "reasoning": true,
+      "thinking_type": "enabled",
+      "supports_effort": false,
+      "default_reasoning_effort": null,
+      "description": "Cheaper premium reasoning alternative to V4-Pro. Tool-capable. Use when V4-Pro is rate-limited or for benchmark parity.",
+      "aliases": ["deepseek-r1-sf"],
+      "supports_vision": false
+    },
+    {
+      "id": "z-ai/GLM-4.6V",
+      "name": "GLM 4.6V Vision (SiliconFlow)",
+      "provider": "siliconflow",
+      "tier": "balanced",
+      "context_window": 131072,
+      "max_output_tokens": 4096,
+      "input_price_per_million": 0.3,
+      "output_price_per_million": 0.9,
+      "reasoning": false,
+      "thinking_type": null,
+      "supports_effort": false,
+      "default_reasoning_effort": null,
+      "description": "Vision-language model with confirmed function-calling support. Use when the turn contains image attachments (router vision-override pending).",
+      "aliases": ["glm-4.6v-sf"],
+      "supports_vision": true
+    },
+    {
+      "id": "gpt-5.4-mini",
+      "name": "GPT-5.4 mini (OpenAI ChatGPT OAuth)",
+      "provider": "openai",
+      "tier": "fast",
+      "context_window": 256000,
+      "max_output_tokens": 32000,
+      "input_price_per_million": 0,
+      "output_price_per_million": 0,
+      "reasoning": true,
+      "thinking_type": "enabled",
+      "supports_effort": true,
+      "default_reasoning_effort": "low",
+      "description": "OpenAI GPT-5.4 mini via ChatGPT subscription OAuth (chatgpt.com/backend-api/codex, Responses API). Subscription-billed — per-token price N/A (0 placeholder). Added to local fallback catalog so OAuth-granted OpenAI models resolve when the CP catalog endpoint is unreachable.",
+      "aliases": ["gpt-5.4-mini-oauth"],
+      "supports_vision": false
+    },
+    {
+      "id": "gpt-5.4",
+      "name": "GPT-5.4 (OpenAI ChatGPT OAuth)",
+      "provider": "openai",
+      "tier": "premium",
+      "context_window": 256000,
+      "max_output_tokens": 32000,
+      "input_price_per_million": 0,
+      "output_price_per_million": 0,
+      "reasoning": true,
+      "thinking_type": "enabled",
+      "supports_effort": true,
+      "default_reasoning_effort": "medium",
+      "description": "OpenAI GPT-5.4 via ChatGPT subscription OAuth (chatgpt.com/backend-api/codex, Responses API). Subscription-billed — per-token price N/A (0 placeholder). Added to local fallback catalog so OAuth-granted OpenAI models resolve when the CP catalog endpoint is unreachable.",
+      "aliases": ["gpt-5.4-oauth"],
+      "supports_vision": false
+    },
+    {
+      "id": "gpt-5.3-codex",
+      "name": "GPT-5.3 Codex (OpenAI ChatGPT OAuth)",
+      "provider": "openai",
+      "tier": "balanced",
+      "context_window": 256000,
+      "max_output_tokens": 32000,
+      "input_price_per_million": 0,
+      "output_price_per_million": 0,
+      "reasoning": true,
+      "thinking_type": "enabled",
+      "supports_effort": true,
+      "default_reasoning_effort": "medium",
+      "description": "OpenAI GPT-5.3 Codex (coding-specialized) via ChatGPT subscription OAuth (chatgpt.com/backend-api/codex, Responses API). Subscription-billed — per-token price N/A (0 placeholder). Added to local fallback catalog so OAuth-granted OpenAI models resolve when the CP catalog endpoint is unreachable.",
+      "aliases": ["gpt-5.3-codex-oauth", "codex"],
+      "supports_vision": false
+    },
+    {
+      "id": "gemini-2.5-flash",
+      "name": "Gemini 2.5 Flash",
+      "provider": "google",
+      "tier": "balanced",
+      "context_window": 1048576,
+      "max_output_tokens": 65536,
+      "input_price_per_million": 0.3,
+      "output_price_per_million": 2.5,
+      "reasoning": true,
+      "thinking_type": "enabled",
+      "supports_effort": false,
+      "default_reasoning_effort": null,
+      "description": "Google Gemini 2.5 Flash via Generative Language API (AI Studio key — console aistudio.google.com/app/apikey, or stored Google OAuth). Multimodal, 1M context, thinking-capable. Pricing per ai.google.dev/pricing (2026-06-04).",
+      "aliases": ["gemini-flash", "gemini-2.5-flash-latest"],
+      "supports_vision": true
+    },
+    {
+      "id": "gemini-2.5-pro",
+      "name": "Gemini 2.5 Pro",
+      "provider": "google",
+      "tier": "premium",
+      "context_window": 1048576,
+      "max_output_tokens": 65536,
+      "input_price_per_million": 1.25,
+      "output_price_per_million": 10.0,
+      "reasoning": true,
+      "thinking_type": "enabled",
+      "supports_effort": false,
+      "default_reasoning_effort": null,
+      "description": "Google Gemini 2.5 Pro via Generative Language API (AI Studio key or stored Google OAuth). Most advanced 2.5 reasoning model, multimodal, 1M context. Pricing per ai.google.dev/pricing (2026-06-04).",
+      "aliases": ["gemini-pro", "gemini-2.5-pro-latest"],
+      "supports_vision": true
+    },
+    {
+      "id": "gemini-3.5-flash",
+      "name": "Gemini 3.5 Flash",
+      "provider": "google",
+      "tier": "fast",
+      "context_window": 1048576,
+      "max_output_tokens": 65536,
+      "input_price_per_million": 0.5,
+      "output_price_per_million": 3.0,
+      "reasoning": true,
+      "thinking_type": "enabled",
+      "supports_effort": false,
+      "default_reasoning_effort": null,
+      "description": "Google Gemini 3.5 Flash via Generative Language API. Frontier agentic/coding flash model, multimodal, 1M context, thinking-capable. Pricing APPROXIMATE (verify at ai.google.dev/pricing before cost-sensitive use).",
+      "aliases": ["gemini-flash-3.5"],
+      "supports_vision": true
+    },
+    {
+      "id": "gemini-3.1-flash-lite",
+      "name": "Gemini 3.1 Flash-Lite",
+      "provider": "google",
+      "tier": "fast",
+      "context_window": 1048576,
+      "max_output_tokens": 65536,
+      "input_price_per_million": 0.15,
+      "output_price_per_million": 0.6,
+      "reasoning": true,
+      "thinking_type": "enabled",
+      "supports_effort": false,
+      "default_reasoning_effort": null,
+      "description": "Google Gemini 3.1 Flash-Lite via Generative Language API. Cheapest multimodal Gemini, 1M context, limited thinking. Pricing APPROXIMATE (verify at ai.google.dev/pricing before cost-sensitive use).",
+      "aliases": ["gemini-flash-lite"],
+      "supports_vision": true
+    },
+    {
+      "id": "gemini-3.1-pro-preview",
+      "name": "Gemini 3.1 Pro Preview",
+      "provider": "google",
+      "tier": "premium",
+      "context_window": 1048576,
+      "max_output_tokens": 65536,
+      "input_price_per_million": 2.0,
+      "output_price_per_million": 12.0,
+      "reasoning": true,
+      "thinking_type": "enabled",
+      "supports_effort": false,
+      "default_reasoning_effort": null,
+      "description": "Google Gemini 3.1 Pro Preview via Generative Language API. Strongest agentic/reasoning Gemini, multimodal, 1M context. Preview model — id/pricing may change; pricing APPROXIMATE (verify at ai.google.dev/pricing).",
+      "aliases": ["gemini-pro-preview"],
+      "supports_vision": true
+    },
+    {
+      "id": "grok-4.3",
+      "name": "Grok 4.3 (xAI)",
+      "provider": "xai",
+      "tier": "premium",
+      "context_window": 1048576,
+      "max_output_tokens": 32768,
+      "input_price_per_million": 1.25,
+      "output_price_per_million": 2.5,
+      "reasoning": true,
+      "thinking_type": "enabled",
+      "supports_effort": true,
+      "default_reasoning_effort": "medium",
+      "description": "xAI's current flagship. Excellent agentic tool-calling, low hallucination, supports reasoning + non-reasoning mode. 1M token context. Recommended default for Grok CLI.",
+      "aliases": ["grok-4.3", "grok-latest", "grok-4"],
+      "supports_vision": true
+    },
+    {
+      "id": "grok-build-0.1",
+      "name": "Grok Build 0.1 (xAI)",
+      "provider": "xai",
+      "tier": "balanced",
+      "context_window": 262144,
+      "max_output_tokens": 65536,
+      "input_price_per_million": 1.0,
+      "output_price_per_million": 2.0,
+      "reasoning": true,
+      "thinking_type": "enabled",
+      "supports_effort": false,
+      "default_reasoning_effort": null,
+      "description": "Specialized fast coding model optimized for agentic coding workflows. The model powering Grok Build CLI. Well-suited for programming, debugging, and multi-step tasks.",
+      "aliases": ["grok-build", "grok-code-fast", "grok-build-0.1"],
+      "supports_vision": true
+    }
+  ]
+}