vtcode_config/models/model_id/
description.rs

1use super::ModelId;
2
3impl ModelId {
4    /// Get a description of the model's characteristics
5    pub fn description(&self) -> &'static str {
6        if let Some(meta) = self.openrouter_metadata() {
7            return meta.description;
8        }
9        match self {
10            // Gemini models
11            ModelId::Gemini31ProPreview => {
12                "Latest Gemini 3.1 Pro flagship model with improved thinking, efficiency, and factual consistency"
13            }
14            ModelId::Gemini31ProPreviewCustomTools => {
15                "Gemini 3.1 Pro variant optimized for agentic workflows using custom tools and bash"
16            }
17            ModelId::Gemini31FlashLitePreview => {
18                "Most cost-efficient Gemini 3.1 model, offering fastest performance for high-frequency, lightweight tasks"
19            }
20            ModelId::Gemini35Flash => {
21                "High-efficiency frontier model for fast inference with excellent quality-to-speed balance"
22            }
23            // OpenAI models
24            ModelId::GPT55 => {
25                "Next-gen OpenAI model with frontier reasoning and long context (2026-04-23 dated release)"
26            }
27            ModelId::GPT54 => {
28                "Mainline frontier GPT model for general-purpose work, coding, long context, and multi-step agents"
29            }
30            ModelId::GPT54Pro => {
31                "Higher-compute GPT-5.4 variant for tougher problems with deeper reasoning"
32            }
33            ModelId::GPT54Nano => {
34                "Lightweight GPT-5.4 variant optimized for speed and cost-efficiency"
35            }
36            ModelId::GPT54Mini => {
37                "Compact GPT-5.4 variant for cost-effective tasks with reduced reasoning overhead"
38            }
39            ModelId::GPT53Codex => {
40                "GPT-5.3 variant optimized for agentic coding tasks with reasoning effort support (low, medium, high, xhigh)"
41            }
42            ModelId::OpenAIGptOss20b => {
43                "OpenAI's open-source 20B parameter GPT-OSS model using harmony tokenization"
44            }
45            ModelId::OpenAIGptOss120b => {
46                "OpenAI's open-source 120B parameter GPT-OSS model using harmony tokenization"
47            }
48            // Anthropic models
49            ModelId::ClaudeOpus48 => {
50                "Anthropic's most capable model for complex reasoning, long-horizon agentic coding, and high-autonomy work"
51            }
52            ModelId::ClaudeSonnet46 => {
53                "Balanced flagship model for coding with budgeted thinking in VT Code's current Anthropic rollout"
54            }
55            ModelId::ClaudeHaiku45 => {
56                "Latest efficient Anthropic model optimized for low-latency agent workflows"
57            }
58            ModelId::ClaudeMythosPreview => {
59                "Invitation-only Anthropic research preview for defensive cybersecurity workflows with adaptive thinking"
60            }
61            ModelId::CopilotAuto => {
62                "GitHub Copilot preview provider with automatic model selection via the official Copilot CLI"
63            }
64            ModelId::CopilotGPT52Codex => {
65                "GitHub Copilot GPT-5.2 Codex option for agentic software engineering workflows"
66            }
67            ModelId::CopilotGPT51CodexMax => {
68                "GitHub Copilot GPT-5.1 Codex Max option for longer-running engineering tasks"
69            }
70            ModelId::CopilotGPT54 => {
71                "GitHub Copilot GPT-5.4 option for complex professional work and long context"
72            }
73            ModelId::CopilotGPT54Mini => {
74                "GitHub Copilot GPT-5.4 Mini option for faster, lighter-weight tasks"
75            }
76            ModelId::CopilotClaudeSonnet46 => {
77                "GitHub Copilot Claude Sonnet 4.6 option for balanced coding and reasoning work"
78            }
79            // DeepSeek models
80            ModelId::DeepSeekV4Pro => {
81                "High-performance reasoning model with advanced thinking capabilities"
82            }
83            ModelId::DeepSeekV4Flash => "Fast inference model for cost-effective reasoning",
84            // Mistral models
85            ModelId::MistralLarge3 => {
86                "State-of-the-art open-weight general-purpose multimodal model with Mixture-of-Experts architecture"
87            }
88            // MiMo models
89            ModelId::MiMoV25Pro => {
90                "Xiaomi's flagship reasoning model with advanced capabilities (1M context)"
91            }
92            ModelId::MiMoV25 => {
93                "Xiaomi's omni-modal model with full-modal understanding and 1M context"
94            }
95            // Z.AI models
96            ModelId::ZaiGlm5 => {
97                "Z.ai flagship GLM-5 foundation model engineered for complex systems design and long-horizon agent workflows"
98            }
99            ModelId::ZaiGlm51 => {
100                "Z.ai next-gen GLM-5.1 foundation model with improved reasoning and agent capabilities"
101            }
102            // Ollama models
103            ModelId::OllamaGptOss20b => {
104                "Local GPT-OSS 20B deployment served via Ollama with no external API dependency"
105            }
106            ModelId::OllamaGptOss20bCloud => {
107                "Cloud-hosted GPT-OSS 20B accessed through Ollama Cloud for efficient reasoning tasks"
108            }
109            ModelId::OllamaGptOss120bCloud => {
110                "Cloud-hosted GPT-OSS 120B accessed through Ollama Cloud for larger reasoning tasks"
111            }
112            ModelId::OllamaQwen317b => {
113                "Qwen3 1.7B served locally through Ollama without external API requirements"
114            }
115            ModelId::OllamaQwen3CoderNext => {
116                "Qwen3-Coder-Next served via Ollama Cloud with 256K context, strong coding/tool-use performance, and non-thinking mode responses"
117            }
118            ModelId::OllamaDeepseekV4FlashCloud => {
119                "DeepSeek V4 Flash cloud deployment via Ollama with fast inference and efficient reasoning"
120            }
121            ModelId::OllamaDeepseekV4ProCloud => {
122                "DeepSeek V4 Pro cloud deployment via Ollama with advanced thinking and strong reasoning"
123            }
124            ModelId::OllamaQwen3Next80bCloud => {
125                "Qwen3 Next generation 80B model via Ollama Cloud with improved reasoning and long context"
126            }
127            ModelId::OllamaGlm5Cloud => "Cloud-hosted GLM-5 model served via Ollama Cloud",
128            ModelId::OllamaGlm51Cloud => "Cloud-hosted GLM-5.1 model served via Ollama Cloud",
129            ModelId::OllamaMinimaxM25Cloud => {
130                "Exceptional multilingual capabilities to elevate code engineering"
131            }
132            ModelId::OllamaGemini3FlashPreviewCloud => {
133                "Gemini 3 Flash offers frontier intelligence built for speed at a fraction of the cost."
134            }
135            ModelId::OllamaKimiK26Cloud => "Cloud-hosted Kimi K2.6 model served via Ollama Cloud",
136            ModelId::OllamaNemotron3SuperCloud => {
137                "NVIDIA Nemotron\u{2122} is a family of open models delivering leading efficiency and accuracy for building specialized AI agents. Nemotron-3-Super (120B) is optimized for collaborative agents and high-volume workloads."
138            }
139            ModelId::OllamaLagunaXs2 => {
140                "Poolside's 33B MoE model with 3B activated parameters, optimized for agentic coding with sliding window attention and native reasoning support (128K context)"
141            }
142            ModelId::LlamaCppQwen3627b => {
143                "Dense Qwen 3.6 local model served through llama.cpp with strong reasoning and coding ability"
144            }
145            ModelId::LlamaCppQwen3635bA3b => {
146                "Qwen 3.6 MoE local model served through llama.cpp for higher reasoning quality with lower active compute"
147            }
148            ModelId::LlamaCppGemma426bA4b => {
149                "Gemma 4 desktop MoE model served through llama.cpp with strong reasoning and fast local inference"
150            }
151            ModelId::LlamaCppGemma4E4b => {
152                "Tiny-footprint Gemma 4 local model served through llama.cpp for phones and low-end laptops"
153            }
154            ModelId::LlamaCppGptOss20b => {
155                "OpenAI's open-weight GPT-OSS 20B model served locally through llama.cpp"
156            }
157            ModelId::LlamaCppStep35Flash => {
158                "StepFun's efficient reasoning model served locally through llama.cpp"
159            }
160            ModelId::OllamaMinimaxM2Cloud => {
161                "Cloud-hosted MiniMax-M2 model accessed through Ollama Cloud for reasoning tasks"
162            }
163            ModelId::OllamaMinimaxM27Cloud => {
164                "Cloud-hosted MiniMax-M2.7 model accessed through Ollama Cloud for reasoning tasks"
165            }
166            ModelId::MinimaxM27 => {
167                "Beginning the journey of recursive self-improvement with 204.8K context and strong reasoning/coding performance"
168            }
169            ModelId::MinimaxM25 => {
170                "Latest MiniMax-M2.5 model with further improvements in reasoning and coding"
171            }
172            // Poolside models
173            ModelId::PoolsideLagunaM1 => {
174                "Poolside's flagship MoE coding agent model with 128K context, optimized for multi-step agentic tasks, tool use, and validation"
175            }
176            ModelId::PoolsideLagunaXs2 => {
177                "Poolside's efficient MoE coding agent model with 128K context, optimized for fast agentic coding with lower resource requirements"
178            }
179            ModelId::MoonshotKimiK26 => {
180                "Kimi K2.6 - Moonshot.ai's latest 1T MoE flagship with 32B active parameters, MLA attention, and MoonViT vision"
181            }
182            ModelId::OpenCodeZenGPT54 => {
183                "OpenCode Zen flagship GPT-5.4 route using OpenCode's curated pay-as-you-go gateway"
184            }
185            ModelId::OpenCodeZenGPT54Mini => {
186                "Lower-cost OpenCode Zen GPT-5.4 Mini option for faster and cheaper tasks"
187            }
188            ModelId::OpenCodeZenClaudeSonnet46 => {
189                "Claude Sonnet 4.6 served through OpenCode Zen's curated Anthropic endpoint"
190            }
191            ModelId::OpenCodeZenGlm51 => {
192                "GLM-5.1 served through OpenCode Zen for lower-cost reasoning and coding work"
193            }
194            ModelId::OpenCodeGoGlm51 => {
195                "GLM-5.1 included with the OpenCode Go subscription for open-model coding workflows"
196            }
197            ModelId::OpenCodeGoMinimaxM25 => {
198                "MiniMax-M2.5 included with the OpenCode Go subscription"
199            }
200            ModelId::OpenCodeGoMinimaxM27 => {
201                "MiniMax-M2.7 included with the OpenCode Go subscription for stronger agentic coding"
202            }
203            ModelId::HuggingFaceOpenAIGptOss20b => "OpenAI GPT-OSS 20B via Hugging Face router",
204            ModelId::HuggingFaceOpenAIGptOss120b => "OpenAI GPT-OSS 120B via Hugging Face router",
205            ModelId::HuggingFaceMinimaxM25Novita => {
206                "MiniMax-M2.5 model via Novita inference provider on HuggingFace router. Enhanced reasoning capabilities."
207            }
208            ModelId::HuggingFaceGlm5Novita => {
209                "Z.ai GLM-5 flagship model via Novita inference provider on HuggingFace router."
210            }
211            ModelId::HuggingFaceGlm51ZaiOrg => {
212                "Z.ai GLM-5.1 model via zai-org inference provider on HuggingFace router."
213            }
214            ModelId::HuggingFaceQwen3CoderNextNovita => {
215                "Qwen3-Coder-Next via Novita inference provider on HuggingFace router. Coding-optimized model with reasoning capabilities."
216            }
217            ModelId::HuggingFaceQwen35397BA17BTogether => {
218                "Qwen3.5-397B-A17B via Together inference provider on HuggingFace router. Vision-language model with linear attention and sparse MoE, 1M context window."
219            }
220            ModelId::HuggingFaceKimiK26Novita => {
221                "Kimi K2.6 via Novita inference provider on HuggingFace router."
222            }
223            ModelId::HuggingFaceDeepseekV4FlashNovita => {
224                "DeepSeek V4 Flash via Novita inference provider on HuggingFace router. Fast inference model for cost-effective reasoning (1M context, 158B params)."
225            }
226            ModelId::HuggingFaceDeepseekV4ProTogether => {
227                "DeepSeek V4 Pro via Together inference provider on HuggingFace router. High-performance reasoning model with advanced thinking capabilities (1M context, 1.6T params)."
228            }
229            ModelId::HuggingFaceStep35Flash => {
230                "Step 3.5 Flash flagship model via HuggingFace router (featherless-ai provider). Supports streaming and fast inference."
231            }
232            ModelId::HuggingFaceGlm51Deepinfra => {
233                "Z.ai GLM-5.1 model via DeepInfra inference provider on HuggingFace router."
234            }
235            ModelId::HuggingFaceMinimaxM27Novita => {
236                "MiniMax-M2.7 model via Novita inference provider on HuggingFace router. Recursive self-improvement with enhanced reasoning."
237            }
238            ModelId::HuggingFaceDeepseekV4ProNovita => {
239                "DeepSeek V4 Pro via Novita inference provider on HuggingFace router. High-performance reasoning model with advanced thinking capabilities (1M context, 1.6T params)."
240            }
241            ModelId::StepFun37Flash => {
242                "StepFun's flagship multimodal reasoning model with 256K context, native image/video input, and tool calling."
243            }
244            // Qwen models
245            ModelId::Qwen37Max => {
246                "Alibaba Cloud's flagship reasoning model with 131K context, advanced thinking, and strong tool-use capabilities"
247            }
248            ModelId::Qwen36Flash => {
249                "Alibaba Cloud's fast inference model with 1M context window, optimized for speed and cost-efficiency"
250            }
251            ModelId::Qwen36Plus => {
252                "Alibaba Cloud's balanced model with 131K context, strong reasoning and coding performance"
253            }
254            ModelId::QwenDeepSeekV4Flash => {
255                "DeepSeek V4 Flash fast inference model served through Qwen Cloud API (1M context)"
256            }
257            ModelId::QwenDeepSeekV4Pro => {
258                "DeepSeek V4 Pro high-performance reasoning model served through Qwen Cloud API (1M context)"
259            }
260            ModelId::QwenGlm51 => {
261                "Z.AI GLM-5.1 next-gen foundation model served through Qwen Cloud API"
262            }
263            ModelId::OpenRouterMinimaxM25 => "MiniMax-M2.5 flagship model via OpenRouter",
264            ModelId::OpenRouterQwen3CoderNext => {
265                "Next-generation Qwen3 coding model optimized for agentic workflows via OpenRouter"
266            }
267            ModelId::OpenRouterMoonshotaiKimiK26 => {
268                "Kimi K2.6 multimodal agentic model for long-horizon coding and design via OpenRouter"
269            }
270            ModelId::OpenRouterZaiGlm51 => "Z.AI GLM-5.1 next-gen foundation model via OpenRouter",
271            ModelId::OpenRouterOpenAIGpt55 => "OpenAI GPT-5.5 model accessed through OpenRouter",
272            _ => unreachable!(),
273        }
274    }
275}
vtcode_config/models/model_id/description.rs

vtcode_config/models/model_id/
description.rs
