vtcode_config/models/model_id.rs
1use serde::{Deserialize, Serialize};
2
3mod as_str;
4mod capabilities;
5mod collection;
6mod defaults;
7mod description;
8mod display;
9mod format;
10mod openrouter;
11mod parse;
12mod provider;
13
14pub use capabilities::{
15 ModelCatalogEntry, ModelPricing, catalog_provider_keys, model_catalog_entry,
16 supported_models_for_provider,
17};
18
19/// Centralized enum for all supported model identifiers
20#[cfg_attr(feature = "schema", derive(schemars::JsonSchema))]
21#[derive(Clone, Copy, Debug, Default, PartialEq, Eq, Hash, Serialize, Deserialize)]
22pub enum ModelId {
23 // Gemini models
24 /// Gemini 3.1 Pro Preview - Latest Gemini 3.1 Pro flagship
25 Gemini31ProPreview,
26 /// Gemini 3.1 Pro Preview Custom Tools - Optimized for custom tools & bash
27 Gemini31ProPreviewCustomTools,
28 /// Gemini 3.1 Flash Lite Preview - Most cost-efficient model, offering fastest performance for high-frequency, lightweight tasks
29 Gemini31FlashLitePreview,
30 /// Gemini 3.5 Flash - High-efficiency frontier model for fast inference
31 Gemini35Flash,
32 /// Gemini 3 Flash Preview - Our most intelligent model built for speed, combining frontier intelligence with superior search and grounding
33 #[default]
34 Gemini3FlashPreview,
35
36 // OpenAI models
37 /// GPT-5.5 - Next-gen OpenAI model dated release (2026-04-23)
38 GPT55,
39 /// GPT-5 - Latest most capable OpenAI model (2025-08-07)
40 GPT5,
41 /// GPT-5.2 - Latest flagship general-purpose OpenAI model (2025-12-11)
42 GPT52,
43 /// GPT-5.2 Codex - Code-focused GPT-5.2 variant optimized for agentic coding with xhigh reasoning support
44 GPT52Codex,
45 /// GPT-5.4 - Mainline frontier GPT model for general-purpose and coding work
46 GPT54,
47 /// GPT-5.4 Pro - Higher-compute GPT-5.4 variant for difficult problems
48 GPT54Pro,
49 /// GPT-5.4 Nano - Lightweight GPT-5.4 variant optimized for speed and cost-efficiency
50 GPT54Nano,
51 /// GPT-5.4 Mini - Compact GPT-5.4 variant for cost-effective tasks with reduced reasoning overhead
52 GPT54Mini,
53 /// GPT-5.3 Codex - Code-focused GPT-5.3 variant optimized for agentic coding with reasoning effort support (low, medium, high, xhigh)
54 GPT53Codex,
55 /// GPT-5.1 Codex - Code-focused GPT-5.1 variant optimized for agentic coding
56 GPT51Codex,
57 /// GPT-5.1 Codex Max - Higher-compute GPT-5.1 Codex variant for longer-running engineering tasks
58 GPT51CodexMax,
59 /// GPT-5 Mini - Latest efficient OpenAI model (2025-08-07)
60 GPT5Mini,
61 /// GPT-5 Nano - Latest most cost-effective OpenAI model (2025-08-07)
62 GPT5Nano,
63 /// GPT-5 Codex - Code-focused GPT-5 variant optimized for agentic coding
64 GPT5Codex,
65 /// GPT-OSS 20B - OpenAI's open-source 20B parameter model using harmony
66 OpenAIGptOss20b,
67 /// GPT-OSS 120B - OpenAI's open-source 120B parameter model using harmony
68 OpenAIGptOss120b,
69
70 // Anthropic models
71 /// Claude Opus 4.8 - Anthropic's most capable model for complex reasoning and agentic coding
72 ClaudeOpus48,
73 /// Claude Opus 4.7 - Previous flagship Anthropic model with adaptive thinking
74 ClaudeOpus47,
75 /// Claude Opus 4.6 - Previous Anthropic flagship retained for compatibility
76 ClaudeOpus46,
77 /// Claude Sonnet 4.6 - Balanced flagship Anthropic model in VT Code's conservative rollout
78 ClaudeSonnet46,
79 /// Claude Haiku 4.5 - Latest efficient Anthropic model (2025-10-15)
80 ClaudeHaiku45,
81 /// Claude Mythos Preview - Invitation-only Anthropic research preview for defensive cybersecurity workflows
82 ClaudeMythosPreview,
83 /// GitHub Copilot auto model selection
84 CopilotAuto,
85 /// GitHub Copilot GPT-5.2 Codex
86 CopilotGPT52Codex,
87 /// GitHub Copilot GPT-5.1 Codex Max
88 CopilotGPT51CodexMax,
89 /// GitHub Copilot GPT-5.4
90 CopilotGPT54,
91 /// GitHub Copilot GPT-5.4 Mini
92 CopilotGPT54Mini,
93 /// GitHub Copilot Claude Sonnet 4.6
94 CopilotClaudeSonnet46,
95
96 // DeepSeek models
97 /// DeepSeek V4 Pro - High-performance reasoning model with advanced thinking
98 DeepSeekV4Pro,
99 /// DeepSeek V4 Flash - Fast inference model for cost-effective reasoning
100 DeepSeekV4Flash,
101
102 // Mistral AI models
103 /// Mistral Large 3 - State-of-the-art open-weight general-purpose multimodal model
104 MistralLarge3,
105 // Hugging Face models
106 /// DeepSeek V3.2 via Hugging Face router
107 HuggingFaceDeepseekV32,
108 /// OpenAI GPT-OSS 20B via Hugging Face router
109 HuggingFaceOpenAIGptOss20b,
110 /// OpenAI GPT-OSS 120B via Hugging Face router
111 HuggingFaceOpenAIGptOss120b,
112 /// DeepSeek V3.2 via Novita on Hugging Face router
113 HuggingFaceDeepseekV32Novita,
114 /// Xiaomi MiMo-V2-Flash via Novita on Hugging Face router
115 HuggingFaceXiaomiMimoV2FlashNovita,
116 /// MiniMax M2.5 via Novita on Hugging Face router
117 HuggingFaceMinimaxM25Novita,
118 /// Z.AI GLM-5 via Novita on Hugging Face router
119 HuggingFaceGlm5Novita,
120 /// Z.AI GLM-5.1 via zai-org provider on Hugging Face router
121 HuggingFaceGlm51ZaiOrg,
122 /// Qwen3-Coder-Next via Novita inference provider on Hugging Face router
123 HuggingFaceQwen3CoderNextNovita,
124 /// Qwen3.5-397B-A17B via Together inference provider on Hugging Face router
125 HuggingFaceQwen35397BA17BTogether,
126 /// Kimi K2.6 via Novita on Hugging Face router
127 HuggingFaceKimiK26Novita,
128 /// DeepSeek V4 Flash via Novita on Hugging Face router
129 HuggingFaceDeepseekV4FlashNovita,
130 /// DeepSeek V4 Pro via Together on Hugging Face router
131 HuggingFaceDeepseekV4ProTogether,
132 /// Step 3.5 Flash via Hugging Face router
133 HuggingFaceStep35Flash,
134
135 /// GLM-5 - Flagship Z.ai foundation model for complex systems
136 ZaiGlm5,
137 /// GLM-5.1 - Next-gen Z.ai foundation model with improved reasoning
138 ZaiGlm51,
139
140 // MiMo models
141 /// MiMo V2.5 Pro - Xiaomi's flagship reasoning model with 1M context
142 MiMoV25Pro,
143 /// MiMo V2.5 - Xiaomi's omni-modal model with full-modal understanding and 1M context
144 MiMoV25,
145 /// MiMo V2 Flash - Xiaomi's lightweight fast model for high-throughput tasks
146 MiMoV2Flash,
147
148 // Moonshot models
149 /// Kimi K2.6 - Moonshot.ai's latest 1T MoE flagship (32B active, MLA, MoonViT vision)
150 MoonshotKimiK26,
151 /// Kimi K2.5 - Moonshot.ai's previous flagship reasoning model
152 MoonshotKimiK25,
153
154 // OpenCode Zen models
155 /// GPT-5.4 - OpenCode Zen default flagship model
156 OpenCodeZenGPT54,
157 /// GPT-5.4 Mini - Lower-cost OpenCode Zen GPT option
158 OpenCodeZenGPT54Mini,
159 /// Claude Sonnet 4.6 - Anthropic-backed OpenCode Zen coding model
160 OpenCodeZenClaudeSonnet46,
161 /// GLM-5.1 - Z.AI model served through OpenCode Zen
162 OpenCodeZenGlm51,
163 /// Kimi K2.5 - Moonshot model served through OpenCode Zen
164 OpenCodeZenKimiK25,
165
166 // OpenCode Go models
167 /// GLM-5.1 - Z.AI model included with OpenCode Go
168 OpenCodeGoGlm51,
169 /// Kimi K2.5 - Moonshot model included with OpenCode Go
170 OpenCodeGoKimiK25,
171 /// MiniMax M2.5 - OpenCode Go subscription model
172 OpenCodeGoMinimaxM25,
173 /// MiniMax M2.7 - Higher-tier OpenCode Go subscription model
174 OpenCodeGoMinimaxM27,
175
176 // Qwen models
177 /// Qwen 3.7 Max - Alibaba Cloud's flagship reasoning model with 131K context
178 Qwen37Max,
179 /// Qwen 3.6 Flash - Alibaba Cloud's fast inference model with 1M context
180 Qwen36Flash,
181 /// Qwen 3.6 Plus - Alibaba Cloud's balanced model with 131K context
182 Qwen36Plus,
183 /// DeepSeek V4 Flash via Qwen Cloud API
184 QwenDeepSeekV4Flash,
185 /// DeepSeek V4 Pro via Qwen Cloud API
186 QwenDeepSeekV4Pro,
187 /// GLM-5.1 via Qwen Cloud API
188 QwenGlm51,
189
190 // Ollama models
191 /// GPT-OSS 20B - Open-weight GPT-OSS 20B model served via Ollama locally
192 OllamaGptOss20b,
193 /// GPT-OSS 20B Cloud - Cloud-hosted GPT-OSS 20B served via Ollama Cloud
194 OllamaGptOss20bCloud,
195 /// GPT-OSS 120B Cloud - Cloud-hosted GPT-OSS 120B served via Ollama Cloud
196 OllamaGptOss120bCloud,
197 /// Qwen3 1.7B - Qwen3 1.7B model served via Ollama
198 OllamaQwen317b,
199 /// Qwen3 Coder Next - Coding-optimized Qwen3 Next model served via Ollama locally
200 OllamaQwen3CoderNext,
201 /// DeepSeek V3.2 Cloud - DeepSeek V3.2 reasoning deployment via Ollama Cloud
202 OllamaDeepseekV32Cloud,
203 /// DeepSeek V4 Flash Cloud - Fast inference DeepSeek V4 Flash model via Ollama Cloud
204 OllamaDeepseekV4FlashCloud,
205 /// DeepSeek V4 Pro Cloud - High-performance DeepSeek V4 Pro model via Ollama Cloud
206 OllamaDeepseekV4ProCloud,
207 /// Qwen3 Next 80B Cloud - Next-generation Qwen3 80B via Ollama Cloud
208 OllamaQwen3Next80bCloud,
209 /// MiniMax-M2 Cloud - Cloud-hosted MiniMax-M2 model served via Ollama Cloud
210 OllamaMinimaxM2Cloud,
211 /// MiniMax-M2.7 Cloud - Cloud-hosted MiniMax-M2.7 model served via Ollama Cloud
212 OllamaMinimaxM27Cloud,
213 /// GLM-5 Cloud - Cloud-hosted GLM-5 model served via Ollama Cloud
214 OllamaGlm5Cloud,
215 /// GLM-5.1 Cloud - Cloud-hosted GLM-5.1 model served via Ollama Cloud
216 OllamaGlm51Cloud,
217 /// MiniMax-M2.5 Cloud - Cloud-hosted MiniMax-M2.5 model served via Ollama Cloud
218 OllamaMinimaxM25Cloud,
219 /// Gemini 3 Flash Preview Cloud - Google Gemini 3 Flash Preview via Ollama Cloud
220 OllamaGemini3FlashPreviewCloud,
221 /// Kimi K2.6 Cloud - Moonshot Kimi K2.6 via Ollama Cloud
222 OllamaKimiK26Cloud,
223 /// Nemotron 3 Super Cloud - NVIDIA Nemotron 3 Super 120B via Ollama Cloud
224 OllamaNemotron3SuperCloud,
225 /// Laguna XS.2 - Poolside's 33B MoE model (3B activated) for agentic coding via Ollama
226 OllamaLagunaXs2,
227
228 // MiniMax models
229 /// MiniMax-M2.7 - Recursive self-improvement flagship with 204.8K context
230 MinimaxM27,
231 /// MiniMax-M2.5 - Latest MiniMax model with further improvements in reasoning and coding
232 MinimaxM25,
233
234 // OpenRouter models
235 /// Qwen3 32B - Dense 32B Qwen3 deployment
236 OpenRouterQwen332b,
237 /// Qwen3 30B A3B - Active-parameter 30B Qwen3 model
238 OpenRouterQwen330bA3b,
239 /// Qwen3 30B A3B Instruct 2507 - Instruction-tuned Qwen3 30B A3B
240 OpenRouterQwen330bA3bInstruct2507,
241 /// Qwen3 30B A3B Thinking 2507 - Deliberative Qwen3 30B A3B release
242 OpenRouterQwen330bA3bThinking2507,
243 /// Qwen3 14B - Lightweight Qwen3 14B model
244 OpenRouterQwen314b,
245 /// Qwen3 8B - Compact Qwen3 8B deployment
246 OpenRouterQwen38b,
247 /// Qwen3 Next 80B A3B Instruct - Next-generation Qwen3 instruction model
248 OpenRouterQwen3Next80bA3bInstruct,
249 /// Qwen3 Next 80B A3B Thinking - Next-generation Qwen3 reasoning release
250 OpenRouterQwen3Next80bA3bThinking,
251 /// Qwen3.5-397B-A17B - Native vision-language model with linear attention and sparse MoE, 1M context window
252 OpenRouterQwen35Plus0215,
253 /// Qwen3 Coder - Qwen3-based coding model tuned for IDE workflows
254 OpenRouterQwen3Coder,
255 /// Qwen3 Coder Plus - Premium Qwen3 coding model with long context
256 OpenRouterQwen3CoderPlus,
257 /// Qwen3 Coder Flash - Latency optimised Qwen3 coding model
258 OpenRouterQwen3CoderFlash,
259 /// Qwen3 Coder 30B A3B Instruct - Large Mixture-of-Experts coding deployment
260 OpenRouterQwen3Coder30bA3bInstruct,
261 /// Qwen3 Coder Next - Next-generation Qwen3 coding model with enhanced reasoning
262 OpenRouterQwen3CoderNext,
263 /// DeepSeek V4 Pro - High-performance reasoning model via OpenRouter
264 OpenRouterDeepSeekV4Pro,
265 /// DeepSeek V4 Flash - Fast inference model via OpenRouter
266 OpenRouterDeepSeekV4Flash,
267 /// DeepSeek R1 - DeepSeek R1 reasoning model with chain-of-thought
268 OpenRouterDeepSeekR1,
269 /// DeepSeek V3.1 - DeepSeek V3.1 chat model via OpenRouter
270 OpenRouterDeepSeekChatV31,
271 /// OpenAI gpt-oss-120b - Open-weight 120B reasoning model via OpenRouter
272 OpenRouterOpenAIGptOss120b,
273 /// OpenAI gpt-oss-120b:free - Open-weight 120B reasoning model free tier via OpenRouter
274 OpenRouterOpenAIGptOss120bFree,
275 /// OpenAI gpt-oss-20b - Open-weight 20B deployment via OpenRouter
276 OpenRouterOpenAIGptOss20b,
277 /// OpenAI GPT-5 - OpenAI GPT-5 model accessed through OpenRouter
278 OpenRouterOpenAIGpt5,
279 /// OpenAI GPT-5.5 - OpenAI GPT-5.5 model accessed through OpenRouter
280 OpenRouterOpenAIGpt55,
281 /// OpenAI GPT-5 Chat - Chat optimised GPT-5 endpoint without tool use
282 OpenRouterOpenAIGpt5Chat,
283
284 /// Gemini 3.1 Pro Preview - Google's latest Gemini 3.1 Pro model via OpenRouter
285 OpenRouterGoogleGemini31ProPreview,
286
287 /// Claude Sonnet 4.5 - Anthropic Claude Sonnet 4.5 listing
288 OpenRouterAnthropicClaudeSonnet45,
289 /// Claude Sonnet 4.6 - Anthropic Claude Sonnet 4.6 listing
290 OpenRouterAnthropicClaudeSonnet46,
291 /// Claude Haiku 4.5 - Anthropic Claude Haiku 4.5 listing
292 OpenRouterAnthropicClaudeHaiku45,
293 /// Mistral Large 3 2512 - Mistral Large 3 2512 model via OpenRouter
294 OpenRouterMistralaiMistralLarge2512,
295 /// DeepSeek V3.1 Nex N1 - Nex AGI DeepSeek V3.1 Nex N1 model via OpenRouter
296 OpenRouterNexAgiDeepseekV31NexN1,
297 /// Step 3.5 Flash (free) - StepFun's most capable open-source reasoning model via OpenRouter
298 OpenRouterStepfunStep35FlashFree,
299 /// Nemotron 3 Super (free) - NVIDIA's 120B hybrid MoE model via OpenRouter
300 OpenRouterNvidiaNemotron3Super120bA12bFree,
301 /// GLM-5 - Z.AI GLM-5 flagship foundation model via OpenRouter
302 OpenRouterZaiGlm5,
303 /// GLM-4.7 - Z.AI GLM-4.7 model via OpenRouter
304 OpenRouterZaiGlm47,
305 /// GLM-5.1 - Z.AI GLM-5.1 next-gen foundation model via OpenRouter
306 OpenRouterZaiGlm51,
307 /// MiniMax-M2.5 - MiniMax flagship model via OpenRouter
308 OpenRouterMinimaxM25,
309 /// Kimi K2.6 - Moonshot AI's next-generation multimodal model via OpenRouter
310 OpenRouterMoonshotaiKimiK26,
311 /// Qwen3.7 Max - Alibaba's flagship Qwen3.7 model for coding and agentic workloads via OpenRouter
312 OpenRouterQwenQwen37Max,
313 /// Hy3 Preview - Tencent's high-efficiency MoE model for agentic workflows via OpenRouter
314 OpenRouterTencentHy3Preview,
315 /// Grok Build 0.1 - xAI's fast coding model for agentic software engineering via OpenRouter
316 OpenRouterXAiGrokBuild01,
317 /// MiMo-V2.5-Pro - Xiaomi's flagship agentic model for complex software engineering via OpenRouter
318 OpenRouterXiaomiMimoV25Pro,
319 /// Laguna XS.2 (free) - Poolside's efficient free coding agent model via OpenRouter
320 OpenRouterPoolsideLagunaXs2Free,
321 /// Laguna M.1 (free) - Poolside's flagship free coding agent model via OpenRouter
322 OpenRouterPoolsideLagunaM1Free,
323
324 // Poolside models
325 /// Laguna M.1 - Poolside's flagship MoE coding agent model
326 PoolsideLagunaM1,
327 /// Laguna XS.2 - Poolside's efficient MoE coding agent model
328 PoolsideLagunaXs2,
329}