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;
13mod table;
14
15pub use capabilities::{
16    ModelCatalogEntry, ModelPricing, catalog_provider_keys, model_catalog_entry, 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, Debug, Default, PartialEq, Eq, Hash, Serialize, Deserialize)]
22pub enum ModelId {
23    // Gemini models
24    /// Gemini 3.6 Flash - Latest flash model with improved capabilities
25    /// Gemini 3.7 Flash - Flash model with 1M context and tunable thinking levels
26    /// Gemini 3.8 Flash - Most intelligent Flash for long-horizon SWE, agents, and enterprise workflows (1M context, 64k output, low/medium/high thinking)
27    Gemini38Flash,
28
29    // OpenAI models
30    /// GPT-6 Astra - Most capable model for hardest end-to-end work with complex reasoning, coding, computer use, and research
31    GPT6Astra,
32    /// GPT-6 Sol - Cost-efficient high-end model in the GPT-6 series for demanding professional work
33    GPT6Sol,
34    /// GPT-6.1 Sol - Near-Astra performance for complex coding, computer use, and professional work at a lower cost
35    GPT61Sol,
36    /// GPT-6 Luna - Fast cost-efficient model in the GPT-6 series for high-volume latency-sensitive workloads
37    GPT6Luna,
38    /// GPT-5.6 Sol - Frontier model for complex professional work in the GPT-5.6 family
39    GPT56Sol,
40    /// GPT-5.6 Terra - GPT-5.6 model that balances intelligence and cost
41    GPT56Terra,
42    /// GPT-5.6 Luna - GPT-5.6 model optimized for cost-sensitive workloads
43    GPT56Luna,
44    /// GPT-OSS 20B - OpenAI's open-source 20B parameter model using harmony
45    OpenAIGptOss20b,
46    /// GPT-OSS 120B - OpenAI's open-source 120B parameter model using harmony
47    OpenAIGptOss120b,
48
49    // Anthropic models
50    /// Claude Sonnet 5 - The best combination of speed and intelligence with adaptive thinking on by default
51    #[default]
52    ClaudeSonnet5,
53    /// Claude Sonnet 5.5 - Latest Sonnet: best speed/intelligence balance, 1M context, 128K output, `between_tools` as the lowest thinking setting, default effort high
54    ClaudeSonnet55,
55    /// Claude Haiku 5.5 - High-volume latency-sensitive work: classification, routing, extraction, subagents; adaptive thinking, 1M context, 128K output, default effort medium
56    ClaudeHaiku55,
57    /// Claude Fable 5 - Anthropic's most capable widely released model for demanding reasoning and long-horizon agentic work
58    ClaudeFable5,
59    /// Claude Fable 5.1 - successor to Fable 5 for demanding reasoning and long-horizon agentic work, 1M context, adaptive thinking always on, cache reads at 1/4 cost
60    ClaudeFable51,
61    /// Claude Opus 5 - Anthropic's newest Opus-tier model with 1M context, thinking on by default
62    ClaudeOpus5,
63    /// Claude Opus 5.5 - Opus-tier successor for long-running agentic coding, adaptive thinking always on, 1M context, 128K output, default effort medium
64    ClaudeOpus55,
65    /// GitHub Copilot auto model selection
66    CopilotAuto,
67    /// GitHub Copilot GPT-5.2 Codex
68    CopilotGPT52Codex,
69    /// GitHub Copilot GPT-5.1 Codex Max
70    CopilotGPT51CodexMax,
71    /// GitHub Copilot GPT-5.4
72    CopilotGPT54,
73    /// GitHub Copilot GPT-5.4 Mini
74    CopilotGPT54Mini,
75
76    // DeepSeek models
77    /// DeepSeek V4.1 Flash - Latest flash model with improved reasoning and efficiency
78    DeepSeekFlash,
79
80    // Official Meta AI models
81    /// Meta Muse Spark 1.1 - Official Meta AI Standard-tier reasoning model
82    MetaMuseSpark11,
83    /// Meta Muse Spark 1.3 - Official Meta AI flagship Standard-tier reasoning model, tuned for agentic workflows
84    MetaMuseSpark13,
85    /// Meta Muse Spark 1.3 Contributor tier - opt-in variant with Meta's discounted data-contribution terms
86    MetaMuseSpark13Contributor,
87
88    // NVIDIA NIM models
89    /// NVIDIA Nemotron 3 Ultra - NVIDIA's flagship agentic reasoning model via NIM
90    NvidiaNemotron3Ultra550bA55b,
91    /// NVIDIA Nemotron 3 Super - Efficient long-context agentic reasoning model via NIM
92    NvidiaNemotron3Super120bA12b,
93    /// NVIDIA Nemotron 3 Nano - Efficient reasoning and tool-use model via NIM
94    NvidiaNemotron3Nano30bA3b,
95
96    // Merge Gateway routes
97    /// Merge Gateway's default route selected by Merge
98    MergeGatewayDefaultRouting,
99    /// Anthropic Claude Opus 5 through Merge Gateway
100    MergeGatewayAnthropicClaudeOpus5,
101    /// Anthropic Claude Opus 5.5 through Merge Gateway
102    MergeGatewayAnthropicClaudeOpus55,
103    /// Anthropic Claude Sonnet 5 through Merge Gateway
104    MergeGatewayAnthropicClaudeSonnet5,
105    /// Anthropic Claude Sonnet 5.5 through Merge Gateway
106    MergeGatewayAnthropicClaudeSonnet55,
107    /// Google Gemini 3.6 Flash through Merge Gateway
108    /// Google Gemini 3.7 Flash through Merge Gateway
109    /// DeepSeek V4.1 Flash through Merge Gateway
110    MergeGatewayDeepseekFlash,
111    /// xAI Grok 4.6 through Merge Gateway
112    MergeGatewayXaiGrok46,
113    /// xAI Grok 4.7 through Merge Gateway
114    MergeGatewayXaiGrok47,
115    /// MiniMax H3 through Merge Gateway
116    MergeGatewayMinimaxH3,
117    /// Moonshot Kimi K3 through Merge Gateway
118    MergeGatewayMoonshotKimiK3,
119    /// Thinking Machines Inkling through Merge Gateway
120    MergeGatewayThinkingMachinesInkling,
121    /// Z.AI GLM-5.3 Flash through Merge Gateway
122    MergeGatewayZaiGlm53Flash,
123    /// Z.AI GLM-5.3 FlashX through Merge Gateway
124    MergeGatewayZaiGlm53Flashx,
125    /// OpenAI GPT-5.6 Luna through Merge Gateway
126    MergeGatewayOpenAIGpt56Luna,
127    /// OpenAI GPT-5.6 Sol through Merge Gateway
128    MergeGatewayOpenAIGpt56Sol,
129    /// OpenAI GPT-5.6 Terra through Merge Gateway
130    MergeGatewayOpenAIGpt56Terra,
131    /// Google Gemini 3.8 Flash through Merge Gateway
132    MergeGatewayGoogleGemini38Flash,
133    /// Anthropic Claude Haiku 4.5 through Merge Gateway
134    MergeGatewayAnthropicClaudeHaiku4520251001,
135    /// Anthropic Claude Haiku 5.5 through Merge Gateway
136    MergeGatewayAnthropicClaudeHaiku55,
137    /// Anthropic Claude Fable 5.1 through Merge Gateway
138    MergeGatewayAnthropicClaudeFable51,
139    /// OpenAI GPT-6 Astra through Merge Gateway
140    MergeGatewayOpenAIGpt6Astra,
141    /// OpenAI GPT-6 Sol through Merge Gateway
142    MergeGatewayOpenAIGpt6Sol,
143    /// OpenAI GPT-6.1 Sol through Merge Gateway
144    MergeGatewayOpenAIGpt61Sol,
145    /// OpenAI GPT-6 Luna through Merge Gateway
146    MergeGatewayOpenAIGpt6Luna,
147    /// Xiaomi MiMo V2.6 Pro through Merge Gateway
148    MergeGatewayXiaomimimoMimoV26Pro,
149    /// Xiaomi MiMo V2.6 Flash through Merge Gateway
150    MergeGatewayXiaomimimoMimoV26Flash,
151    /// Mistral Large 4 through Merge Gateway
152    MergeGatewayMistralLarge4,
153
154    // Mistral AI models
155    /// Mistral Large 3 - State-of-the-art open-weight general-purpose multimodal model
156    MistralLarge3,
157    /// Mistral Large 4 - Open-weight MoE flagship (49B active / 1.05T total) with 1M context (Public Preview)
158    MistralLarge4,
159    // Hugging Face models
160    /// OpenAI GPT-OSS 20B via Hugging Face router
161    HuggingFaceOpenAIGptOss20b,
162    /// OpenAI GPT-OSS 120B via Hugging Face router
163    HuggingFaceOpenAIGptOss120b,
164    /// Z.AI GLM-5.2 via Novita inference provider on Hugging Face router
165    /// Z.AI GLM-5.3 Flash via Together inference provider on Hugging Face router
166    HuggingFaceGlm53FlashTogether,
167    /// Z.AI GLM-5.3 via Together inference provider on Hugging Face router
168    HuggingFaceGlm53Together,
169    /// Kimi K3 via Together on Hugging Face router
170    HuggingFaceKimiK3Together,
171    /// MiniMax M3 via Novita on Hugging Face router
172    HuggingFaceMinimaxM3Novita,
173
174    // StepFun models
175    /// Step 3.7 Flash - StepFun's flagship multimodal reasoning model with tool calling
176    StepFun37Flash,
177    /// Step 5 Preview - StepFun's frontier model for production-scale Agent applications with 1M context
178    StepFun5Preview,
179
180    // Evolink gateway models (namespaced as `evolink/<model>`)
181    /// GPT-5.2 served through the Evolink gateway
182    /// GPT-5.5 served through the Evolink gateway
183    /// Gemini 3.1 Pro served through the Evolink gateway (OpenAI SDK format)
184    EvolinkGemini31Pro,
185    /// Gemini 3.5 Flash served through the Evolink gateway (OpenAI SDK format)
186    /// MiniMax-M3 served through the Evolink gateway (OpenAI Chat Completions format)
187    EvolinkMinimaxM3,
188    /// Claude Haiku 4.5 served through the Evolink gateway (Anthropic Messages API)
189    EvolinkClaudeHaiku45,
190
191    /// GLM-5.3 - Z.ai flagship coding model with frontier long-horizon agentic performance
192    ZaiGlm53,
193    /// GLM-5.3 Flash - Z.ai efficient multimodal model with hybrid sparse+linear attention, 320B total / 18B active, 1M context, native vision
194    ZaiGlm53Flash,
195    /// GLM-5.3 FlashX - Z.ai high-speed Flash variant with faster inference (up to 200 tok/s), 320B total / 18B active, 1M context, native vision
196    ZaiGlm53Flashx,
197
198    // MiMo models
199    /// MiMo V2.6 Pro - Xiaomi's flagship reasoning model with 1M context
200    MiMoV26Pro,
201    /// MiMo V2.6 Flash - Xiaomi's efficient high-volume model with 1M context
202    MiMoV26Flash,
203    /// MiMo V2.6 Pro UltraSpeed - Xiaomi's fastest flagship variant with 1M context
204    MiMoV26ProUltraspeed,
205
206    // Moonshot models
207    /// Kimi K3 - Moonshot.ai's 2.8T parameter flagship with Delta Attention, native vision, 1M context
208    MoonshotKimiK3,
209
210    // OpenCode Zen models
211
212    // OpenCode Go models (20 models - https://opencode.ai/docs/go)
213    /// GLM-5.3 - Z.AI flagship for frontier long-horizon coding on OpenCode Go
214    OpenCodeGoGlm53,
215    /// GLM-5.2 - Z.AI flagship model included with OpenCode Go
216    /// GPT-5.6 Luna - OpenAI cost-efficient frontier model on OpenCode Go
217    OpenCodeGoGpt56Luna,
218    /// Kimi K3 - Moonshot flagship 2.8T agentic model on OpenCode Go
219    OpenCodeGoKimiK3,
220    /// MiniMax M3 - Frontier multimodal coding model on OpenCode Go
221    OpenCodeGoMinimaxM3,
222    /// Muse Spark 1.2 Contributor - Meta long-context reasoning on OpenCode Go (limited regions)
223
224    // Qwen models (non-Qwen3 only)
225
226    // Ollama models
227    /// GPT-OSS 20B - Open-weight GPT-OSS 20B model served via Ollama locally
228    OllamaGptOss20b,
229    /// GPT-OSS 20B Cloud - Cloud-hosted GPT-OSS 20B served via Ollama Cloud
230    OllamaGptOss20bCloud,
231    /// GPT-OSS 120B Cloud - Cloud-hosted GPT-OSS 120B served via Ollama Cloud
232    OllamaGptOss120bCloud,
233    /// MiniMax-M3 Cloud - Cloud-hosted MiniMax-M3 model served via Ollama Cloud
234    OllamaMinimaxM3Cloud,
235    /// GLM-5.2 Cloud - Cloud-hosted GLM-5.2 flagship model served via Ollama Cloud
236    /// GLM-5.3 Cloud - Cloud-hosted GLM-5.3 flagship model served via Ollama Cloud
237    OllamaGlm53Cloud,
238    /// Kimi K3 Cloud - Moonshot Kimi K3 via Ollama Cloud
239    OllamaKimiK3Cloud,
240    /// Gemma 4 - Google Gemma 4 model served via Ollama
241    OllamaGemma4,
242    /// Laguna XS.2 - Poolside's 33B MoE model (3B activated) for agentic coding via Ollama
243
244    // llama.cpp models
245    /// Gemma 4 26B A4B - Desktop Gemma 4 MoE model served through llama.cpp
246    LlamaCppGemma426bA4b,
247    /// Gemma 4 E4B - Tiny-footprint Gemma 4 model served through llama.cpp
248    LlamaCppGemma4E4b,
249    /// GPT-OSS 20B - OpenAI open-weight model served through llama.cpp
250    LlamaCppGptOss20b,
251
252    // MiniMax models
253    /// MiniMax-M3 - Frontier multimodal coding model with 1M context
254    MinimaxM3,
255
256    // OpenRouter models
257    /// DeepSeek V4.1 Flash - Latest flash model via OpenRouter
258    OpenRouterDeepSeekFlash,
259    /// OpenAI gpt-oss-120b - Open-weight 120B reasoning model via OpenRouter
260    OpenRouterOpenAIGptOss120b,
261    /// OpenAI gpt-oss-120b:free - Open-weight 120B reasoning model free tier via OpenRouter
262    OpenRouterOpenAIGptOss120bFree,
263    /// OpenAI gpt-oss-20b - Open-weight 20B deployment via OpenRouter
264    OpenRouterOpenAIGptOss20b,
265    /// OpenAI GPT-6 Astra - OpenAI's flagship model for demanding end-to-end work via OpenRouter
266    OpenRouterOpenAIGpt6Astra,
267    /// GPT-6 Sol - Cost-efficient high-end model in the GPT-6 series via OpenRouter
268    OpenRouterOpenAIGpt6Sol,
269    /// GPT-6 Luna - Fast cost-efficient model in the GPT-6 series via OpenRouter
270    OpenRouterOpenAIGpt6Luna,
271
272    /// Meta Muse Glimmer 30B via OpenRouter
273    OpenRouterMetaMuseGlimmer30b,
274    /// Meta Muse Spark 1.2 via OpenRouter
275    /// Meta Muse Spark 1.3 via OpenRouter
276    OpenRouterMetaMuseSpark13,
277    /// Gemini 3.7 Flash - Flash model with 1M context and tunable thinking levels via OpenRouter
278    /// Gemini 3.8 Flash - Most intelligent Flash for long-horizon SWE/agents with 1M context via OpenRouter
279    OpenRouterGoogleGemini38Flash,
280
281    /// Claude Sonnet 5 - Anthropic Claude Sonnet 5 listing
282    OpenRouterAnthropicClaudeSonnet5,
283    /// Mistral Large 3 2512 - Mistral Large 3 2512 model via OpenRouter
284    OpenRouterMistralaiMistralLarge2512,
285    /// DeepSeek V3.1 Nex N1 - Nex AGI DeepSeek V3.1 Nex N1 model via OpenRouter
286    OpenRouterNexAgiDeepseekV31NexN1,
287    /// GLM-5.2 - Z.AI GLM-5.2 flagship model for long-horizon tasks via OpenRouter
288    /// GLM-5.3 Flash - Z.AI efficient multimodal model with hybrid sparse+linear attention via OpenRouter
289    OpenRouterZaiGlm53Flash,
290    /// GLM-5.3 FlashX - Z.AI high-speed Flash variant with faster inference via OpenRouter
291    OpenRouterZaiGlm53Flashx,
292    /// Kimi K3 - Moonshot AI's 2.8T parameter flagship via OpenRouter
293    OpenRouterMoonshotaiKimiK3,
294    /// Grok 4.6 - xAI's flagship reasoning model with reasoning_effort support via OpenRouter
295    OpenRouterXAiGrok46,
296    /// Grok 4.7 - xAI's flagship reasoning model with reasoning_effort support via OpenRouter
297    OpenRouterXAiGrok47,
298    /// MiMo-V2.6-Pro - Xiaomi's flagship agentic model for complex software engineering via OpenRouter
299    OpenRouterXiaomiMimoV26Pro,
300    /// MiMo-V2.6-Flash - Xiaomi's efficient high-volume agentic model via OpenRouter
301    OpenRouterXiaomiMimoV26Flash,
302    /// MiMo-V2.6-Pro-UltraSpeed - Xiaomi's fastest flagship variant via OpenRouter
303    OpenRouterXiaomiMimoV26ProUltraspeed,
304    /// Step 5 Preview - StepFun's flagship agentic model via OpenRouter
305    OpenRouterStepfunStep5Preview,
306
307    // Vercel AI Gateway models (namespaced as `vendor/model` on the gateway)
308    /// Claude Sonnet 5 served through the Vercel AI Gateway
309    VercelAnthropicClaudeSonnet5,
310    /// Claude Opus 5 served through the Vercel AI Gateway
311    VercelAnthropicClaudeOpus5,
312    /// Claude Opus 5.5 served through the Vercel AI Gateway
313    VercelAnthropicClaudeOpus55,
314    /// Claude Haiku 4.5 served through the Vercel AI Gateway
315    VercelAnthropicClaudeHaiku45,
316    /// GPT-5.6 Sol served through the Vercel AI Gateway
317    VercelOpenAiGpt56Sol,
318    /// GPT-6 Astra served through the Vercel AI Gateway
319    VercelOpenAiGpt6Astra,
320    /// GPT-5.6 Luna served through the Vercel AI Gateway
321    VercelOpenAiGpt56Luna,
322    /// GPT-5.3 Codex served through the Vercel AI Gateway
323    /// Gemini 3.1 Pro Preview served through the Vercel AI Gateway
324    /// Gemini 3.8 Flash served through the Vercel AI Gateway
325    VercelGoogleGemini38Flash,
326    /// DeepSeek V4.1 Flash served through the Vercel AI Gateway
327    VercelDeepseekFlash,
328    /// Kimi K3 served through the Vercel AI Gateway
329    VercelMoonshotaiKimiK3,
330    /// MiniMax M3 served through the Vercel AI Gateway
331    VercelMinimaxM3,
332    /// Grok 4.7 served through the Vercel AI Gateway
333    VercelSpacexaiGrok47,
334    // xAI models
335    /// Grok 4.6 - xAI's flagship reasoning model with reasoning_effort support (500k context)
336    XaiGrok46,
337    /// Grok 4.7 - xAI's flagship reasoning model with reasoning_effort support (500k context)
338    XaiGrok47,
339
340    /// User-defined model not in the hardcoded catalog.
341    /// Carries the provider key string and model identifier string.
342    Custom(String, String),
343}
vtcode_config/models/model_id.rs

vtcode_config/models/
model_id.rs
