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    /// Meta Muse Wonder 1.0 - Frontier coding, long-running agents, and advanced computer use
88    MetaMuseWonder10,
89    /// Meta Muse Wonder Full 1.0 - Most advanced Meta model for critical workflows (Muse Code subscription)
90    MetaMuseWonderFull10,
91
92    // NVIDIA NIM models
93    /// NVIDIA Nemotron 3 Ultra - NVIDIA's flagship agentic reasoning model via NIM
94    NvidiaNemotron3Ultra550bA55b,
95    /// NVIDIA Nemotron 3 Super - Efficient long-context agentic reasoning model via NIM
96    NvidiaNemotron3Super120bA12b,
97    /// NVIDIA Nemotron 3 Nano - Efficient reasoning and tool-use model via NIM
98    NvidiaNemotron3Nano30bA3b,
99
100    // Merge Gateway routes
101    /// Merge Gateway's default route selected by Merge
102    MergeGatewayDefaultRouting,
103    /// Anthropic Claude Opus 5 through Merge Gateway
104    MergeGatewayAnthropicClaudeOpus5,
105    /// Anthropic Claude Opus 5.5 through Merge Gateway
106    MergeGatewayAnthropicClaudeOpus55,
107    /// Anthropic Claude Sonnet 5 through Merge Gateway
108    MergeGatewayAnthropicClaudeSonnet5,
109    /// Anthropic Claude Sonnet 5.5 through Merge Gateway
110    MergeGatewayAnthropicClaudeSonnet55,
111    /// Google Gemini 3.6 Flash through Merge Gateway
112    /// Google Gemini 3.7 Flash through Merge Gateway
113    /// DeepSeek V4.1 Flash through Merge Gateway
114    MergeGatewayDeepseekFlash,
115    /// xAI Grok 4.6 through Merge Gateway
116    MergeGatewayXaiGrok46,
117    /// xAI Grok 4.7 through Merge Gateway
118    MergeGatewayXaiGrok47,
119    /// MiniMax H3 through Merge Gateway
120    MergeGatewayMinimaxH3,
121    /// Moonshot Kimi K3 through Merge Gateway
122    MergeGatewayMoonshotKimiK3,
123    /// Thinking Machines Inkling through Merge Gateway
124    MergeGatewayThinkingMachinesInkling,
125    /// Z.AI GLM-5.3 Flash through Merge Gateway
126    MergeGatewayZaiGlm53Flash,
127    /// Z.AI GLM-5.3 FlashX through Merge Gateway
128    MergeGatewayZaiGlm53Flashx,
129    /// OpenAI GPT-5.6 Luna through Merge Gateway
130    MergeGatewayOpenAIGpt56Luna,
131    /// OpenAI GPT-5.6 Sol through Merge Gateway
132    MergeGatewayOpenAIGpt56Sol,
133    /// OpenAI GPT-5.6 Terra through Merge Gateway
134    MergeGatewayOpenAIGpt56Terra,
135    /// Google Gemini 3.8 Flash through Merge Gateway
136    MergeGatewayGoogleGemini38Flash,
137    /// Anthropic Claude Haiku 4.5 through Merge Gateway
138    MergeGatewayAnthropicClaudeHaiku4520251001,
139    /// Anthropic Claude Haiku 5.5 through Merge Gateway
140    MergeGatewayAnthropicClaudeHaiku55,
141    /// Anthropic Claude Fable 5.1 through Merge Gateway
142    MergeGatewayAnthropicClaudeFable51,
143    /// OpenAI GPT-6 Astra through Merge Gateway
144    MergeGatewayOpenAIGpt6Astra,
145    /// OpenAI GPT-6 Sol through Merge Gateway
146    MergeGatewayOpenAIGpt6Sol,
147    /// OpenAI GPT-6.1 Sol through Merge Gateway
148    MergeGatewayOpenAIGpt61Sol,
149    /// OpenAI GPT-6 Luna through Merge Gateway
150    MergeGatewayOpenAIGpt6Luna,
151    /// Xiaomi MiMo V2.6 Pro through Merge Gateway
152    MergeGatewayXiaomimimoMimoV26Pro,
153    /// Xiaomi MiMo V2.6 Flash through Merge Gateway
154    MergeGatewayXiaomimimoMimoV26Flash,
155    /// Mistral Large 4 through Merge Gateway
156    MergeGatewayMistralLarge4,
157
158    // Mistral AI models
159    /// Mistral Large 3 - State-of-the-art open-weight general-purpose multimodal model
160    MistralLarge3,
161    /// Mistral Large 4 - Open-weight MoE flagship (49B active / 1.05T total) with 1M context (Public Preview)
162    MistralLarge4,
163    // Hugging Face models
164    /// OpenAI GPT-OSS 20B via Hugging Face router
165    HuggingFaceOpenAIGptOss20b,
166    /// OpenAI GPT-OSS 120B via Hugging Face router
167    HuggingFaceOpenAIGptOss120b,
168    /// Z.AI GLM-5.2 via Novita inference provider on Hugging Face router
169    /// Z.AI GLM-5.3 Flash via Together inference provider on Hugging Face router
170    HuggingFaceGlm53FlashTogether,
171    /// Z.AI GLM-5.3 via Together inference provider on Hugging Face router
172    HuggingFaceGlm53Together,
173    /// Kimi K3 via Together on Hugging Face router
174    HuggingFaceKimiK3Together,
175    /// MiniMax M3 via Novita on Hugging Face router
176    HuggingFaceMinimaxM3Novita,
177
178    // StepFun models
179    /// Step 3.7 Flash - StepFun's flagship multimodal reasoning model with tool calling
180    StepFun37Flash,
181    /// Step 5 Preview - StepFun's frontier model for production-scale Agent applications with 1M context
182    StepFun5Preview,
183
184    // Evolink gateway models (namespaced as `evolink/<model>`)
185    /// GPT-5.2 served through the Evolink gateway
186    /// GPT-5.5 served through the Evolink gateway
187    /// Gemini 3.1 Pro served through the Evolink gateway (OpenAI SDK format)
188    EvolinkGemini31Pro,
189    /// Gemini 3.5 Flash served through the Evolink gateway (OpenAI SDK format)
190    /// MiniMax-M3 served through the Evolink gateway (OpenAI Chat Completions format)
191    EvolinkMinimaxM3,
192    /// Claude Haiku 4.5 served through the Evolink gateway (Anthropic Messages API)
193    EvolinkClaudeHaiku45,
194
195    /// GLM-5.3 - Z.ai flagship coding model with frontier long-horizon agentic performance
196    ZaiGlm53,
197    /// GLM-5.3 Flash - Z.ai efficient multimodal model with hybrid sparse+linear attention, 320B total / 18B active, 1M context, native vision
198    ZaiGlm53Flash,
199    /// 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
200    ZaiGlm53Flashx,
201
202    // MiMo models
203    /// MiMo V2.6 Pro - Xiaomi's flagship reasoning model with 1M context
204    MiMoV26Pro,
205    /// MiMo V2.6 Flash - Xiaomi's efficient high-volume model with 1M context
206    MiMoV26Flash,
207    /// MiMo V2.6 Pro UltraSpeed - Xiaomi's fastest flagship variant with 1M context
208    MiMoV26ProUltraspeed,
209
210    // Moonshot models
211    /// Kimi K3 - Moonshot.ai's 2.8T parameter flagship with Delta Attention, native vision, 1M context
212    MoonshotKimiK3,
213
214    // OpenCode Zen models
215
216    // OpenCode Go models (20 models - https://opencode.ai/docs/go)
217    /// GLM-5.3 - Z.AI flagship for frontier long-horizon coding on OpenCode Go
218    OpenCodeGoGlm53,
219    /// GLM-5.2 - Z.AI flagship model included with OpenCode Go
220    /// GPT-5.6 Luna - OpenAI cost-efficient frontier model on OpenCode Go
221    OpenCodeGoGpt56Luna,
222    /// Kimi K3 - Moonshot flagship 2.8T agentic model on OpenCode Go
223    OpenCodeGoKimiK3,
224    /// MiniMax M3 - Frontier multimodal coding model on OpenCode Go
225    OpenCodeGoMinimaxM3,
226    /// Muse Spark 1.2 Contributor - Meta long-context reasoning on OpenCode Go (limited regions)
227
228    // Qwen models (non-Qwen3 only)
229
230    // Ollama models
231    /// GPT-OSS 20B - Open-weight GPT-OSS 20B model served via Ollama locally
232    OllamaGptOss20b,
233    /// GPT-OSS 20B Cloud - Cloud-hosted GPT-OSS 20B served via Ollama Cloud
234    OllamaGptOss20bCloud,
235    /// GPT-OSS 120B Cloud - Cloud-hosted GPT-OSS 120B served via Ollama Cloud
236    OllamaGptOss120bCloud,
237    /// MiniMax-M3 Cloud - Cloud-hosted MiniMax-M3 model served via Ollama Cloud
238    OllamaMinimaxM3Cloud,
239    /// GLM-5.2 Cloud - Cloud-hosted GLM-5.2 flagship model served via Ollama Cloud
240    /// GLM-5.3 Cloud - Cloud-hosted GLM-5.3 flagship model served via Ollama Cloud
241    OllamaGlm53Cloud,
242    /// Kimi K3 Cloud - Moonshot Kimi K3 via Ollama Cloud
243    OllamaKimiK3Cloud,
244    /// Gemma 4 - Google Gemma 4 model served via Ollama
245    OllamaGemma4,
246    /// Laguna XS.2 - Poolside's 33B MoE model (3B activated) for agentic coding via Ollama
247
248    // llama.cpp models
249    /// Gemma 4 26B A4B - Desktop Gemma 4 MoE model served through llama.cpp
250    LlamaCppGemma426bA4b,
251    /// Gemma 4 E4B - Tiny-footprint Gemma 4 model served through llama.cpp
252    LlamaCppGemma4E4b,
253    /// GPT-OSS 20B - OpenAI open-weight model served through llama.cpp
254    LlamaCppGptOss20b,
255
256    // MiniMax models
257    /// MiniMax-M3 - Frontier multimodal coding model with 1M context
258    MinimaxM3,
259
260    // OpenRouter models
261    /// DeepSeek V4.1 Flash - Latest flash model via OpenRouter
262    OpenRouterDeepSeekFlash,
263    /// OpenAI gpt-oss-120b - Open-weight 120B reasoning model via OpenRouter
264    OpenRouterOpenAIGptOss120b,
265    /// OpenAI gpt-oss-120b:free - Open-weight 120B reasoning model free tier via OpenRouter
266    OpenRouterOpenAIGptOss120bFree,
267    /// OpenAI gpt-oss-20b - Open-weight 20B deployment via OpenRouter
268    OpenRouterOpenAIGptOss20b,
269    /// OpenAI GPT-6 Astra - OpenAI's flagship model for demanding end-to-end work via OpenRouter
270    OpenRouterOpenAIGpt6Astra,
271    /// GPT-6 Sol - Cost-efficient high-end model in the GPT-6 series via OpenRouter
272    OpenRouterOpenAIGpt6Sol,
273    /// GPT-6 Luna - Fast cost-efficient model in the GPT-6 series via OpenRouter
274    OpenRouterOpenAIGpt6Luna,
275
276    /// Meta Muse Glimmer 30B via OpenRouter
277    OpenRouterMetaMuseGlimmer30b,
278    /// Meta Muse Spark 1.2 via OpenRouter
279    /// Meta Muse Spark 1.3 via OpenRouter
280    OpenRouterMetaMuseSpark13,
281    /// Gemini 3.7 Flash - Flash model with 1M context and tunable thinking levels via OpenRouter
282    /// Gemini 3.8 Flash - Most intelligent Flash for long-horizon SWE/agents with 1M context via OpenRouter
283    OpenRouterGoogleGemini38Flash,
284
285    /// Claude Sonnet 5 - Anthropic Claude Sonnet 5 listing
286    OpenRouterAnthropicClaudeSonnet5,
287    /// Mistral Large 3 2512 - Mistral Large 3 2512 model via OpenRouter
288    OpenRouterMistralaiMistralLarge2512,
289    /// DeepSeek V3.1 Nex N1 - Nex AGI DeepSeek V3.1 Nex N1 model via OpenRouter
290    OpenRouterNexAgiDeepseekV31NexN1,
291    /// GLM-5.2 - Z.AI GLM-5.2 flagship model for long-horizon tasks via OpenRouter
292    /// GLM-5.3 Flash - Z.AI efficient multimodal model with hybrid sparse+linear attention via OpenRouter
293    OpenRouterZaiGlm53Flash,
294    /// GLM-5.3 FlashX - Z.AI high-speed Flash variant with faster inference via OpenRouter
295    OpenRouterZaiGlm53Flashx,
296    /// Kimi K3 - Moonshot AI's 2.8T parameter flagship via OpenRouter
297    OpenRouterMoonshotaiKimiK3,
298    /// Grok 4.6 - xAI's flagship reasoning model with reasoning_effort support via OpenRouter
299    OpenRouterXAiGrok46,
300    /// Grok 4.7 - xAI's flagship reasoning model with reasoning_effort support via OpenRouter
301    OpenRouterXAiGrok47,
302    /// MiMo-V2.6-Pro - Xiaomi's flagship agentic model for complex software engineering via OpenRouter
303    OpenRouterXiaomiMimoV26Pro,
304    /// MiMo-V2.6-Flash - Xiaomi's efficient high-volume agentic model via OpenRouter
305    OpenRouterXiaomiMimoV26Flash,
306    /// MiMo-V2.6-Pro-UltraSpeed - Xiaomi's fastest flagship variant via OpenRouter
307    OpenRouterXiaomiMimoV26ProUltraspeed,
308    /// Step 5 Preview - StepFun's flagship agentic model via OpenRouter
309    OpenRouterStepfunStep5Preview,
310
311    // Vercel AI Gateway models (namespaced as `vendor/model` on the gateway)
312    /// Claude Sonnet 5 served through the Vercel AI Gateway
313    VercelAnthropicClaudeSonnet5,
314    /// Claude Opus 5 served through the Vercel AI Gateway
315    VercelAnthropicClaudeOpus5,
316    /// Claude Opus 5.5 served through the Vercel AI Gateway
317    VercelAnthropicClaudeOpus55,
318    /// Claude Haiku 4.5 served through the Vercel AI Gateway
319    VercelAnthropicClaudeHaiku45,
320    /// GPT-5.6 Sol served through the Vercel AI Gateway
321    VercelOpenAiGpt56Sol,
322    /// GPT-6 Astra served through the Vercel AI Gateway
323    VercelOpenAiGpt6Astra,
324    /// GPT-5.6 Luna served through the Vercel AI Gateway
325    VercelOpenAiGpt56Luna,
326    /// GPT-5.3 Codex served through the Vercel AI Gateway
327    /// Gemini 3.1 Pro Preview served through the Vercel AI Gateway
328    /// Gemini 3.8 Flash served through the Vercel AI Gateway
329    VercelGoogleGemini38Flash,
330    /// DeepSeek V4.1 Flash served through the Vercel AI Gateway
331    VercelDeepseekFlash,
332    /// Kimi K3 served through the Vercel AI Gateway
333    VercelMoonshotaiKimiK3,
334    /// MiniMax M3 served through the Vercel AI Gateway
335    VercelMinimaxM3,
336    /// Grok 4.7 served through the Vercel AI Gateway
337    VercelSpacexaiGrok47,
338    // xAI models
339    /// Grok 4.6 - xAI's flagship reasoning model with reasoning_effort support (500k context)
340    XaiGrok46,
341    /// Grok 4.7 - xAI's flagship reasoning model with reasoning_effort support (500k context)
342    XaiGrok47,
343
344    /// User-defined model not in the hardcoded catalog.
345    /// Carries the provider key string and model identifier string.
346    Custom(String, String),
347}
vtcode_config/models/model_id.rs

vtcode_config/models/
model_id.rs
