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    /// Gemini 4 Argon - Flagship reasoning model for the hardest coding, agentic, math, and science problems (unified 1M context window, 64K sync output, long decode, low/medium/high thinking)
29    Gemini4Argon,
30
31    // OpenAI models
32    /// GPT-6 Astra - Most capable model for hardest end-to-end work with complex reasoning, coding, computer use, and research
33    GPT6Astra,
34    /// GPT-6 Sol - Cost-efficient high-end model in the GPT-6 series for demanding professional work
35    GPT6Sol,
36    /// GPT-6.1 Sol - Near-Astra performance for complex coding, computer use, and professional work at a lower cost
37    GPT61Sol,
38    /// GPT-6 Luna - Fast cost-efficient model in the GPT-6 series for high-volume latency-sensitive workloads
39    GPT6Luna,
40    /// GPT-5.6 Sol - Frontier model for complex professional work in the GPT-5.6 family
41    GPT56Sol,
42    /// GPT-5.6 Terra - GPT-5.6 model that balances intelligence and cost
43    GPT56Terra,
44    /// GPT-5.6 Luna - GPT-5.6 model optimized for cost-sensitive workloads
45    GPT56Luna,
46    /// GPT-OSS 20B - OpenAI's open-source 20B parameter model using harmony
47    OpenAIGptOss20b,
48    /// GPT-OSS 120B - OpenAI's open-source 120B parameter model using harmony
49    OpenAIGptOss120b,
50
51    // Anthropic models
52    /// Claude Sonnet 5 - The best combination of speed and intelligence with adaptive thinking on by default
53    #[default]
54    ClaudeSonnet5,
55    /// Claude Sonnet 5.5 - Latest Sonnet: best speed/intelligence balance, 1M context, 128K output, `between_tools` as the lowest thinking setting, default effort high
56    ClaudeSonnet55,
57    /// Claude Haiku 5.5 - High-volume latency-sensitive work: classification, routing, extraction, subagents; adaptive thinking, 1M context, 128K output, default effort medium
58    ClaudeHaiku55,
59    /// Claude Fable 5 - Anthropic's most capable widely released model for demanding reasoning and long-horizon agentic work
60    ClaudeFable5,
61    /// 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
62    ClaudeFable51,
63    /// Claude Opus 5 - Anthropic's newest Opus-tier model with 1M context, thinking on by default
64    ClaudeOpus5,
65    /// Claude Opus 5.5 - Opus-tier successor for long-running agentic coding, adaptive thinking always on, 1M context, 128K output, default effort medium
66    ClaudeOpus55,
67    /// GitHub Copilot auto model selection
68    CopilotAuto,
69    /// GitHub Copilot GPT-5.2 Codex
70    CopilotGPT52Codex,
71    /// GitHub Copilot GPT-5.1 Codex Max
72    CopilotGPT51CodexMax,
73    /// GitHub Copilot GPT-5.4
74    CopilotGPT54,
75    /// GitHub Copilot GPT-5.4 Mini
76    CopilotGPT54Mini,
77
78    // DeepSeek models
79    /// DeepSeek V4.1 Flash - Latest flash model with improved reasoning and efficiency
80    DeepSeekFlash,
81
82    // Official Meta AI models
83    /// Meta Muse Spark 1.1 - Official Meta AI Standard-tier reasoning model
84    MetaMuseSpark11,
85    /// Meta Muse Spark 1.3 - Official Meta AI flagship Standard-tier reasoning model, tuned for agentic workflows
86    MetaMuseSpark13,
87    /// Meta Muse Spark 1.3 Contributor tier - opt-in variant with Meta's discounted data-contribution terms
88    MetaMuseSpark13Contributor,
89    /// Meta Muse Wonder 1.0 - Frontier coding, long-running agents, and advanced computer use
90    MetaMuseWonder10,
91    /// Meta Muse Wonder Full 1.0 - Most advanced Meta model for critical workflows (Muse Code subscription)
92    MetaMuseWonderFull10,
93
94    // NVIDIA NIM models
95    /// NVIDIA Nemotron 3 Ultra - NVIDIA's flagship agentic reasoning model via NIM
96    NvidiaNemotron3Ultra550bA55b,
97    /// NVIDIA Nemotron 3 Super - Efficient long-context agentic reasoning model via NIM
98    NvidiaNemotron3Super120bA12b,
99    /// NVIDIA Nemotron 3 Nano - Efficient reasoning and tool-use model via NIM
100    NvidiaNemotron3Nano30bA3b,
101
102    // Merge Gateway routes
103    /// Merge Gateway's default route selected by Merge
104    MergeGatewayDefaultRouting,
105    /// Anthropic Claude Opus 5 through Merge Gateway
106    MergeGatewayAnthropicClaudeOpus5,
107    /// Anthropic Claude Opus 5.5 through Merge Gateway
108    MergeGatewayAnthropicClaudeOpus55,
109    /// Anthropic Claude Sonnet 5 through Merge Gateway
110    MergeGatewayAnthropicClaudeSonnet5,
111    /// Anthropic Claude Sonnet 5.5 through Merge Gateway
112    MergeGatewayAnthropicClaudeSonnet55,
113    /// Google Gemini 3.6 Flash through Merge Gateway
114    /// Google Gemini 3.7 Flash through Merge Gateway
115    /// DeepSeek V4.1 Flash through Merge Gateway
116    MergeGatewayDeepseekFlash,
117    /// xAI Grok 4.6 through Merge Gateway
118    MergeGatewayXaiGrok46,
119    /// xAI Grok 4.7 through Merge Gateway
120    MergeGatewayXaiGrok47,
121    /// MiniMax H3 through Merge Gateway
122    MergeGatewayMinimaxH3,
123    /// MiniMax M3.1 Flash through Merge Gateway
124    MergeGatewayMinimaxM31Flash,
125    /// Moonshot Kimi K3 through Merge Gateway
126    MergeGatewayMoonshotKimiK3,
127    /// Thinking Machines Inkling through Merge Gateway
128    MergeGatewayThinkingMachinesInkling,
129    /// Z.AI GLM-5.3 Flash through Merge Gateway
130    MergeGatewayZaiGlm53Flash,
131    /// Z.AI GLM-5.3 FlashX through Merge Gateway
132    MergeGatewayZaiGlm53Flashx,
133    /// OpenAI GPT-5.6 Luna through Merge Gateway
134    MergeGatewayOpenAIGpt56Luna,
135    /// OpenAI GPT-5.6 Sol through Merge Gateway
136    MergeGatewayOpenAIGpt56Sol,
137    /// OpenAI GPT-5.6 Terra through Merge Gateway
138    MergeGatewayOpenAIGpt56Terra,
139    /// Google Gemini 3.8 Flash through Merge Gateway
140    MergeGatewayGoogleGemini38Flash,
141    /// Google Gemini 4 Argon through Merge Gateway
142    MergeGatewayGoogleGemini4Argon,
143    /// Anthropic Claude Haiku 4.5 through Merge Gateway
144    MergeGatewayAnthropicClaudeHaiku4520251001,
145    /// Anthropic Claude Haiku 5.5 through Merge Gateway
146    MergeGatewayAnthropicClaudeHaiku55,
147    /// Anthropic Claude Fable 5.1 through Merge Gateway
148    MergeGatewayAnthropicClaudeFable51,
149    /// OpenAI GPT-6 Astra through Merge Gateway
150    MergeGatewayOpenAIGpt6Astra,
151    /// OpenAI GPT-6 Sol through Merge Gateway
152    MergeGatewayOpenAIGpt6Sol,
153    /// OpenAI GPT-6.1 Sol through Merge Gateway
154    MergeGatewayOpenAIGpt61Sol,
155    /// OpenAI GPT-6 Luna through Merge Gateway
156    MergeGatewayOpenAIGpt6Luna,
157    /// Xiaomi MiMo V2.6 Pro through Merge Gateway
158    MergeGatewayXiaomimimoMimoV26Pro,
159    /// Xiaomi MiMo V2.6 Flash through Merge Gateway
160    MergeGatewayXiaomimimoMimoV26Flash,
161    /// Mistral Large 4 through Merge Gateway
162    MergeGatewayMistralLarge4,
163    /// Meta Muse Spark Open (816B/42B) through Merge Gateway
164    MergeGatewayMetaMuseSparkOpen,
165
166    // Mistral AI models
167    /// Mistral Large 3 - State-of-the-art open-weight general-purpose multimodal model
168    MistralLarge3,
169    /// Mistral Large 4 - Open-weight MoE flagship (49B active / 1.05T total) with 1M context (Public Preview)
170    MistralLarge4,
171    // Hugging Face models
172    /// OpenAI GPT-OSS 20B via Hugging Face router
173    HuggingFaceOpenAIGptOss20b,
174    /// OpenAI GPT-OSS 120B via Hugging Face router
175    HuggingFaceOpenAIGptOss120b,
176    /// Z.AI GLM-5.2 via Novita inference provider on Hugging Face router
177    /// Z.AI GLM-5.3 Flash via Together inference provider on Hugging Face router
178    HuggingFaceGlm53FlashTogether,
179    /// Z.AI GLM-5.3 via Together inference provider on Hugging Face router
180    HuggingFaceGlm53Together,
181    /// Kimi K3 via Together on Hugging Face router
182    HuggingFaceKimiK3Together,
183    /// MiniMax M3 via Novita on Hugging Face router
184    HuggingFaceMinimaxM3Novita,
185    /// Meta Muse Spark Open 816B-A42B via Baseten on Hugging Face router
186    HuggingFaceMuseSparkOpenBaseten,
187
188    // StepFun models
189    /// Step 3.7 Flash - StepFun's flagship multimodal reasoning model with tool calling
190    StepFun37Flash,
191    /// Step 5 Preview - StepFun's frontier model for production-scale Agent applications with 1M context
192    StepFun5Preview,
193
194    // Evolink gateway models (namespaced as `evolink/<model>`)
195    /// GPT-5.2 served through the Evolink gateway
196    /// GPT-5.5 served through the Evolink gateway
197    /// Gemini 3.1 Pro served through the Evolink gateway (OpenAI SDK format)
198    EvolinkGemini31Pro,
199    /// Gemini 3.5 Flash served through the Evolink gateway (OpenAI SDK format)
200    /// MiniMax-M3 served through the Evolink gateway (OpenAI Chat Completions format)
201    EvolinkMinimaxM3,
202    /// Claude Haiku 4.5 served through the Evolink gateway (Anthropic Messages API)
203    EvolinkClaudeHaiku45,
204
205    /// GLM-5.3 - Z.ai flagship coding model with frontier long-horizon agentic performance
206    ZaiGlm53,
207    /// GLM-5.3 Flash - Z.ai efficient multimodal model with hybrid sparse+linear attention, 320B total / 18B active, 1M context, native vision
208    ZaiGlm53Flash,
209    /// 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
210    ZaiGlm53Flashx,
211
212    // MiMo models
213    /// MiMo V2.6 Pro - Xiaomi's flagship reasoning model with 1M context
214    MiMoV26Pro,
215    /// MiMo V2.6 Flash - Xiaomi's efficient high-volume model with 1M context
216    MiMoV26Flash,
217    /// MiMo V2.6 Pro UltraSpeed - Xiaomi's fastest flagship variant with 1M context
218    MiMoV26ProUltraspeed,
219
220    // Moonshot models
221    /// Kimi K3 - Moonshot.ai's 2.8T parameter flagship with Delta Attention, native vision, 1M context
222    MoonshotKimiK3,
223
224    // OpenCode Zen models
225
226    // OpenCode Go models (20 models - https://opencode.ai/docs/go)
227    /// GLM-5.3 - Z.AI flagship for frontier long-horizon coding on OpenCode Go
228    OpenCodeGoGlm53,
229    /// GLM-5.2 - Z.AI flagship model included with OpenCode Go
230    /// GPT-5.6 Luna - OpenAI cost-efficient frontier model on OpenCode Go
231    OpenCodeGoGpt56Luna,
232    /// Kimi K3 - Moonshot flagship 2.8T agentic model on OpenCode Go
233    OpenCodeGoKimiK3,
234    /// MiniMax M3 - Frontier multimodal coding model on OpenCode Go
235    OpenCodeGoMinimaxM3,
236    /// Muse Spark 1.2 Contributor - Meta long-context reasoning on OpenCode Go (limited regions)
237
238    // Qwen models (non-Qwen3 only)
239
240    // Ollama models
241    /// GPT-OSS 20B - Open-weight GPT-OSS 20B model served via Ollama locally
242    OllamaGptOss20b,
243    /// GPT-OSS 20B Cloud - Cloud-hosted GPT-OSS 20B served via Ollama Cloud
244    OllamaGptOss20bCloud,
245    /// GPT-OSS 120B Cloud - Cloud-hosted GPT-OSS 120B served via Ollama Cloud
246    OllamaGptOss120bCloud,
247    /// MiniMax-M3 Cloud - Cloud-hosted MiniMax-M3 model served via Ollama Cloud
248    OllamaMinimaxM3Cloud,
249    /// MiniMax M3.1 Flash Cloud - ultra-efficient multimodal MiniMax M3.1 Flash served via Ollama Cloud
250    OllamaMinimaxM31FlashCloud,
251    /// GLM-5.2 Cloud - Cloud-hosted GLM-5.2 flagship model served via Ollama Cloud
252    /// GLM-5.3 Cloud - Cloud-hosted GLM-5.3 flagship model served via Ollama Cloud
253    OllamaGlm53Cloud,
254    /// GLM-5.3 Flash Cloud - Multimodal GLM-5.3 Flash served via Ollama Cloud
255    OllamaGlm53FlashCloud,
256    /// Mistral Large 4 Cloud - Multimodal Mistral Large 4 served via Ollama Cloud
257    OllamaMistralLarge4Cloud,
258    /// Muse Spark Open Cloud - Multimodal Muse Spark Open served via Ollama Cloud
259    OllamaMuseSparkOpenCloud,
260    /// DeepSeek V4.1 Flash Cloud - Multimodal DeepSeek V4.1 Flash served via Ollama Cloud
261    OllamaDeepSeekV41FlashCloud,
262    /// Kimi K3 Cloud - Moonshot Kimi K3 via Ollama Cloud
263    OllamaKimiK3Cloud,
264    /// Gemma 4 - Google Gemma 4 model served via Ollama
265    OllamaGemma4,
266    /// Laguna XS.2 - Poolside's 33B MoE model (3B activated) for agentic coding via Ollama
267
268    // llama.cpp models
269    /// Gemma 4 26B A4B - Desktop Gemma 4 MoE model served through llama.cpp
270    LlamaCppGemma426bA4b,
271    /// Gemma 4 E4B - Tiny-footprint Gemma 4 model served through llama.cpp
272    LlamaCppGemma4E4b,
273    /// GPT-OSS 20B - OpenAI open-weight model served through llama.cpp
274    LlamaCppGptOss20b,
275
276    // MiniMax models
277    /// MiniMax-M3 - Frontier multimodal coding model with 1M context
278    MinimaxM3,
279    /// MiniMax-M3.1-Flash - Latest ultra-efficient multimodal model via OpenRouter
280    OpenRouterMinimaxM31FlashFree,
281    /// MiniMax-M3.1-Flash - Latest ultra-efficient multimodal model with 1M context and tunable thinking depth
282    MinimaxM31Flash,
283
284    // OpenRouter models
285    /// DeepSeek V4.1 Flash 1017 - Latest flash model via OpenRouter
286    OpenRouterDeepSeekFlash,
287    /// OpenAI gpt-oss-120b - Open-weight 120B reasoning model via OpenRouter
288    OpenRouterOpenAIGptOss120b,
289    /// OpenAI gpt-oss-120b:free - Open-weight 120B reasoning model free tier via OpenRouter
290    OpenRouterOpenAIGptOss120bFree,
291    /// OpenAI gpt-oss-20b - Open-weight 20B deployment via OpenRouter
292    OpenRouterOpenAIGptOss20b,
293    /// OpenAI GPT-6 Astra - OpenAI's flagship model for demanding end-to-end work via OpenRouter
294    OpenRouterOpenAIGpt6Astra,
295    /// GPT-6 Sol - Cost-efficient high-end model in the GPT-6 series via OpenRouter
296    OpenRouterOpenAIGpt6Sol,
297    /// GPT-6 Luna - Fast cost-efficient model in the GPT-6 series via OpenRouter
298    OpenRouterOpenAIGpt6Luna,
299
300    /// Meta Muse Glimmer 30B via OpenRouter
301    OpenRouterMetaMuseGlimmer30b,
302    /// Meta Muse Spark 1.2 via OpenRouter
303    /// Meta Muse Spark 1.3 via OpenRouter
304    OpenRouterMetaMuseSpark13,
305    /// Meta Muse Spark Open 816B-A42B via OpenRouter
306    OpenRouterMetaMuseSparkOpen,
307    /// Gemini 3.7 Flash - Flash model with 1M context and tunable thinking levels via OpenRouter
308    /// Gemini 3.8 Flash - Most intelligent Flash for long-horizon SWE/agents with 1M context via OpenRouter
309    OpenRouterGoogleGemini38Flash,
310
311    /// Claude Sonnet 5 - Anthropic Claude Sonnet 5 listing
312    OpenRouterAnthropicClaudeSonnet5,
313    /// Mistral Large 3 2512 - Mistral Large 3 2512 model via OpenRouter
314    OpenRouterMistralaiMistralLarge2512,
315    /// DeepSeek V3.1 Nex N1 - Nex AGI DeepSeek V3.1 Nex N1 model via OpenRouter
316    OpenRouterNexAgiDeepseekV31NexN1,
317    /// GLM-5.2 - Z.AI GLM-5.2 flagship model for long-horizon tasks via OpenRouter
318    /// GLM-5.3 Flash - Z.AI efficient multimodal model with hybrid sparse+linear attention via OpenRouter
319    OpenRouterZaiGlm53Flash,
320    /// GLM-5.3 FlashX - Z.AI high-speed Flash variant with faster inference via OpenRouter
321    OpenRouterZaiGlm53Flashx,
322    /// Kimi K3 - Moonshot AI's 2.8T parameter flagship via OpenRouter
323    OpenRouterMoonshotaiKimiK3,
324    /// Grok 4.6 - xAI's flagship reasoning model with reasoning_effort support via OpenRouter
325    OpenRouterXAiGrok46,
326    /// Grok 4.7 - xAI's flagship reasoning model with reasoning_effort support via OpenRouter
327    OpenRouterXAiGrok47,
328    /// MiMo-V2.6-Pro - Xiaomi's flagship agentic model for complex software engineering via OpenRouter
329    OpenRouterXiaomiMimoV26Pro,
330    /// MiMo-V2.6-Flash - Xiaomi's efficient high-volume agentic model via OpenRouter
331    OpenRouterXiaomiMimoV26Flash,
332    /// MiMo-V2.6-Pro-UltraSpeed - Xiaomi's fastest flagship variant via OpenRouter
333    OpenRouterXiaomiMimoV26ProUltraspeed,
334    /// Step 5 Preview - StepFun's flagship agentic model via OpenRouter
335    OpenRouterStepfunStep5Preview,
336
337    // Vercel AI Gateway models (namespaced as `vendor/model` on the gateway)
338    /// Claude Sonnet 5 served through the Vercel AI Gateway
339    VercelAnthropicClaudeSonnet5,
340    /// Claude Opus 5 served through the Vercel AI Gateway
341    VercelAnthropicClaudeOpus5,
342    /// Claude Opus 5.5 served through the Vercel AI Gateway
343    VercelAnthropicClaudeOpus55,
344    /// Claude Haiku 4.5 served through the Vercel AI Gateway
345    VercelAnthropicClaudeHaiku45,
346    /// GPT-5.6 Sol served through the Vercel AI Gateway
347    VercelOpenAiGpt56Sol,
348    /// GPT-6 Astra served through the Vercel AI Gateway
349    VercelOpenAiGpt6Astra,
350    /// GPT-5.6 Luna served through the Vercel AI Gateway
351    VercelOpenAiGpt56Luna,
352    /// GPT-6 Sol served through the Vercel AI Gateway
353    VercelOpenAiGpt6Sol,
354    /// GPT-6 Luna served through the Vercel AI Gateway
355    VercelOpenAiGpt6Luna,
356    /// GPT-5.3 Codex served through the Vercel AI Gateway
357    /// Gemini 3.1 Pro Preview served through the Vercel AI Gateway
358    /// Gemini 3.8 Flash served through the Vercel AI Gateway
359    VercelGoogleGemini38Flash,
360    /// DeepSeek V4.1 Flash served through the Vercel AI Gateway
361    VercelDeepseekFlash,
362    /// Kimi K3 served through the Vercel AI Gateway
363    VercelMoonshotaiKimiK3,
364    /// MiniMax M3 served through the Vercel AI Gateway
365    VercelMinimaxM3,
366    /// Grok 4.7 served through the Vercel AI Gateway
367    VercelSpacexaiGrok47,
368    // xAI models
369    /// Grok 4.6 - xAI's flagship reasoning model with reasoning_effort support (500k context)
370    XaiGrok46,
371    /// Grok 4.7 - xAI's flagship reasoning model with reasoning_effort support (500k context)
372    XaiGrok47,
373
374    /// User-defined model not in the hardcoded catalog.
375    /// Carries the provider key string and model identifier string.
376    Custom(String, String),
377}
vtcode_config/models/model_id.rs

vtcode_config/models/
model_id.rs
