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-6.1 Terra - Fast, efficient model for visual reasoning, precise computer use, and clear, natural writing
39    GPT61Terra,
40    /// GPT-6 Luna - Fast cost-efficient model in the GPT-6 series for high-volume latency-sensitive workloads
41    GPT6Luna,
42    /// GPT-5.6 Sol - Frontier model for complex professional work in the GPT-5.6 family
43    GPT56Sol,
44    /// GPT-5.6 Terra - GPT-5.6 model that balances intelligence and cost
45    GPT56Terra,
46    /// GPT-5.6 Luna - GPT-5.6 model optimized for cost-sensitive workloads
47    GPT56Luna,
48    /// GPT-OSS 20B - OpenAI's open-source 20B parameter model using harmony
49    OpenAIGptOss20b,
50    /// GPT-OSS 120B - OpenAI's open-source 120B parameter model using harmony
51    OpenAIGptOss120b,
52
53    // Anthropic models
54    /// Claude Sonnet 5 - The best combination of speed and intelligence with adaptive thinking on by default
55    #[default]
56    ClaudeSonnet5,
57    /// Claude Sonnet 5.5 - Latest Sonnet: best speed/intelligence balance, 1M context, 128K output, `between_tools` as the lowest thinking setting, default effort high
58    ClaudeSonnet55,
59    /// Claude Haiku 5.5 - High-volume latency-sensitive work: classification, routing, extraction, subagents; adaptive thinking, 1M context, 128K output, default effort medium
60    ClaudeHaiku55,
61    /// Claude Fable 5 - Anthropic's most capable widely released model for demanding reasoning and long-horizon agentic work
62    ClaudeFable5,
63    /// 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
64    ClaudeFable51,
65    /// Claude Opus 5 - Anthropic's newest Opus-tier model with 1M context, thinking on by default
66    ClaudeOpus5,
67    /// Claude Opus 5.5 - Opus-tier successor for long-running agentic coding, adaptive thinking always on, 1M context, 128K output, default effort medium
68    ClaudeOpus55,
69    /// GitHub Copilot auto model selection
70    CopilotAuto,
71    /// GitHub Copilot GPT-5.2 Codex
72    CopilotGPT52Codex,
73    /// GitHub Copilot GPT-5.1 Codex Max
74    CopilotGPT51CodexMax,
75    /// GitHub Copilot GPT-5.4
76    CopilotGPT54,
77    /// GitHub Copilot GPT-5.4 Mini
78    CopilotGPT54Mini,
79
80    // DeepSeek models
81    /// DeepSeek V4.1 Flash - Latest flash model with improved reasoning and efficiency
82    DeepSeekFlash,
83
84    // Official Meta AI models
85    /// Meta Muse Spark 1.1 - Official Meta AI Standard-tier reasoning model
86    MetaMuseSpark11,
87    /// Meta Muse Spark 1.3 - Official Meta AI flagship Standard-tier reasoning model, tuned for agentic workflows
88    MetaMuseSpark13,
89    /// Meta Muse Spark 1.3 Contributor tier - opt-in variant with Meta's discounted data-contribution terms
90    MetaMuseSpark13Contributor,
91    /// Meta Muse Wonder 1.0 - Frontier coding, long-running agents, and advanced computer use
92    MetaMuseWonder10,
93    /// Meta Muse Wonder Full 1.0 - Most advanced Meta model for critical workflows (Muse Code subscription)
94    MetaMuseWonderFull10,
95
96    // NVIDIA NIM models
97    /// NVIDIA Nemotron 3 Ultra - NVIDIA's flagship agentic reasoning model via NIM
98    NvidiaNemotron3Ultra550bA55b,
99    /// NVIDIA Nemotron 3 Super - Efficient long-context agentic reasoning model via NIM
100    NvidiaNemotron3Super120bA12b,
101    /// NVIDIA Nemotron 3 Nano - Efficient reasoning and tool-use model via NIM
102    NvidiaNemotron3Nano30bA3b,
103
104    // Merge Gateway routes
105    /// Merge Gateway's default route selected by Merge
106    MergeGatewayDefaultRouting,
107    /// Anthropic Claude Opus 5 through Merge Gateway
108    MergeGatewayAnthropicClaudeOpus5,
109    /// Anthropic Claude Opus 5.5 through Merge Gateway
110    MergeGatewayAnthropicClaudeOpus55,
111    /// Anthropic Claude Sonnet 5 through Merge Gateway
112    MergeGatewayAnthropicClaudeSonnet5,
113    /// Anthropic Claude Sonnet 5.5 through Merge Gateway
114    MergeGatewayAnthropicClaudeSonnet55,
115    /// Google Gemini 3.6 Flash through Merge Gateway
116    /// Google Gemini 3.7 Flash through Merge Gateway
117    /// DeepSeek V4.1 Flash through Merge Gateway
118    MergeGatewayDeepseekFlash,
119    /// xAI Grok 4.6 through Merge Gateway
120    MergeGatewayXaiGrok46,
121    /// xAI Grok 4.7 through Merge Gateway
122    MergeGatewayXaiGrok47,
123    /// MiniMax H3 through Merge Gateway
124    MergeGatewayMinimaxH3,
125    /// MiniMax M3.1 Flash through Merge Gateway
126    MergeGatewayMinimaxM31Flash,
127    /// Moonshot Kimi K3 through Merge Gateway
128    MergeGatewayMoonshotKimiK3,
129    /// Moonshot Kimi K3.1 Pro through Merge Gateway
130    MergeGatewayMoonshotKimiK31Pro,
131    /// Thinking Machines Inkling through Merge Gateway
132    MergeGatewayThinkingMachinesInkling,
133    /// Z.AI GLM-5.3 Flash through Merge Gateway
134    MergeGatewayZaiGlm53Flash,
135    /// Z.AI GLM-5.3 FlashX through Merge Gateway
136    MergeGatewayZaiGlm53Flashx,
137    /// OpenAI GPT-5.6 Luna through Merge Gateway
138    MergeGatewayOpenAIGpt56Luna,
139    /// OpenAI GPT-5.6 Sol through Merge Gateway
140    MergeGatewayOpenAIGpt56Sol,
141    /// OpenAI GPT-5.6 Terra through Merge Gateway
142    MergeGatewayOpenAIGpt56Terra,
143    /// Google Gemini 3.8 Flash through Merge Gateway
144    MergeGatewayGoogleGemini38Flash,
145    /// Google Gemini 4 Argon through Merge Gateway
146    MergeGatewayGoogleGemini4Argon,
147    /// Anthropic Claude Haiku 4.5 through Merge Gateway
148    MergeGatewayAnthropicClaudeHaiku4520251001,
149    /// Anthropic Claude Haiku 5.5 through Merge Gateway
150    MergeGatewayAnthropicClaudeHaiku55,
151    /// Anthropic Claude Fable 5.1 through Merge Gateway
152    MergeGatewayAnthropicClaudeFable51,
153    /// OpenAI GPT-6 Astra through Merge Gateway
154    MergeGatewayOpenAIGpt6Astra,
155    /// OpenAI GPT-6 Sol through Merge Gateway
156    MergeGatewayOpenAIGpt6Sol,
157    /// OpenAI GPT-6.1 Sol through Merge Gateway
158    MergeGatewayOpenAIGpt61Sol,
159    /// OpenAI GPT-6.1 Terra through Merge Gateway
160    MergeGatewayOpenAIGpt61Terra,
161    /// OpenAI GPT-6 Luna through Merge Gateway
162    MergeGatewayOpenAIGpt6Luna,
163    /// Xiaomi MiMo V2.6 Pro through Merge Gateway
164    MergeGatewayXiaomimimoMimoV26Pro,
165    /// Xiaomi MiMo V2.6 Flash through Merge Gateway
166    MergeGatewayXiaomimimoMimoV26Flash,
167    /// Mistral Large 4 through Merge Gateway
168    MergeGatewayMistralLarge4,
169    /// Meta Muse Spark Open (816B/42B) through Merge Gateway
170    MergeGatewayMetaMuseSparkOpen,
171
172    // Mistral AI models
173    /// Mistral Large 3 - State-of-the-art open-weight general-purpose multimodal model
174    MistralLarge3,
175    /// Mistral Large 4 - Open-weight MoE flagship (49B active / 1.05T total) with 1M context (Public Preview)
176    MistralLarge4,
177    // Hugging Face models
178    /// OpenAI GPT-OSS 20B via Hugging Face router
179    HuggingFaceOpenAIGptOss20b,
180    /// OpenAI GPT-OSS 120B via Hugging Face router
181    HuggingFaceOpenAIGptOss120b,
182    /// Z.AI GLM-5.2 via Novita inference provider on Hugging Face router
183    /// Z.AI GLM-5.3 Flash via Together inference provider on Hugging Face router
184    HuggingFaceGlm53FlashTogether,
185    /// Z.AI GLM-5.3 via Together inference provider on Hugging Face router
186    HuggingFaceGlm53Together,
187    /// Kimi K3 via Together on Hugging Face router
188    HuggingFaceKimiK3Together,
189    /// MiniMax M3 via Novita on Hugging Face router
190    HuggingFaceMinimaxM3Novita,
191    /// Meta Muse Spark Open 816B-A42B via Baseten on Hugging Face router
192    HuggingFaceMuseSparkOpenBaseten,
193
194    // StepFun models
195    /// Step 3.7 Flash - StepFun's flagship multimodal reasoning model with tool calling
196    StepFun37Flash,
197    /// Step 5 Preview - StepFun's frontier model for production-scale Agent applications with 1M context
198    StepFun5Preview,
199
200    // Evolink gateway models (namespaced as `evolink/<model>`)
201    /// GPT-5.2 served through the Evolink gateway
202    /// GPT-5.5 served through the Evolink gateway
203    /// Gemini 3.1 Pro served through the Evolink gateway (OpenAI SDK format)
204    EvolinkGemini31Pro,
205    /// Gemini 3.5 Flash served through the Evolink gateway (OpenAI SDK format)
206    /// MiniMax-M3 served through the Evolink gateway (OpenAI Chat Completions format)
207    EvolinkMinimaxM3,
208    /// Claude Haiku 4.5 served through the Evolink gateway (Anthropic Messages API)
209    EvolinkClaudeHaiku45,
210
211    /// GLM-5.3 - Z.ai flagship coding model with frontier long-horizon agentic performance
212    ZaiGlm53,
213    /// GLM-5.3 Flash - Z.ai efficient multimodal model with hybrid sparse+linear attention, 320B total / 18B active, 1M context, native vision
214    ZaiGlm53Flash,
215    /// 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
216    ZaiGlm53Flashx,
217
218    // MiMo models
219    /// MiMo V2.6 Pro - Xiaomi's flagship reasoning model with 1M context
220    MiMoV26Pro,
221    /// MiMo V2.6 Flash - Xiaomi's efficient high-volume model with 1M context
222    MiMoV26Flash,
223    /// MiMo V2.6 Pro UltraSpeed - Xiaomi's fastest flagship variant with 1M context
224    MiMoV26ProUltraspeed,
225
226    // Moonshot models
227    /// Kimi K3.1 Pro - Moonshot.ai's 3T-class open-weight flagship post-trained from K3, 1M context
228    MoonshotKimiK31Pro,
229    /// Kimi K3 - Moonshot.ai's 2.8T parameter flagship with Delta Attention, native vision, 1M context
230    MoonshotKimiK3,
231
232    // OpenCode Zen models
233
234    // OpenCode Go models (20 models - https://opencode.ai/docs/go)
235    /// GLM-5.3 - Z.AI flagship for frontier long-horizon coding on OpenCode Go
236    OpenCodeGoGlm53,
237    /// GLM-5.2 - Z.AI flagship model included with OpenCode Go
238    /// GPT-5.6 Luna - OpenAI cost-efficient frontier model on OpenCode Go
239    OpenCodeGoGpt56Luna,
240    /// Kimi K3 - Moonshot flagship 2.8T agentic model on OpenCode Go
241    OpenCodeGoKimiK3,
242    /// MiniMax M3 - Frontier multimodal coding model on OpenCode Go
243    OpenCodeGoMinimaxM3,
244    /// Muse Spark 1.2 Contributor - Meta long-context reasoning on OpenCode Go (limited regions)
245
246    // Qwen models (non-Qwen3 only)
247
248    // Ollama models
249    /// GPT-OSS 20B - Open-weight GPT-OSS 20B model served via Ollama locally
250    OllamaGptOss20b,
251    /// GPT-OSS 20B Cloud - Cloud-hosted GPT-OSS 20B served via Ollama Cloud
252    OllamaGptOss20bCloud,
253    /// GPT-OSS 120B Cloud - Cloud-hosted GPT-OSS 120B served via Ollama Cloud
254    OllamaGptOss120bCloud,
255    /// MiniMax-M3 Cloud - Cloud-hosted MiniMax-M3 model served via Ollama Cloud
256    OllamaMinimaxM3Cloud,
257    /// MiniMax M3.1 Flash Cloud - ultra-efficient multimodal MiniMax M3.1 Flash served via Ollama Cloud
258    OllamaMinimaxM31FlashCloud,
259    /// GLM-5.2 Cloud - Cloud-hosted GLM-5.2 flagship model served via Ollama Cloud
260    /// GLM-5.3 Cloud - Cloud-hosted GLM-5.3 flagship model served via Ollama Cloud
261    OllamaGlm53Cloud,
262    /// GLM-5.3 Flash Cloud - Multimodal GLM-5.3 Flash served via Ollama Cloud
263    OllamaGlm53FlashCloud,
264    /// Mistral Large 4 Cloud - Multimodal Mistral Large 4 served via Ollama Cloud
265    OllamaMistralLarge4Cloud,
266    /// Muse Spark Open Cloud - Multimodal Muse Spark Open served via Ollama Cloud
267    OllamaMuseSparkOpenCloud,
268    /// DeepSeek V4.1 Flash Cloud - Multimodal DeepSeek V4.1 Flash served via Ollama Cloud
269    OllamaDeepSeekV41FlashCloud,
270    /// Kimi K3 Cloud - Moonshot Kimi K3 via Ollama Cloud
271    OllamaKimiK3Cloud,
272    /// Gemma 4 - Google Gemma 4 model served via Ollama
273    OllamaGemma4,
274    /// Laguna XS.2 - Poolside's 33B MoE model (3B activated) for agentic coding via Ollama
275
276    // llama.cpp models
277    /// Gemma 4 26B A4B - Desktop Gemma 4 MoE model served through llama.cpp
278    LlamaCppGemma426bA4b,
279    /// Gemma 4 E4B - Tiny-footprint Gemma 4 model served through llama.cpp
280    LlamaCppGemma4E4b,
281    /// GPT-OSS 20B - OpenAI open-weight model served through llama.cpp
282    LlamaCppGptOss20b,
283
284    // MiniMax models
285    /// MiniMax-M3 - Frontier multimodal coding model with 1M context
286    MinimaxM3,
287    /// MiniMax-M3.1-Flash - Latest ultra-efficient multimodal model via OpenRouter
288    OpenRouterMinimaxM31FlashFree,
289    /// MiniMax-M3.1-Flash - Latest ultra-efficient multimodal model with 1M context and tunable thinking depth
290    MinimaxM31Flash,
291
292    // OpenRouter models
293    /// DeepSeek V4.1 Flash 1017 - Latest flash model via OpenRouter
294    OpenRouterDeepSeekFlash,
295    /// OpenAI gpt-oss-120b - Open-weight 120B reasoning model via OpenRouter
296    OpenRouterOpenAIGptOss120b,
297    /// OpenAI gpt-oss-120b:free - Open-weight 120B reasoning model free tier via OpenRouter
298    OpenRouterOpenAIGptOss120bFree,
299    /// OpenAI gpt-oss-20b - Open-weight 20B deployment via OpenRouter
300    OpenRouterOpenAIGptOss20b,
301    /// OpenAI GPT-6 Astra - OpenAI's flagship model for demanding end-to-end work via OpenRouter
302    OpenRouterOpenAIGpt6Astra,
303    /// GPT-6 Sol - Cost-efficient high-end model in the GPT-6 series via OpenRouter
304    OpenRouterOpenAIGpt6Sol,
305    /// GPT-6 Luna - Fast cost-efficient model in the GPT-6 series via OpenRouter
306    OpenRouterOpenAIGpt6Luna,
307
308    /// Meta Muse Glimmer 30B via OpenRouter
309    OpenRouterMetaMuseGlimmer30b,
310    /// Meta Muse Spark 1.2 via OpenRouter
311    /// Meta Muse Spark 1.3 via OpenRouter
312    OpenRouterMetaMuseSpark13,
313    /// Meta Muse Spark Open 816B-A42B via OpenRouter
314    OpenRouterMetaMuseSparkOpen,
315    /// Gemini 3.7 Flash - Flash model with 1M context and tunable thinking levels via OpenRouter
316    /// Gemini 3.8 Flash - Most intelligent Flash for long-horizon SWE/agents with 1M context via OpenRouter
317    OpenRouterGoogleGemini38Flash,
318
319    /// Claude Sonnet 5 - Anthropic Claude Sonnet 5 listing
320    OpenRouterAnthropicClaudeSonnet5,
321    /// Mistral Large 3 2512 - Mistral Large 3 2512 model via OpenRouter
322    OpenRouterMistralaiMistralLarge2512,
323    /// DeepSeek V3.1 Nex N1 - Nex AGI DeepSeek V3.1 Nex N1 model via OpenRouter
324    OpenRouterNexAgiDeepseekV31NexN1,
325    /// GLM-5.2 - Z.AI GLM-5.2 flagship model for long-horizon tasks via OpenRouter
326    /// GLM-5.3 Flash - Z.AI efficient multimodal model with hybrid sparse+linear attention via OpenRouter
327    OpenRouterZaiGlm53Flash,
328    /// GLM-5.3 FlashX - Z.AI high-speed Flash variant with faster inference via OpenRouter
329    OpenRouterZaiGlm53Flashx,
330    /// Kimi K3 - Moonshot AI's 2.8T parameter flagship via OpenRouter
331    OpenRouterMoonshotaiKimiK3,
332    /// Grok 4.6 - xAI's flagship reasoning model with reasoning_effort support via OpenRouter
333    OpenRouterXAiGrok46,
334    /// Grok 4.7 - xAI's flagship reasoning model with reasoning_effort support via OpenRouter
335    OpenRouterXAiGrok47,
336    /// MiMo-V2.6-Pro - Xiaomi's flagship agentic model for complex software engineering via OpenRouter
337    OpenRouterXiaomiMimoV26Pro,
338    /// MiMo-V2.6-Flash - Xiaomi's efficient high-volume agentic model via OpenRouter
339    OpenRouterXiaomiMimoV26Flash,
340    /// MiMo-V2.6-Pro-UltraSpeed - Xiaomi's fastest flagship variant via OpenRouter
341    OpenRouterXiaomiMimoV26ProUltraspeed,
342    /// Step 5 Preview - StepFun's flagship agentic model via OpenRouter
343    OpenRouterStepfunStep5Preview,
344
345    // Vercel AI Gateway models (namespaced as `vendor/model` on the gateway)
346    /// Claude Sonnet 5 served through the Vercel AI Gateway
347    VercelAnthropicClaudeSonnet5,
348    /// Claude Opus 5 served through the Vercel AI Gateway
349    VercelAnthropicClaudeOpus5,
350    /// Claude Opus 5.5 served through the Vercel AI Gateway
351    VercelAnthropicClaudeOpus55,
352    /// Claude Haiku 4.5 served through the Vercel AI Gateway
353    VercelAnthropicClaudeHaiku45,
354    /// GPT-5.6 Sol served through the Vercel AI Gateway
355    VercelOpenAiGpt56Sol,
356    /// GPT-6 Astra served through the Vercel AI Gateway
357    VercelOpenAiGpt6Astra,
358    /// GPT-5.6 Luna served through the Vercel AI Gateway
359    VercelOpenAiGpt56Luna,
360    /// GPT-6 Sol served through the Vercel AI Gateway
361    VercelOpenAiGpt6Sol,
362    /// GPT-6 Luna served through the Vercel AI Gateway
363    VercelOpenAiGpt6Luna,
364    /// GPT-5.3 Codex served through the Vercel AI Gateway
365    /// Gemini 3.1 Pro Preview served through the Vercel AI Gateway
366    /// Gemini 3.8 Flash served through the Vercel AI Gateway
367    VercelGoogleGemini38Flash,
368    /// DeepSeek V4.1 Flash served through the Vercel AI Gateway
369    VercelDeepseekFlash,
370    /// Kimi K3 served through the Vercel AI Gateway
371    VercelMoonshotaiKimiK3,
372    /// MiniMax M3 served through the Vercel AI Gateway
373    VercelMinimaxM3,
374    /// Grok 4.7 served through the Vercel AI Gateway
375    VercelSpacexaiGrok47,
376    // xAI models
377    /// Grok 4.6 - xAI's flagship reasoning model with reasoning_effort support (500k context)
378    XaiGrok46,
379    /// Grok 4.7 - xAI's flagship reasoning model with reasoning_effort support (500k context)
380    XaiGrok47,
381
382    /// User-defined model not in the hardcoded catalog.
383    /// Carries the provider key string and model identifier string.
384    Custom(String, String),
385}
vtcode_config/models/model_id.rs

vtcode_config/models/
model_id.rs
