diff --git a/README.md b/README.md
index ae82e64a..5a30e059 100644
--- a/README.md
+++ b/README.md
@@ -174,6 +174,9 @@ deepseek --provider fireworks --model deepseek-v4-pro
 
 # Self-hosted SGLang
 SGLANG_BASE_URL="http://localhost:30000/v1" deepseek --provider sglang --model deepseek-v4-flash
+
+# Self-hosted vLLM
+VLLM_BASE_URL="http://localhost:8000/v1" deepseek --provider vllm --model deepseek-v4-flash
 ```
 
 ---
@@ -257,11 +260,12 @@ Key environment variables:
 | `DEEPSEEK_API_KEY` | API key |
 | `DEEPSEEK_BASE_URL` | API base URL |
 | `DEEPSEEK_MODEL` | Default model |
-| `DEEPSEEK_PROVIDER` | `deepseek` (default), `nvidia-nim`, `fireworks`, `sglang` |
+| `DEEPSEEK_PROVIDER` | `deepseek` (default), `nvidia-nim`, `fireworks`, `sglang`, `vllm` |
 | `DEEPSEEK_PROFILE` | Config profile name |
 | `DEEPSEEK_MEMORY` | Set to `on` to enable user memory |
-| `NVIDIA_API_KEY` / `FIREWORKS_API_KEY` / `SGLANG_API_KEY` | Provider auth |
+| `NVIDIA_API_KEY` / `FIREWORKS_API_KEY` / `SGLANG_API_KEY` / `VLLM_API_KEY` | Provider auth |
 | `SGLANG_BASE_URL` | Self-hosted SGLang endpoint |
+| `VLLM_BASE_URL` | Self-hosted vLLM endpoint |
 | `NO_ANIMATIONS=1` | Force accessibility mode at startup |
 | `SSL_CERT_FILE` | Custom CA bundle for corporate proxies |
 
diff --git a/README.zh-CN.md b/README.zh-CN.md
index b041be8b..1fcb548a 100644
--- a/README.zh-CN.md
+++ b/README.zh-CN.md
@@ -169,6 +169,9 @@ deepseek --provider fireworks --model deepseek-v4-pro
 
 # 自托管 SGLang
 SGLANG_BASE_URL="http://localhost:30000/v1" deepseek --provider sglang --model deepseek-v4-flash
+
+# 自托管 vLLM
+VLLM_BASE_URL="http://localhost:8000/v1" deepseek --provider vllm --model deepseek-v4-flash
 ```
 
 ---
@@ -252,11 +255,12 @@ deepseek mcp-server                            # 启动 dispatcher MCP stdio 服
 | `DEEPSEEK_API_KEY` | DeepSeek API key |
 | `DEEPSEEK_BASE_URL` | API base URL |
 | `DEEPSEEK_MODEL` | 默认模型 |
-| `DEEPSEEK_PROVIDER` | `deepseek`（默认）、`nvidia-nim`、`fireworks`、`sglang` |
+| `DEEPSEEK_PROVIDER` | `deepseek`（默认）、`nvidia-nim`、`fireworks`、`sglang`、`vllm` |
 | `DEEPSEEK_PROFILE` | 配置 profile 名称 |
 | `DEEPSEEK_MEMORY` | 设为 `on` 启用用户记忆 |
-| `NVIDIA_API_KEY` / `FIREWORKS_API_KEY` / `SGLANG_API_KEY` | 提供商认证 |
+| `NVIDIA_API_KEY` / `FIREWORKS_API_KEY` / `SGLANG_API_KEY` / `VLLM_API_KEY` | 提供商认证 |
 | `SGLANG_BASE_URL` | 自托管 SGLang 端点 |
+| `VLLM_BASE_URL` | 自托管 vLLM 端点 |
 | `NO_ANIMATIONS=1` | 启动时强制无障碍模式 |
 | `SSL_CERT_FILE` | 企业代理的自定义 CA 包 |
 
diff --git a/config.example.toml b/config.example.toml
index 2bee9a0b..632bc7a7 100644
--- a/config.example.toml
+++ b/config.example.toml
@@ -12,10 +12,10 @@
 # Choose which provider to use by default. Per-provider credentials live in the
 # `[providers.*]` sections near the bottom of
 # this file — keeping both stored at once means `/provider deepseek` and
-# `/provider nvidia-nim` (or `--provider fireworks`, `/provider sglang`) toggle without having to
+# `/provider nvidia-nim` (or `--provider fireworks`, `/provider sglang`, `/provider vllm`) toggle without having to
 # re-enter keys. Top-level `api_key` / `base_url` are still read as DeepSeek
 # defaults when `[providers.deepseek]` is absent (backward compatibility).
-provider = "deepseek" # deepseek | nvidia-nim | openrouter | novita | fireworks | sglang
+provider = "deepseek" # deepseek | nvidia-nim | openrouter | novita | fireworks | sglang | vllm
 api_key = "YOUR_DEEPSEEK_API_KEY" # must be non-empty
 base_url = "https://api.deepseek.com"
 # base_url = "https://api.deepseeki.com"         # China users
@@ -148,6 +148,7 @@ max_subagents = 10 # optional (1-20)
 #             (or NVIDIA_NIM_BASE_URL / NVIDIA_BASE_URL), NVIDIA_NIM_MODEL
 #   Fireworks: FIREWORKS_API_KEY, FIREWORKS_BASE_URL
 #   SGLang:    SGLANG_BASE_URL, SGLANG_MODEL, optional SGLANG_API_KEY
+#   vLLM:      VLLM_BASE_URL, VLLM_MODEL, optional VLLM_API_KEY
 
 # DeepSeek Platform (https://platform.deepseek.com)
 [providers.deepseek]
@@ -173,6 +174,12 @@ max_subagents = 10 # optional (1-20)
 # base_url = "http://localhost:30000/v1"
 # model = "deepseek-ai/DeepSeek-V4-Pro"     # or deepseek-ai/DeepSeek-V4-Flash
 
+# Self-hosted vLLM OpenAI-compatible server
+[providers.vllm]
+# api_key = "OPTIONAL_VLLM_TOKEN"
+# base_url = "http://localhost:8000/v1"
+# model = "deepseek-ai/DeepSeek-V4-Pro"     # or deepseek-ai/DeepSeek-V4-Flash
+
 # ─────────────────────────────────────────────────────────────────────────────────
 # Network Policy (#135)
 # ─────────────────────────────────────────────────────────────────────────────────
diff --git a/crates/agent/src/lib.rs b/crates/agent/src/lib.rs
index eb1e0622..3d2644d1 100644
--- a/crates/agent/src/lib.rs
+++ b/crates/agent/src/lib.rs
@@ -163,6 +163,28 @@ impl Default for ModelRegistry {
                 supports_tools: true,
                 supports_reasoning: true,
             },
+            ModelInfo {
+                id: "deepseek-ai/DeepSeek-V4-Pro".to_string(),
+                provider: ProviderKind::Vllm,
+                aliases: vec![
+                    "deepseek-v4-pro".to_string(),
+                    "vllm-deepseek-v4-pro".to_string(),
+                ],
+                supports_tools: true,
+                supports_reasoning: true,
+            },
+            ModelInfo {
+                id: "deepseek-ai/DeepSeek-V4-Flash".to_string(),
+                provider: ProviderKind::Vllm,
+                aliases: vec![
+                    "deepseek-v4-flash".to_string(),
+                    "deepseek-chat".to_string(),
+                    "deepseek-reasoner".to_string(),
+                    "vllm-deepseek-v4-flash".to_string(),
+                ],
+                supports_tools: true,
+                supports_reasoning: true,
+            },
         ];
         Self::new(models)
     }
@@ -366,4 +388,22 @@ mod tests {
         assert_eq!(resolved.resolved.provider, ProviderKind::Sglang);
         assert_eq!(resolved.resolved.id, "deepseek-ai/DeepSeek-V4-Flash");
     }
+
+    #[test]
+    fn vllm_default_uses_canonical_model_id() {
+        let registry = ModelRegistry::default();
+        let resolved = registry.resolve(None, Some(ProviderKind::Vllm));
+
+        assert_eq!(resolved.resolved.provider, ProviderKind::Vllm);
+        assert_eq!(resolved.resolved.id, "deepseek-ai/DeepSeek-V4-Pro");
+    }
+
+    #[test]
+    fn deepseek_v4_flash_alias_resolves_to_vllm_when_provider_hinted() {
+        let registry = ModelRegistry::default();
+        let resolved = registry.resolve(Some("deepseek-v4-flash"), Some(ProviderKind::Vllm));
+
+        assert_eq!(resolved.resolved.provider, ProviderKind::Vllm);
+        assert_eq!(resolved.resolved.id, "deepseek-ai/DeepSeek-V4-Flash");
+    }
 }
diff --git a/crates/cli/src/lib.rs b/crates/cli/src/lib.rs
index 87ee2d7c..cb2a5cf5 100644
--- a/crates/cli/src/lib.rs
+++ b/crates/cli/src/lib.rs
@@ -28,6 +28,7 @@ enum ProviderArg {
     Novita,
     Fireworks,
     Sglang,
+    Vllm,
 }
 
 impl From<ProviderArg> for ProviderKind {
@@ -40,6 +41,7 @@ impl From<ProviderArg> for ProviderKind {
             ProviderArg::Novita => ProviderKind::Novita,
             ProviderArg::Fireworks => ProviderKind::Fireworks,
             ProviderArg::Sglang => ProviderKind::Sglang,
+            ProviderArg::Vllm => ProviderKind::Vllm,
         }
     }
 }
@@ -561,17 +563,19 @@ fn provider_slot(provider: ProviderKind) -> &'static str {
         ProviderKind::Novita => "novita",
         ProviderKind::Fireworks => "fireworks",
         ProviderKind::Sglang => "sglang",
+        ProviderKind::Vllm => "vllm",
     }
 }
 
 /// Provider order used by the `auth list` and `auth status` outputs.
-const PROVIDER_LIST: [ProviderKind; 7] = [
+const PROVIDER_LIST: [ProviderKind; 8] = [
     ProviderKind::Deepseek,
     ProviderKind::NvidiaNim,
     ProviderKind::Openrouter,
     ProviderKind::Novita,
     ProviderKind::Fireworks,
     ProviderKind::Sglang,
+    ProviderKind::Vllm,
     ProviderKind::Openai,
 ];
 
@@ -1045,9 +1049,10 @@ fn delegate_to_tui(
             | ProviderKind::Novita
             | ProviderKind::Fireworks
             | ProviderKind::Sglang
+            | ProviderKind::Vllm
     ) {
         bail!(
-            "The interactive TUI supports DeepSeek, NVIDIA NIM, OpenRouter, Novita, Fireworks, and SGLang providers. Remove --provider {} or use `deepseek model ...` for provider registry inspection.",
+            "The interactive TUI supports DeepSeek, NVIDIA NIM, OpenRouter, Novita, Fireworks, SGLang, and vLLM providers. Remove --provider {} or use `deepseek model ...` for provider registry inspection.",
             resolved_runtime.provider.as_str()
         );
     }
@@ -1562,6 +1567,16 @@ mod tests {
             }))
         ));
 
+        let cli = parse_ok(&["deepseek", "auth", "get", "--provider", "vllm"]);
+        assert!(matches!(
+            cli.command,
+            Some(Commands::Auth(AuthArgs {
+                command: AuthCommand::Get {
+                    provider: ProviderArg::Vllm
+                }
+            }))
+        ));
+
         let cli = parse_ok(&["deepseek", "auth", "list"]);
         assert!(matches!(
             cli.command,
diff --git a/crates/config/src/lib.rs b/crates/config/src/lib.rs
index dc491669..9d7cbfb8 100644
--- a/crates/config/src/lib.rs
+++ b/crates/config/src/lib.rs
@@ -26,6 +26,9 @@ const DEFAULT_OPENROUTER_BASE_URL: &str = "https://openrouter.ai/api/v1";
 const DEFAULT_NOVITA_BASE_URL: &str = "https://api.novita.ai/v1";
 const DEFAULT_FIREWORKS_BASE_URL: &str = "https://api.fireworks.ai/inference/v1";
 const DEFAULT_SGLANG_BASE_URL: &str = "http://localhost:30000/v1";
+const DEFAULT_VLLM_MODEL: &str = "deepseek-ai/DeepSeek-V4-Pro";
+const DEFAULT_VLLM_FLASH_MODEL: &str = "deepseek-ai/DeepSeek-V4-Flash";
+const DEFAULT_VLLM_BASE_URL: &str = "http://localhost:8000/v1";
 
 #[derive(Debug, Clone, Copy, Serialize, Deserialize, PartialEq, Eq, Default)]
 #[serde(rename_all = "kebab-case")]
@@ -38,6 +41,7 @@ pub enum ProviderKind {
     Novita,
     Fireworks,
     Sglang,
+    Vllm,
 }
 
 impl ProviderKind {
@@ -51,6 +55,7 @@ impl ProviderKind {
             Self::Novita => "novita",
             Self::Fireworks => "fireworks",
             Self::Sglang => "sglang",
+            Self::Vllm => "vllm",
         }
     }
 
@@ -64,6 +69,7 @@ impl ProviderKind {
             "novita" => Some(Self::Novita),
             "fireworks" | "fireworks-ai" => Some(Self::Fireworks),
             "sglang" | "sg-lang" => Some(Self::Sglang),
+            "vllm" | "v-llm" => Some(Self::Vllm),
             _ => None,
         }
     }
@@ -92,6 +98,8 @@ pub struct ProvidersToml {
     pub fireworks: ProviderConfigToml,
     #[serde(default)]
     pub sglang: ProviderConfigToml,
+    #[serde(default)]
+    pub vllm: ProviderConfigToml,
 }
 
 impl ProvidersToml {
@@ -105,6 +113,7 @@ impl ProvidersToml {
             ProviderKind::Novita => &self.novita,
             ProviderKind::Fireworks => &self.fireworks,
             ProviderKind::Sglang => &self.sglang,
+            ProviderKind::Vllm => &self.vllm,
         }
     }
 
@@ -117,6 +126,7 @@ impl ProvidersToml {
             ProviderKind::Novita => &mut self.novita,
             ProviderKind::Fireworks => &mut self.fireworks,
             ProviderKind::Sglang => &mut self.sglang,
+            ProviderKind::Vllm => &mut self.vllm,
         }
     }
 }
@@ -318,6 +328,7 @@ impl ConfigToml {
         merge_provider_config(&mut self.providers.novita, &project.providers.novita);
         merge_provider_config(&mut self.providers.fireworks, &project.providers.fireworks);
         merge_provider_config(&mut self.providers.sglang, &project.providers.sglang);
+        merge_provider_config(&mut self.providers.vllm, &project.providers.vllm);
 
         if project.network.is_some() {
             self.network = project.network;
@@ -373,6 +384,9 @@ impl ConfigToml {
             "providers.sglang.api_key" => self.providers.sglang.api_key.clone(),
             "providers.sglang.base_url" => self.providers.sglang.base_url.clone(),
             "providers.sglang.model" => self.providers.sglang.model.clone(),
+            "providers.vllm.api_key" => self.providers.vllm.api_key.clone(),
+            "providers.vllm.base_url" => self.providers.vllm.base_url.clone(),
+            "providers.vllm.model" => self.providers.vllm.model.clone(),
             _ => self.extras.get(key).map(toml::Value::to_string),
         }
     }
@@ -460,6 +474,15 @@ impl ConfigToml {
             "providers.sglang.model" => {
                 self.providers.sglang.model = Some(value.to_string());
             }
+            "providers.vllm.api_key" => {
+                self.providers.vllm.api_key = Some(value.to_string());
+            }
+            "providers.vllm.base_url" => {
+                self.providers.vllm.base_url = Some(value.to_string());
+            }
+            "providers.vllm.model" => {
+                self.providers.vllm.model = Some(value.to_string());
+            }
             _ => {
                 self.extras
                     .insert(key.to_string(), toml::Value::String(value.to_string()));
@@ -513,6 +536,9 @@ impl ConfigToml {
             "providers.sglang.api_key" => self.providers.sglang.api_key = None,
             "providers.sglang.base_url" => self.providers.sglang.base_url = None,
             "providers.sglang.model" => self.providers.sglang.model = None,
+            "providers.vllm.api_key" => self.providers.vllm.api_key = None,
+            "providers.vllm.base_url" => self.providers.vllm.base_url = None,
+            "providers.vllm.model" => self.providers.vllm.model = None,
             _ => {
                 self.extras.remove(key);
             }
@@ -624,6 +650,15 @@ impl ConfigToml {
         if let Some(v) = self.providers.sglang.model.as_ref() {
             out.insert("providers.sglang.model".to_string(), v.clone());
         }
+        if let Some(v) = self.providers.vllm.api_key.as_ref() {
+            out.insert("providers.vllm.api_key".to_string(), redact_secret(v));
+        }
+        if let Some(v) = self.providers.vllm.base_url.as_ref() {
+            out.insert("providers.vllm.base_url".to_string(), v.clone());
+        }
+        if let Some(v) = self.providers.vllm.model.as_ref() {
+            out.insert("providers.vllm.model".to_string(), v.clone());
+        }
 
         for (k, v) in &self.extras {
             out.insert(k.clone(), v.to_string());
@@ -695,6 +730,7 @@ impl ConfigToml {
                 ProviderKind::Novita => DEFAULT_NOVITA_BASE_URL.to_string(),
                 ProviderKind::Fireworks => DEFAULT_FIREWORKS_BASE_URL.to_string(),
                 ProviderKind::Sglang => DEFAULT_SGLANG_BASE_URL.to_string(),
+                ProviderKind::Vllm => DEFAULT_VLLM_BASE_URL.to_string(),
             });
 
         let model = cli
@@ -712,6 +748,7 @@ impl ConfigToml {
                 ProviderKind::Novita => DEFAULT_NOVITA_MODEL.to_string(),
                 ProviderKind::Fireworks => DEFAULT_FIREWORKS_MODEL.to_string(),
                 ProviderKind::Sglang => DEFAULT_SGLANG_MODEL.to_string(),
+                ProviderKind::Vllm => DEFAULT_VLLM_MODEL.to_string(),
             });
         let model = normalize_model_for_provider(provider, &model);
 
@@ -822,6 +859,14 @@ fn normalize_model_for_provider(provider: ProviderKind, model: &str) -> String {
             "deepseek-v4-flash" | "deepseek-v4flash" | "deepseek-chat" | "deepseek-reasoner"
             | "deepseek-r1" | "deepseek-v3" | "deepseek-v3.2",
         ) => DEFAULT_SGLANG_FLASH_MODEL.to_string(),
+        (ProviderKind::Vllm, "deepseek-v4-pro" | "deepseek-v4pro") => {
+            DEFAULT_VLLM_MODEL.to_string()
+        }
+        (
+            ProviderKind::Vllm,
+            "deepseek-v4-flash" | "deepseek-v4flash" | "deepseek-chat" | "deepseek-reasoner"
+            | "deepseek-r1" | "deepseek-v3" | "deepseek-v3.2",
+        ) => DEFAULT_VLLM_FLASH_MODEL.to_string(),
         _ => model.to_string(),
     }
 }
@@ -973,6 +1018,7 @@ struct EnvRuntimeOverrides {
     novita_base_url: Option<String>,
     fireworks_base_url: Option<String>,
     sglang_base_url: Option<String>,
+    vllm_base_url: Option<String>,
 }
 
 impl EnvRuntimeOverrides {
@@ -1013,6 +1059,9 @@ impl EnvRuntimeOverrides {
             sglang_base_url: std::env::var("SGLANG_BASE_URL")
                 .ok()
                 .filter(|v| !v.trim().is_empty()),
+            vllm_base_url: std::env::var("VLLM_BASE_URL")
+                .ok()
+                .filter(|v| !v.trim().is_empty()),
         }
     }
 
@@ -1027,6 +1076,7 @@ impl EnvRuntimeOverrides {
             ProviderKind::Novita => self.novita_base_url.clone(),
             ProviderKind::Fireworks => self.fireworks_base_url.clone(),
             ProviderKind::Sglang => self.sglang_base_url.clone(),
+            ProviderKind::Vllm => self.vllm_base_url.clone(),
         }
     }
 }
@@ -1061,6 +1111,8 @@ mod tests {
         fireworks_base_url: Option<OsString>,
         sglang_api_key: Option<OsString>,
         sglang_base_url: Option<OsString>,
+        vllm_api_key: Option<OsString>,
+        vllm_base_url: Option<OsString>,
     }
 
     impl EnvGuard {
@@ -1083,6 +1135,8 @@ mod tests {
                 fireworks_base_url: env::var_os("FIREWORKS_BASE_URL"),
                 sglang_api_key: env::var_os("SGLANG_API_KEY"),
                 sglang_base_url: env::var_os("SGLANG_BASE_URL"),
+                vllm_api_key: env::var_os("VLLM_API_KEY"),
+                vllm_base_url: env::var_os("VLLM_BASE_URL"),
             };
             // Safety: test-only environment mutation guarded by a module mutex.
             unsafe {
@@ -1103,6 +1157,8 @@ mod tests {
                 env::remove_var("FIREWORKS_BASE_URL");
                 env::remove_var("SGLANG_API_KEY");
                 env::remove_var("SGLANG_BASE_URL");
+                env::remove_var("VLLM_API_KEY");
+                env::remove_var("VLLM_BASE_URL");
             }
             guard
         }
@@ -1137,6 +1193,8 @@ mod tests {
                 Self::restore_var("FIREWORKS_BASE_URL", self.fireworks_base_url.take());
                 Self::restore_var("SGLANG_API_KEY", self.sglang_api_key.take());
                 Self::restore_var("SGLANG_BASE_URL", self.sglang_base_url.take());
+                Self::restore_var("VLLM_API_KEY", self.vllm_api_key.take());
+                Self::restore_var("VLLM_BASE_URL", self.vllm_base_url.take());
             }
         }
     }
@@ -1320,6 +1378,8 @@ mod tests {
             Some(ProviderKind::Fireworks)
         );
         assert_eq!(ProviderKind::parse("sg-lang"), Some(ProviderKind::Sglang));
+        assert_eq!(ProviderKind::parse("v-llm"), Some(ProviderKind::Vllm));
+        assert_eq!(ProviderKind::parse("vllm"), Some(ProviderKind::Vllm));
     }
 
     #[test]
@@ -1386,6 +1446,22 @@ mod tests {
         assert_eq!(resolved.model, DEFAULT_SGLANG_MODEL);
     }
 
+    #[test]
+    fn vllm_provider_defaults_to_local_endpoint_and_model() {
+        let _lock = env_lock();
+        let _env = EnvGuard::without_deepseek_runtime_overrides();
+        let config = ConfigToml {
+            provider: ProviderKind::Vllm,
+            ..ConfigToml::default()
+        };
+
+        let resolved = config.resolve_runtime_options(&CliRuntimeOverrides::default());
+
+        assert_eq!(resolved.provider, ProviderKind::Vllm);
+        assert_eq!(resolved.base_url, DEFAULT_VLLM_BASE_URL);
+        assert_eq!(resolved.model, DEFAULT_VLLM_MODEL);
+    }
+
     #[test]
     fn openrouter_env_api_key_falls_back_when_config_missing() {
         let _lock = env_lock();
@@ -1488,6 +1564,22 @@ mod tests {
         assert_eq!(resolved.model, DEFAULT_SGLANG_FLASH_MODEL);
     }
 
+    #[test]
+    fn vllm_provider_normalizes_flash_aliases() {
+        let _lock = env_lock();
+        let _env = EnvGuard::without_deepseek_runtime_overrides();
+        let cli = CliRuntimeOverrides {
+            provider: Some(ProviderKind::Vllm),
+            model: Some("deepseek-v4-flash".to_string()),
+            ..CliRuntimeOverrides::default()
+        };
+
+        let resolved = ConfigToml::default().resolve_runtime_options(&cli);
+
+        assert_eq!(resolved.provider, ProviderKind::Vllm);
+        assert_eq!(resolved.model, DEFAULT_VLLM_FLASH_MODEL);
+    }
+
     #[test]
     fn openrouter_provider_specific_config_overrides_env() {
         let _lock = env_lock();
diff --git a/crates/secrets/src/lib.rs b/crates/secrets/src/lib.rs
index e0c54241..ed4cb32d 100644
--- a/crates/secrets/src/lib.rs
+++ b/crates/secrets/src/lib.rs
@@ -411,6 +411,7 @@ pub fn env_for(name: &str) -> Option<String> {
         }
         "fireworks" | "fireworks-ai" => &["FIREWORKS_API_KEY"],
         "sglang" | "sg-lang" => &["SGLANG_API_KEY"],
+        "vllm" | "v-llm" => &["VLLM_API_KEY"],
         "openai" => &["OPENAI_API_KEY"],
         _ => return None,
     };
@@ -447,6 +448,7 @@ mod tests {
             "NVIDIA_NIM_API_KEY",
             "FIREWORKS_API_KEY",
             "SGLANG_API_KEY",
+            "VLLM_API_KEY",
             "OPENAI_API_KEY",
         ] {
             // Safety: tests serialise on env_lock(); the broader
@@ -563,6 +565,19 @@ mod tests {
         unsafe { std::env::remove_var("SGLANG_API_KEY") };
     }
 
+    #[test]
+    fn vllm_env_aliases_resolve() {
+        let _lock = env_lock();
+        clear_known_envs();
+        // Safety: env mutation guarded by env_lock().
+        unsafe { std::env::set_var("VLLM_API_KEY", "vllm-key") };
+
+        assert_eq!(env_for("vllm").as_deref(), Some("vllm-key"));
+        assert_eq!(env_for("v-llm").as_deref(), Some("vllm-key"));
+        // Safety: env mutation guarded by env_lock().
+        unsafe { std::env::remove_var("VLLM_API_KEY") };
+    }
+
     #[cfg(unix)]
     #[test]
     fn file_store_round_trips_with_secure_perms() {
diff --git a/crates/tui/src/client.rs b/crates/tui/src/client.rs
index e0421063..832c3ae0 100644
--- a/crates/tui/src/client.rs
+++ b/crates/tui/src/client.rs
@@ -754,7 +754,8 @@ pub(super) fn apply_reasoning_effort(
             | ApiProvider::Openrouter
             | ApiProvider::Novita
             | ApiProvider::Fireworks
-            | ApiProvider::Sglang => {
+            | ApiProvider::Sglang
+            | ApiProvider::Vllm => {
                 body["thinking"] = json!({ "type": "disabled" });
             }
             ApiProvider::NvidiaNim => {
@@ -769,7 +770,8 @@ pub(super) fn apply_reasoning_effort(
             | ApiProvider::Openrouter
             | ApiProvider::Novita
             | ApiProvider::Fireworks
-            | ApiProvider::Sglang => {
+            | ApiProvider::Sglang
+            | ApiProvider::Vllm => {
                 body["reasoning_effort"] = json!("high");
                 body["thinking"] = json!({ "type": "enabled" });
             }
@@ -786,7 +788,8 @@ pub(super) fn apply_reasoning_effort(
             | ApiProvider::Openrouter
             | ApiProvider::Novita
             | ApiProvider::Fireworks
-            | ApiProvider::Sglang => {
+            | ApiProvider::Sglang
+            | ApiProvider::Vllm => {
                 body["reasoning_effort"] = json!("max");
                 body["thinking"] = json!({ "type": "enabled" });
             }
diff --git a/crates/tui/src/commands/provider.rs b/crates/tui/src/commands/provider.rs
index ac279035..1ec5a025 100644
--- a/crates/tui/src/commands/provider.rs
+++ b/crates/tui/src/commands/provider.rs
@@ -27,7 +27,7 @@ pub fn provider(app: &mut App, args: Option<&str>) -> CommandResult {
 
     let Some(target) = ApiProvider::parse(name) else {
         return CommandResult::error(format!(
-            "Unknown provider '{name}'. Expected: deepseek, nvidia-nim, openrouter, novita, fireworks, or sglang."
+            "Unknown provider '{name}'. Expected: deepseek, nvidia-nim, openrouter, novita, fireworks, sglang, or vllm."
         ));
     };
 
@@ -164,6 +164,19 @@ mod tests {
         }
     }
 
+    #[test]
+    fn switch_to_vllm_flash_emits_action() {
+        let mut app = create_test_app();
+        let result = provider(&mut app, Some("vllm flash"));
+        match result.action {
+            Some(AppAction::SwitchProvider { provider, model }) => {
+                assert_eq!(provider, ApiProvider::Vllm);
+                assert_eq!(model.as_deref(), Some("deepseek-v4-flash"));
+            }
+            other => panic!("expected SwitchProvider, got {other:?}"),
+        }
+    }
+
     #[test]
     fn switching_to_active_provider_without_model_is_a_noop() {
         let mut app = create_test_app();
diff --git a/crates/tui/src/config.rs b/crates/tui/src/config.rs
index 83db0131..30f72a70 100644
--- a/crates/tui/src/config.rs
+++ b/crates/tui/src/config.rs
@@ -30,6 +30,9 @@ pub const DEFAULT_FIREWORKS_BASE_URL: &str = "https://api.fireworks.ai/inference
 pub const DEFAULT_SGLANG_MODEL: &str = "deepseek-ai/DeepSeek-V4-Pro";
 pub const DEFAULT_SGLANG_FLASH_MODEL: &str = "deepseek-ai/DeepSeek-V4-Flash";
 pub const DEFAULT_SGLANG_BASE_URL: &str = "http://localhost:30000/v1";
+pub const DEFAULT_VLLM_MODEL: &str = "deepseek-ai/DeepSeek-V4-Pro";
+pub const DEFAULT_VLLM_FLASH_MODEL: &str = "deepseek-ai/DeepSeek-V4-Flash";
+pub const DEFAULT_VLLM_BASE_URL: &str = "http://localhost:8000/v1";
 pub const DEFAULT_DEEPSEEKCN_BASE_URL: &str = "https://api.deepseeki.com";
 const API_KEYRING_SENTINEL: &str = "__KEYRING__";
 pub const COMMON_DEEPSEEK_MODELS: &[&str] = &[
@@ -51,6 +54,7 @@ pub enum ApiProvider {
     Novita,
     Fireworks,
     Sglang,
+    Vllm,
 }
 
 impl ApiProvider {
@@ -66,6 +70,7 @@ impl ApiProvider {
             "novita" => Some(Self::Novita),
             "fireworks" | "fireworks-ai" => Some(Self::Fireworks),
             "sglang" | "sg-lang" => Some(Self::Sglang),
+            "vllm" | "v-llm" => Some(Self::Vllm),
             _ => None,
         }
     }
@@ -80,6 +85,7 @@ impl ApiProvider {
             Self::Novita => "novita",
             Self::Fireworks => "fireworks",
             Self::Sglang => "sglang",
+            Self::Vllm => "vllm",
         }
     }
 
@@ -94,6 +100,7 @@ impl ApiProvider {
             Self::Novita => "Novita AI",
             Self::Fireworks => "Fireworks AI",
             Self::Sglang => "SGLang",
+            Self::Vllm => "vLLM",
         }
     }
 
@@ -108,6 +115,7 @@ impl ApiProvider {
             Self::Novita,
             Self::Fireworks,
             Self::Sglang,
+            Self::Vllm,
         ]
     }
 }
@@ -902,6 +910,8 @@ pub struct ProvidersConfig {
     pub fireworks: ProviderConfig,
     #[serde(default)]
     pub sglang: ProviderConfig,
+    #[serde(default)]
+    pub vllm: ProviderConfig,
 }
 
 #[derive(Debug, Clone, Deserialize, Default)]
@@ -961,7 +971,7 @@ impl Config {
             && ApiProvider::parse(provider).is_none()
         {
             anyhow::bail!(
-                "Invalid provider '{provider}': expected deepseek, deepseek-cn, nvidia-nim, openrouter, novita, fireworks, or sglang."
+                "Invalid provider '{provider}': expected deepseek, deepseek-cn, nvidia-nim, openrouter, novita, fireworks, sglang, or vllm."
             );
         }
         if let Some(ref key) = self.api_key
@@ -1079,6 +1089,7 @@ impl Config {
             ApiProvider::Novita => &providers.novita,
             ApiProvider::Fireworks => &providers.fireworks,
             ApiProvider::Sglang => &providers.sglang,
+            ApiProvider::Vllm => &providers.vllm,
         })
     }
 
@@ -1114,6 +1125,7 @@ impl Config {
             ApiProvider::Novita => DEFAULT_NOVITA_MODEL,
             ApiProvider::Fireworks => DEFAULT_FIREWORKS_MODEL,
             ApiProvider::Sglang => DEFAULT_SGLANG_MODEL,
+            ApiProvider::Vllm => DEFAULT_VLLM_MODEL,
         }
         .to_string()
     }
@@ -1139,7 +1151,8 @@ impl Config {
             ApiProvider::Openrouter
             | ApiProvider::Novita
             | ApiProvider::Fireworks
-            | ApiProvider::Sglang => None,
+            | ApiProvider::Sglang
+            | ApiProvider::Vllm => None,
         };
         let base = provider_base.or(root_base).unwrap_or_else(|| {
             match provider {
@@ -1150,6 +1163,7 @@ impl Config {
                 ApiProvider::Novita => DEFAULT_NOVITA_BASE_URL,
                 ApiProvider::Fireworks => DEFAULT_FIREWORKS_BASE_URL,
                 ApiProvider::Sglang => DEFAULT_SGLANG_BASE_URL,
+                ApiProvider::Vllm => DEFAULT_VLLM_BASE_URL,
             }
             .to_string()
         });
@@ -1173,6 +1187,7 @@ impl Config {
             ApiProvider::Novita => "novita",
             ApiProvider::Fireworks => "fireworks",
             ApiProvider::Sglang => "sglang",
+            ApiProvider::Vllm => "vllm",
         };
 
         // 0. Explicit in-memory override (set by onboarding / provider
@@ -1236,7 +1251,7 @@ impl Config {
             // Self-hosted SGLang deployments commonly run without auth on
             // localhost. Return an empty key and let the client omit the
             // Authorization header.
-            ApiProvider::Sglang => Ok(String::new()),
+            ApiProvider::Sglang | ApiProvider::Vllm => Ok(String::new()),
         }
     }
 
@@ -1678,11 +1693,26 @@ fn apply_env_overrides(config: &mut Config) {
             .sglang
             .base_url = Some(value);
     }
+    if matches!(config.api_provider(), ApiProvider::Vllm)
+        && let Ok(value) = std::env::var("VLLM_BASE_URL")
+        && !value.trim().is_empty()
+    {
+        config
+            .providers
+            .get_or_insert_with(ProvidersConfig::default)
+            .vllm
+            .base_url = Some(value);
+    }
     if matches!(config.api_provider(), ApiProvider::Sglang)
         && let Ok(value) = std::env::var("SGLANG_MODEL")
     {
         config.default_text_model = Some(value);
     }
+    if matches!(config.api_provider(), ApiProvider::Vllm)
+        && let Ok(value) = std::env::var("VLLM_MODEL")
+    {
+        config.default_text_model = Some(value);
+    }
     if let Ok(value) =
         std::env::var("DEEPSEEK_MODEL").or_else(|_| std::env::var("DEEPSEEK_DEFAULT_TEXT_MODEL"))
     {
@@ -1902,6 +1932,11 @@ fn normalize_model_config(config: &mut Config) {
         {
             providers.sglang.model = Some(normalized);
         }
+        if let Some(model) = providers.vllm.model.as_deref()
+            && let Some(normalized) = normalize_model_for_provider(ApiProvider::Vllm, model)
+        {
+            providers.vllm.model = Some(normalized);
+        }
     }
 }
 
@@ -1927,6 +1962,8 @@ fn model_for_provider(provider: ApiProvider, normalized: String) -> String {
         }
         (ApiProvider::Sglang, "deepseek-v4-pro") => DEFAULT_SGLANG_MODEL.to_string(),
         (ApiProvider::Sglang, "deepseek-v4-flash") => DEFAULT_SGLANG_FLASH_MODEL.to_string(),
+        (ApiProvider::Vllm, "deepseek-v4-pro") => DEFAULT_VLLM_MODEL.to_string(),
+        (ApiProvider::Vllm, "deepseek-v4-flash") => DEFAULT_VLLM_FLASH_MODEL.to_string(),
         _ => normalized,
     }
 }
@@ -2067,6 +2104,7 @@ fn merge_providers(
             novita: merge_provider_config(base.novita, override_cfg.novita),
             fireworks: merge_provider_config(base.fireworks, override_cfg.fireworks),
             sglang: merge_provider_config(base.sglang, override_cfg.sglang),
+            vllm: merge_provider_config(base.vllm, override_cfg.vllm),
         }),
     }
 }
@@ -2436,6 +2474,7 @@ pub fn has_api_key_for(config: &Config, provider: ApiProvider) -> bool {
         ApiProvider::Novita => "NOVITA_API_KEY",
         ApiProvider::Fireworks => "FIREWORKS_API_KEY",
         ApiProvider::Sglang => "SGLANG_API_KEY",
+        ApiProvider::Vllm => "VLLM_API_KEY",
     };
     if std::env::var(env_var).is_ok_and(|k| !k.trim().is_empty()) {
         return true;
@@ -2447,7 +2486,7 @@ pub fn has_api_key_for(config: &Config, provider: ApiProvider) -> bool {
     }
 
     // SGLang is self-hosted and typically runs without authentication.
-    if matches!(provider, ApiProvider::Sglang) {
+    if matches!(provider, ApiProvider::Sglang | ApiProvider::Vllm) {
         return true;
     }
 
@@ -2494,6 +2533,7 @@ pub fn save_api_key_for(provider: ApiProvider, api_key: &str) -> Result<PathBuf>
         ApiProvider::Novita => "providers.novita",
         ApiProvider::Fireworks => "providers.fireworks",
         ApiProvider::Sglang => "providers.sglang",
+        ApiProvider::Vllm => "providers.vllm",
     };
 
     // Parse existing TOML (or start fresh) so we can edit the right table
@@ -2521,6 +2561,7 @@ pub fn save_api_key_for(provider: ApiProvider, api_key: &str) -> Result<PathBuf>
         ApiProvider::Novita => "novita",
         ApiProvider::Fireworks => "fireworks",
         ApiProvider::Sglang => "sglang",
+        ApiProvider::Vllm => "vllm",
     };
     let entry = providers
         .entry(key_inside.to_string())
@@ -2634,6 +2675,9 @@ mod tests {
         sglang_api_key: Option<OsString>,
         sglang_base_url: Option<OsString>,
         sglang_model: Option<OsString>,
+        vllm_api_key: Option<OsString>,
+        vllm_base_url: Option<OsString>,
+        vllm_model: Option<OsString>,
     }
 
     impl EnvGuard {
@@ -2664,6 +2708,9 @@ mod tests {
             let sglang_api_key_prev = env::var_os("SGLANG_API_KEY");
             let sglang_base_url_prev = env::var_os("SGLANG_BASE_URL");
             let sglang_model_prev = env::var_os("SGLANG_MODEL");
+            let vllm_api_key_prev = env::var_os("VLLM_API_KEY");
+            let vllm_base_url_prev = env::var_os("VLLM_BASE_URL");
+            let vllm_model_prev = env::var_os("VLLM_MODEL");
             // Safety: test-only environment mutation guarded by a global mutex.
             unsafe {
                 env::set_var("HOME", &home_str);
@@ -2689,6 +2736,9 @@ mod tests {
                 env::remove_var("SGLANG_API_KEY");
                 env::remove_var("SGLANG_BASE_URL");
                 env::remove_var("SGLANG_MODEL");
+                env::remove_var("VLLM_API_KEY");
+                env::remove_var("VLLM_BASE_URL");
+                env::remove_var("VLLM_MODEL");
             }
             Self {
                 home: home_prev,
@@ -2714,6 +2764,9 @@ mod tests {
                 sglang_api_key: sglang_api_key_prev,
                 sglang_base_url: sglang_base_url_prev,
                 sglang_model: sglang_model_prev,
+                vllm_api_key: vllm_api_key_prev,
+                vllm_base_url: vllm_base_url_prev,
+                vllm_model: vllm_model_prev,
             }
         }
     }
@@ -2748,6 +2801,9 @@ mod tests {
                 Self::restore_var("SGLANG_API_KEY", self.sglang_api_key.take());
                 Self::restore_var("SGLANG_BASE_URL", self.sglang_base_url.take());
                 Self::restore_var("SGLANG_MODEL", self.sglang_model.take());
+                Self::restore_var("VLLM_API_KEY", self.vllm_api_key.take());
+                Self::restore_var("VLLM_BASE_URL", self.vllm_base_url.take());
+                Self::restore_var("VLLM_MODEL", self.vllm_model.take());
             }
         }
     }
diff --git a/crates/tui/src/main.rs b/crates/tui/src/main.rs
index c5978bf8..33270850 100644
--- a/crates/tui/src/main.rs
+++ b/crates/tui/src/main.rs
@@ -1301,6 +1301,10 @@ fn run_setup_status(config: &Config, workspace: &Path) -> Result<()> {
                     "SGLANG_API_KEY",
                     "deepseek auth set --provider sglang --api-key \"...\"",
                 ),
+                crate::config::ApiProvider::Vllm => (
+                    "VLLM_API_KEY",
+                    "deepseek auth set --provider vllm --api-key \"...\"",
+                ),
                 crate::config::ApiProvider::Deepseek | crate::config::ApiProvider::DeepseekCN => {
                     ("DEEPSEEK_API_KEY", "deepseek auth set --provider deepseek")
                 }
@@ -1314,6 +1318,7 @@ fn run_setup_status(config: &Config, workspace: &Path) -> Result<()> {
                     crate::config::ApiProvider::Novita => "novita",
                     crate::config::ApiProvider::Fireworks => "fireworks",
                     crate::config::ApiProvider::Sglang => "sglang",
+                    crate::config::ApiProvider::Vllm => "vllm",
                     crate::config::ApiProvider::Deepseek
                     | crate::config::ApiProvider::DeepseekCN => "deepseek",
                 }
@@ -1544,6 +1549,11 @@ async fn run_doctor(config: &Config, workspace: &Path, config_path_override: Opt
             "sglang",
             &["SGLANG_API_KEY"][..],
         ),
+        (
+            crate::config::ApiProvider::Vllm,
+            "vllm",
+            &["VLLM_API_KEY"][..],
+        ),
     ] {
         let in_env = env_names.iter().any(|n| {
             std::env::var(n)
diff --git a/crates/tui/src/tui/provider_picker.rs b/crates/tui/src/tui/provider_picker.rs
index ac13a7b3..9eead1b6 100644
--- a/crates/tui/src/tui/provider_picker.rs
+++ b/crates/tui/src/tui/provider_picker.rs
@@ -92,6 +92,7 @@ impl ProviderPickerView {
             ApiProvider::Novita => "NOVITA_API_KEY",
             ApiProvider::Fireworks => "FIREWORKS_API_KEY",
             ApiProvider::Sglang => "SGLANG_API_KEY",
+            ApiProvider::Vllm => "VLLM_API_KEY",
         }
     }
 
@@ -374,7 +375,8 @@ mod tests {
                 "OpenRouter",
                 "Novita AI",
                 "Fireworks AI",
-                "SGLang"
+                "SGLang",
+                "vLLM"
             ]
         );
     }
diff --git a/crates/tui/src/tui/ui.rs b/crates/tui/src/tui/ui.rs
index d7cfacec..14f4d8e5 100644
--- a/crates/tui/src/tui/ui.rs
+++ b/crates/tui/src/tui/ui.rs
@@ -4507,6 +4507,7 @@ async fn execute_command_input(
             providers.novita.api_key = None;
             providers.fireworks.api_key = None;
             providers.sglang.api_key = None;
+            providers.vllm.api_key = None;
         }
         app.api_key_env_only = crate::config::active_provider_uses_env_only_api_key(config);
     }
@@ -4884,6 +4885,7 @@ fn render(f: &mut Frame, app: &mut App) {
             crate::config::ApiProvider::Novita => Some("Novita"),
             crate::config::ApiProvider::Fireworks => Some("Fireworks"),
             crate::config::ApiProvider::Sglang => Some("SGLang"),
+            crate::config::ApiProvider::Vllm => Some("vLLM"),
         };
         let header_data = HeaderData::new(
             app.mode,
@@ -5510,6 +5512,7 @@ async fn apply_provider_picker_api_key(
             ApiProvider::Novita => &mut providers.novita,
             ApiProvider::Fireworks => &mut providers.fireworks,
             ApiProvider::Sglang => &mut providers.sglang,
+            ApiProvider::Vllm => &mut providers.vllm,
         };
         entry.api_key = Some(api_key);
     }
diff --git a/docs/CONFIGURATION.md b/docs/CONFIGURATION.md
index 9768e5a5..b30e9ff1 100644
--- a/docs/CONFIGURATION.md
+++ b/docs/CONFIGURATION.md
@@ -56,12 +56,12 @@ the legacy `deepseek login --api-key ...` alias) saves the key to
 to the TUI as `DEEPSEEK_MODEL`.
 
 For hosted or self-hosted DeepSeek V4 providers, set `provider = "nvidia-nim"`,
-`"fireworks"`, or `"sglang"` or pass `deepseek --provider <name>`. The facade
+`"fireworks"`, `"sglang"`, or `"vllm"` or pass `deepseek --provider <name>`. The facade
 saves provider credentials to the shared user config and forwards the resolved
 key, base URL, provider, and model to the TUI process. Use
 `deepseek auth set --provider nvidia-nim --api-key "YOUR_NVIDIA_API_KEY"` or
 `deepseek auth set --provider fireworks --api-key "YOUR_FIREWORKS_API_KEY"` to
-save hosted-provider keys through the facade. SGLang is self-hosted and can run
+save hosted-provider keys through the facade. SGLang and vLLM are self-hosted and can run
 without an API key by default.
 
 To bootstrap MCP and skills directories at their resolved paths, run `deepseek-tui setup`.
@@ -99,6 +99,11 @@ default_text_model = "accounts/fireworks/models/deepseek-v4-pro"
 provider = "sglang"
 base_url = "http://localhost:30000/v1"
 default_text_model = "deepseek-ai/DeepSeek-V4-Pro"
+
+[profiles.vllm]
+provider = "vllm"
+base_url = "http://localhost:8000/v1"
+default_text_model = "deepseek-ai/DeepSeek-V4-Pro"
 ```
 
 Select a profile with:
@@ -114,7 +119,7 @@ These override config values:
 
 - `DEEPSEEK_API_KEY`
 - `DEEPSEEK_BASE_URL`
-- `DEEPSEEK_PROVIDER` (`deepseek|nvidia-nim|openrouter|novita|fireworks|sglang`)
+- `DEEPSEEK_PROVIDER` (`deepseek|nvidia-nim|openrouter|novita|fireworks|sglang|vllm`)
 - `DEEPSEEK_MODEL` or `DEEPSEEK_DEFAULT_TEXT_MODEL`
 - `NVIDIA_API_KEY` or `NVIDIA_NIM_API_KEY` (preferred when provider is `nvidia-nim`; falls back to `DEEPSEEK_API_KEY`)
 - `NVIDIA_NIM_BASE_URL`, `NIM_BASE_URL`, or `NVIDIA_BASE_URL`
@@ -124,6 +129,9 @@ These override config values:
 - `SGLANG_BASE_URL`
 - `SGLANG_MODEL`
 - `SGLANG_API_KEY` (optional; many localhost SGLang servers do not require auth)
+- `VLLM_BASE_URL`
+- `VLLM_MODEL`
+- `VLLM_API_KEY` (optional; many localhost vLLM servers do not require auth)
 - `DEEPSEEK_LOG_LEVEL` or `RUST_LOG` (`info`/`debug`/`trace` enables lightweight verbose logs)
 - `DEEPSEEK_SKILLS_DIR`
 - `DEEPSEEK_MCP_CONFIG`
@@ -294,7 +302,7 @@ If you are upgrading from older releases:
 
 ### Core keys (used by the TUI/engine)
 
-- `provider` (string, optional): `deepseek` (default), `nvidia-nim`, `openrouter`, `novita`, `fireworks`, or `sglang`. `nvidia-nim` targets NVIDIA's NIM-hosted DeepSeek endpoints through `https://integrate.api.nvidia.com/v1`; `fireworks` targets `https://api.fireworks.ai/inference/v1`; `sglang` targets a self-hosted OpenAI-compatible endpoint, defaulting to `http://localhost:30000/v1`.
+- `provider` (string, optional): `deepseek` (default), `nvidia-nim`, `openrouter`, `novita`, `fireworks`, `sglang`, or `vllm`. `nvidia-nim` targets NVIDIA's NIM-hosted DeepSeek endpoints through `https://integrate.api.nvidia.com/v1`; `fireworks` targets `https://api.fireworks.ai/inference/v1`; `sglang` targets a self-hosted OpenAI-compatible endpoint, defaulting to `http://localhost:30000/v1`; `vllm` targets a self-hosted vLLM OpenAI-compatible endpoint, defaulting to `http://localhost:8000/v1`.
 - `api_key` (string, required): must be non-empty (or set `DEEPSEEK_API_KEY`).
 - `base_url` (string, optional): defaults to `https://api.deepseek.com` for DeepSeek's OpenAI-compatible Chat Completions API, or `https://integrate.api.nvidia.com/v1` for `provider = "nvidia-nim"`. `https://api.deepseek.com/v1` is also accepted for SDK compatibility; use `https://api.deepseek.com/beta` only for DeepSeek beta features such as strict tool mode, chat prefix completion, and FIM completion.
 - `default_text_model` (string, optional): defaults to `deepseek-v4-pro` for DeepSeek, `deepseek-ai/deepseek-v4-pro` for NVIDIA NIM, `accounts/fireworks/models/deepseek-v4-pro` for Fireworks, and `deepseek-ai/DeepSeek-V4-Pro` for SGLang. Current public DeepSeek IDs are `deepseek-v4-pro` and `deepseek-v4-flash`, both with 1M context windows and thinking mode enabled by default. Legacy `deepseek-chat` and `deepseek-reasoner` remain compatibility aliases for `deepseek-v4-flash`. Provider-specific mappings translate `deepseek-v4-pro` / `deepseek-v4-flash` to each provider's model ID where supported. Use `/models` or `deepseek models` to discover live IDs from your configured endpoint. `DEEPSEEK_MODEL` overrides this for a single process.