feat(config): allow provider path suffix overrides

crates/config/src/lib.rs

@@ -158,6 +158,7 @@ impl ProviderKind {
 pub struct ProviderConfigToml {
     pub api_key: Option<String>,
     pub base_url: Option<String>,
+    pub path_suffix: Option<String>,
     pub model: Option<String>,
     pub auth_mode: Option<String>,
     #[serde(default)]
@@ -1355,6 +1356,9 @@ impl ConfigToml {
 }
 
 fn merge_project_provider_config(target: &mut ProviderConfigToml, source: &ProviderConfigToml) {
+    if source.path_suffix.is_some() {
+        target.path_suffix = source.path_suffix.clone();
+    }
     if source.model.is_some() {
         target.model = source.model.clone();
     }

crates/tui/src/client.rs

@@ -125,6 +125,7 @@ pub struct DeepSeekClient {
     pub(super) http_client: reqwest::Client,
     api_key: String,
     pub(super) base_url: String,
+    api_path_suffix: Option<String>,
     pub(super) api_provider: ApiProvider,
     retry: RetryPolicy,
     default_model: String,
@@ -291,6 +292,7 @@ impl Clone for DeepSeekClient {
             http_client: self.http_client.clone(),
             api_key: self.api_key.clone(),
             base_url: self.base_url.clone(),
+            api_path_suffix: self.api_path_suffix.clone(),
             api_provider: self.api_provider,
             retry: self.retry.clone(),
             default_model: self.default_model.clone(),
@@ -390,8 +392,25 @@ fn is_version_segment(segment: &str) -> bool {
             .is_some_and(|rest| !rest.is_empty() && rest.chars().all(|ch| ch.is_ascii_digit()))
 }
 
+#[cfg(test)]
 pub(super) fn api_url(base_url: &str, path: &str) -> String {
+    api_url_with_path_suffix(base_url, None, path)
+}
+
+pub(super) fn api_url_with_path_suffix(
+    base_url: &str,
+    path_suffix: Option<&str>,
+    path: &str,
+) -> String {
     let path = path.trim_start_matches('/');
+    if let Some(path_suffix) = path_suffix {
+        let base = base_url.trim_end_matches('/');
+        let suffix = path_suffix.trim().trim_matches('/');
+        if suffix.is_empty() {
+            return format!("{base}/{path}");
+        }
+        return format!("{base}/{suffix}/{path}");
+    }
     if path.starts_with("beta/") {
         return format!("{}/{}", unversioned_base_url(base_url), path);
     }
@@ -465,10 +484,15 @@ fn add_extra_root_certs(
 }
 
 impl DeepSeekClient {
+    fn api_url(&self, path: &str) -> String {
+        api_url_with_path_suffix(&self.base_url, self.api_path_suffix.as_deref(), path)
+    }
+
     /// Create a DeepSeek client from CLI configuration.
     pub fn new(config: &Config) -> Result<Self> {
         let api_key = config.deepseek_api_key()?;
         let base_url = config.deepseek_base_url();
+        let api_path_suffix = config.api_path_suffix();
         let api_provider = config.api_provider();
         validate_base_url_security(&base_url)?;
         let retry = config.retry_policy();
@@ -494,6 +518,7 @@ impl DeepSeekClient {
             http_client,
             api_key,
             base_url,
+            api_path_suffix,
             api_provider,
             retry,
             default_model,
@@ -585,7 +610,7 @@ impl DeepSeekClient {
         model: &str,
         target_language: &str,
     ) -> Result<String> {
-        let url = api_url(&self.base_url, "chat/completions");
+        let url = self.api_url("chat/completions");
         let model = wire_model_for_provider(self.api_provider, model);
         let mut body = serde_json::json!({
             "model": model,
@@ -632,7 +657,7 @@ impl DeepSeekClient {
 
     /// List available models from the provider.
     pub async fn list_models(&self) -> Result<Vec<AvailableModel>> {
-        let url = api_url(&self.base_url, "models");
+        let url = self.api_url("models");
         let response = self.send_with_retry(|| self.http_client.get(&url)).await?;
 
         let status = response.status();
@@ -679,7 +704,7 @@ impl DeepSeekClient {
         if !should_probe {
             return;
         }
-        let health_url = api_url(&self.base_url, "models");
+        let health_url = self.api_url("models");
         let probe = self.http_client.get(health_url).send().await;
         match probe {
             Ok(resp) if resp.status().is_success() => {
@@ -790,7 +815,7 @@ impl LlmClient for DeepSeekClient {
     }
 
     async fn health_check(&self) -> Result<bool> {
-        let health_url = api_url(&self.base_url, "models");
+        let health_url = self.api_url("models");
         self.wait_for_rate_limit().await;
         let response = self.http_client.get(health_url).send().await;
         match response {
@@ -1096,7 +1121,7 @@ impl DeepSeekClient {
         suffix: &str,
         max_tokens: u32,
     ) -> anyhow::Result<String> {
-        let url = api_url(&self.base_url, "beta/completions");
+        let url = self.api_url("beta/completions");
         let model = wire_model_for_provider(self.api_provider, model);
         let body = json!({
             "model": model,
@@ -1236,6 +1261,34 @@ mod tests {
         );
     }
 
+    #[test]
+    fn api_url_respects_explicit_path_suffix() {
+        assert_eq!(
+            api_url_with_path_suffix(
+                "https://openai-compatible.example",
+                Some(""),
+                "chat/completions"
+            ),
+            "https://openai-compatible.example/chat/completions"
+        );
+        assert_eq!(
+            api_url_with_path_suffix(
+                "https://openai-compatible.example",
+                Some("/openai/v2/"),
+                "chat/completions"
+            ),
+            "https://openai-compatible.example/openai/v2/chat/completions"
+        );
+        assert_eq!(
+            api_url_with_path_suffix(
+                "https://openai-compatible.example",
+                None,
+                "chat/completions"
+            ),
+            "https://openai-compatible.example/v1/chat/completions"
+        );
+    }
+
     #[test]
     fn api_url_routes_beta_paths_from_any_deepseek_base() {
         assert_eq!(

crates/tui/src/client/chat.rs

@@ -69,7 +69,7 @@ use crate::models::{
 
 use super::{
     DeepSeekClient, ERROR_BODY_MAX_BYTES, SSE_BACKPRESSURE_HIGH_WATERMARK,
-    SSE_BACKPRESSURE_SLEEP_MS, SSE_MAX_LINES_PER_CHUNK, acquire_stream_buffer, api_url,
+    SSE_BACKPRESSURE_SLEEP_MS, SSE_MAX_LINES_PER_CHUNK, acquire_stream_buffer,
     apply_reasoning_effort, bounded_error_text, from_api_tool_name, parse_usage,
     release_stream_buffer, system_to_instructions, to_api_tool_name,
 };
@@ -136,7 +136,7 @@ impl DeepSeekClient {
             self.api_provider,
         );
 
-        let url = api_url(&self.base_url, "chat/completions");
+        let url = self.api_url("chat/completions");
         let open_timeout = stream_open_timeout();
         let response = match tokio_timeout(
             open_timeout,
@@ -239,7 +239,7 @@ impl DeepSeekClient {
             self.api_provider,
         );
 
-        let url = api_url(&self.base_url, "chat/completions");
+        let url = self.api_url("chat/completions");
         let response = self
             .send_with_retry(|| self.http_client.post(&url).json(&body))
             .await?;

crates/tui/src/config.rs

@@ -1663,6 +1663,7 @@ impl LspConfigToml {
 pub struct ProviderConfig {
     pub api_key: Option<String>,
     pub base_url: Option<String>,
+    pub path_suffix: Option<String>,
     pub model: Option<String>,
     pub auth_mode: Option<String>,
     pub http_headers: Option<HashMap<String, String>>,
@@ -2166,6 +2167,13 @@ impl Config {
         normalize_base_url(&base)
     }
 
+    #[must_use]
+    pub fn api_path_suffix(&self) -> Option<String> {
+        self.provider_config()
+            .and_then(|provider| provider.path_suffix.as_deref())
+            .map(|suffix| suffix.trim().trim_matches('/').to_string())
+    }
+
     fn active_provider_preserves_custom_base_url_model(&self) -> bool {
         let provider = self.api_provider();
         provider_preserves_custom_base_url_model(provider, &self.deepseek_base_url())
@@ -3830,6 +3838,7 @@ fn merge_provider_config(base: ProviderConfig, override_cfg: ProviderConfig) ->
     ProviderConfig {
         api_key: override_cfg.api_key.or(base.api_key),
         base_url: override_cfg.base_url.or(base.base_url),
+        path_suffix: override_cfg.path_suffix.or(base.path_suffix),
         model: override_cfg.model.or(base.model),
         auth_mode: override_cfg.auth_mode.or(base.auth_mode),
         http_headers: override_cfg.http_headers.or(base.http_headers),
@@ -7169,6 +7178,7 @@ model = "account-model-id"
 [providers.openai]
 api_key = "openai-table-key"
 base_url = "https://openai-compatible.example/api/coding/paas/v4"
+path_suffix = ""
 model = "glm-5"
 "#,
         )?;
@@ -7180,6 +7190,7 @@ model = "glm-5"
             config.deepseek_base_url(),
             "https://openai-compatible.example/api/coding/paas/v4"
         );
+        assert_eq!(config.api_path_suffix().as_deref(), Some(""));
         assert_eq!(config.default_model(), "glm-5");
         Ok(())
     }

docs/CONFIGURATION.md

@@ -116,6 +116,16 @@ api_key = "YOUR_OPENAI_COMPATIBLE_API_KEY"
 base_url = "https://your-gateway.example/v1"
 ```
 
+If your gateway expects unversioned paths such as `/chat/completions` instead
+of `/v1/chat/completions`, keep `base_url` at the host root and set an explicit
+empty path suffix:
+
+```toml
+[providers.openai]
+base_url = "https://your-gateway.example"
+path_suffix = ""
+```
+
 Do not invent a custom provider name; `provider` must be one of the known
 providers listed above. Put the endpoint under `[providers.openai]`, not the
 legacy top-level `base_url`, so the OpenAI-compatible provider receives it.
@@ -590,6 +600,7 @@ If you are upgrading from older releases:
 - `provider` (string, optional): `deepseek` (default), `nvidia-nim`, `openai`, `atlascloud`, `wanjie-ark`, `openrouter`, `xiaomi-mimo`, `novita`, `fireworks`, `siliconflow`, `moonshot`, `sglang`, `vllm`, or `ollama`. Legacy `deepseek-cn` configs are still accepted as an alias for `deepseek`; DeepSeek uses the same official host [`https://api.deepseek.com`](https://api-docs.deepseek.com/) worldwide. `nvidia-nim` targets NVIDIA's NIM-hosted DeepSeek endpoints through `https://integrate.api.nvidia.com/v1`; `openai` targets a generic OpenAI-compatible endpoint, defaulting to `https://api.openai.com/v1`; `atlascloud` targets AtlasCloud's OpenAI-compatible endpoint at `https://api.atlascloud.ai/v1`; `wanjie-ark` targets Wanjie Ark's OpenAI-compatible endpoint at `https://maas-openapi.wanjiedata.com/api/v1`; `openrouter` targets `https://openrouter.ai/api/v1`; `xiaomi-mimo` targets Xiaomi MiMo's OpenAI-compatible endpoint at `https://api.xiaomimimo.com/v1`; `novita` targets `https://api.novita.ai/v1`; `fireworks` targets `https://api.fireworks.ai/inference/v1`; `siliconflow` targets SiliconFlow, defaulting to `https://api.siliconflow.com/v1`; `moonshot` targets Moonshot/Kimi, defaulting to `https://api.moonshot.ai/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`; `ollama` targets Ollama's OpenAI-compatible endpoint, defaulting to `http://localhost:11434/v1`.
 - `api_key` (string, required for hosted providers): must be non-empty for DeepSeek/hosted providers (or set the provider API key env var). Self-hosted SGLang, vLLM, and Ollama can omit it.
 - `base_url` (string, optional): defaults to `https://api.deepseek.com/beta` for DeepSeek's OpenAI-compatible Chat Completions API, including legacy `provider = "deepseek-cn"` configs. Other defaults are `https://integrate.api.nvidia.com/v1` for `nvidia-nim`, `https://api.openai.com/v1` for `openai`, `https://api.atlascloud.ai/v1` for `atlascloud`, `https://maas-openapi.wanjiedata.com/api/v1` for `wanjie-ark`, `https://openrouter.ai/api/v1` for `openrouter`, `https://api.xiaomimimo.com/v1` for `xiaomi-mimo`, `https://api.novita.ai/v1` for `novita`, `https://api.fireworks.ai/inference/v1` for `fireworks`, `https://api.siliconflow.com/v1` for `siliconflow`, `https://api.moonshot.ai/v1` for `moonshot`, `http://localhost:30000/v1` for `sglang`, `http://localhost:8000/v1` for `vllm`, and `http://localhost:11434/v1` for `ollama`. Set `https://api.deepseek.com` or `https://api.deepseek.com/v1` explicitly to opt out of DeepSeek beta features.
+- `path_suffix` (string, optional): provider-table override for the path segment inserted between `base_url` and API paths. When unset, CodeWhale keeps the current `/v1` behavior. Set `path_suffix = ""` for gateways that expect `/chat/completions` directly under `base_url`.
 - `default_text_model` (string, optional): defaults to `deepseek-v4-pro` for DeepSeek and generic OpenAI-compatible endpoints, `deepseek-ai/deepseek-v4-pro` for NVIDIA NIM, `deepseek-ai/deepseek-v4-flash` for AtlasCloud, `deepseek-reasoner` for Wanjie Ark, `deepseek/deepseek-v4-pro` for OpenRouter and Novita, `mimo-v2.5-pro` for Xiaomi MiMo, `accounts/fireworks/models/deepseek-v4-pro` for Fireworks, `deepseek-ai/DeepSeek-V4-Pro` for SiliconFlow, `kimi-k2.6` for Moonshot, `deepseek-ai/DeepSeek-V4-Pro` for SGLang/vLLM, and `deepseek-coder:1.3b` for Ollama. Current public DeepSeek IDs are `deepseek-v4-pro` and `deepseek-v4-flash`, both with 1M context windows, 384K max output, and thinking mode enabled by default. Legacy `deepseek-chat` and `deepseek-reasoner` remain compatibility aliases for `deepseek-v4-flash` until July 24, 2026, except SiliconFlow maps `deepseek-reasoner` and `deepseek-r1` to its Pro model while `deepseek-chat` and `deepseek-v3` map to Flash. Provider-specific mappings translate `deepseek-v4-pro` / `deepseek-v4-flash` to each provider's model ID where supported. OpenRouter also recognizes recent large IDs such as `arcee-ai/trinity-large-thinking`, `minimax/minimax-m3`, `xiaomi/mimo-v2.5-pro`, `qwen/qwen3.6-35b-a3b`, `google/gemma-4-31b-it`, and `moonshotai/kimi-k2.6`. Generic `openai`, `atlascloud`, `wanjie-ark`, `xiaomi-mimo`, and Ollama model IDs are passed through unchanged. OpenRouter and SiliconFlow provider configs with a custom `base_url` also preserve explicit model values, which lets OpenAI-compatible gateways accept bare model IDs. Use `/models` or `codewhale models` to discover live IDs from your configured endpoint. `CODEWHALE_MODEL` overrides this for a single process; `DEEPSEEK_MODEL` is the legacy alias.
 - `reasoning_effort` (string, optional): `off`, `low`, `medium`, `high`, or `max`; defaults to the configured UI tier. DeepSeek Platform receives top-level `thinking` / `reasoning_effort` fields. NVIDIA NIM receives equivalent settings through `chat_template_kwargs`.
 - `allow_shell` (bool, optional): defaults to `true` (sandboxed).