feat(providers): add Gondola as declarative OpenAI-compatible provider (#11421)

Signed-off-by: Abhijay Jain <Abhijay007j@gmail.com>
2026-08-21 01:47:18 +00:00
parent f45ccd46f7
commit 8e8b674cf0
5 changed files with 99 additions and 0 deletions
@@ -0,0 +1,75 @@
+use super::api_client::{ApiClient, AuthMethod};
+use super::base::{ConfigKey, ProviderDef, ProviderMetadata};
+use super::openai_compatible::OpenAiCompatibleProvider;
+use anyhow::Result;
+use futures::future::BoxFuture;
+
+const GONDOLA_PROVIDER_NAME: &str = "gondola";
+pub const GONDOLA_API_HOST: &str = "https://api.gondola-ai.com/v1";
+pub const GONDOLA_DEFAULT_MODEL: &str = "deepseek-v4-flash";
+pub const GONDOLA_KNOWN_MODELS: &[&str] = &[
+    "claude-opus-5",
+    "claude-sonnet-5",
+    "deepseek-v3.2",
+    "deepseek-v4-flash",
+    "deepseek-v4-pro",
+    "e2ee-deepseek-v4-flash",
+    "e2ee-qwen3-6-35b-a3b",
+    "kimi-k2-5",
+    "kimi-k3",
+    "llama-3.3-70b",
+    "qwen3-235b-a22b-instruct-2507",
+];
+pub const GONDOLA_DOC_URL: &str = "https://gondola-ai.com/guides";
+
+pub struct GondolaProvider;
+
+impl goose_providers::base::ProviderDescriptor for GondolaProvider {
+    fn metadata() -> ProviderMetadata {
+        ProviderMetadata::new(
+            GONDOLA_PROVIDER_NAME,
+            "Gondola",
+            "Pay-per-request inference via Venice AI, settled in USDC. Privacy-preserving with TEE model options.",
+            GONDOLA_DEFAULT_MODEL,
+            GONDOLA_KNOWN_MODELS.to_vec(),
+            GONDOLA_DOC_URL,
+            vec![
+                ConfigKey::new("GONDOLA_API_KEY", true, true, None, true),
+                ConfigKey::new("GONDOLA_HOST", false, false, Some(GONDOLA_API_HOST), false),
+            ],
+        )
+        .with_setup(
+            crate::providers::catalog::ProviderSetupMetadata::api_key(
+                crate::providers::catalog::ProviderSetupGroup::Default,
+            )
+            .with_docs_url(GONDOLA_DOC_URL),
+        )
+    }
+}
+
+impl ProviderDef for GondolaProvider {
+    type Provider = OpenAiCompatibleProvider;
+
+    fn from_env(
+        _extensions: Vec<crate::config::ExtensionConfig>,
+        tls_config: Option<crate::providers::api_client::TlsConfig>,
+    ) -> BoxFuture<'static, Result<OpenAiCompatibleProvider>> {
+        Box::pin(async move {
+            let config = crate::config::Config::global();
+            let api_key: String = config.get_secret("GONDOLA_API_KEY")?;
+            let host: String = config
+                .get_param("GONDOLA_HOST")
+                .unwrap_or_else(|_| GONDOLA_API_HOST.to_string());
+
+            let api_client =
+                ApiClient::new_with_tls(host, AuthMethod::BearerToken(api_key), tls_config)?
+                    .with_request_builder(crate::session_context::session_id_request_builder());
+
+            Ok(OpenAiCompatibleProvider::new(
+                GONDOLA_PROVIDER_NAME.to_string(),
+                api_client,
+                String::new(),
+            ))
+        })
+    }
+}
@@ -23,6 +23,7 @@ use super::{
    gemini_cli::GeminiCliProvider,
    gemini_oauth::GeminiOAuthProvider,
    githubcopilot::GithubCopilotProvider,
+    gondola::GondolaProvider,
    huggingface::HuggingFaceProvider,
    kimicode::KimiCodeProvider,
    litellm::LiteLLMProvider,
@@ -121,6 +122,7 @@ async fn init_registry() -> RwLock<ProviderRegistry> {
            false,
            Some(registrations::refresh_only()),
        );
+        registry.register::<GondolaProvider>(false);
        registry.register_with_inventory::<GoogleProviderDef>(
            true,
            Some(registrations::google_inventory()),
@@ -329,6 +331,21 @@ mod tests {
            .any(|key| key.name == "HF_TOKEN" && key.secret));
    }

+    #[tokio::test]
+    async fn test_gondola_provider_registry_wiring() {
+        let gondola = get_from_registry("gondola")
+            .await
+            .expect("gondola provider should be registered");
+        let meta = gondola.metadata();
+
+        assert_eq!(meta.name, "gondola");
+        assert_eq!(meta.default_model, "deepseek-v4-flash");
+        assert!(meta
+            .config_keys
+            .iter()
+            .any(|key| key.name == "GONDOLA_API_KEY" && key.secret));
+    }
+
    #[tokio::test]
    async fn test_openai_compatible_providers_config_keys() {
        let providers_list = providers().await;
@@ -42,6 +42,7 @@ pub mod githubcopilot;
 pub mod google {
    pub use goose_providers::google::*;
 }
+pub mod gondola;
 pub mod google_def;
 pub mod http_status {
    pub use goose_providers::http_status::*;
@@ -38,6 +38,7 @@ goose is compatible with a wide range of LLM providers, allowing you to choose a
 | [Gemini](https://ai.google.dev/gemini-api/docs)                             | Advanced LLMs by Google with multimodal capabilities (text, images). Gemini 3 models support configurable [thinking levels](#gemini-3-thinking-levels).                                                                                                | `GOOGLE_API_KEY`, `GEMINI3_THINKING_LEVEL` (optional)                                                                                                                              |
 | [GCP Vertex AI](https://cloud.google.com/vertex-ai)                         | Google Cloud's Vertex AI platform, supporting Gemini and Claude models. **Credentials must be [configured in advance](https://cloud.google.com/vertex-ai/docs/authentication).** Filters for allowed models by organization policy (if configured). | `GCP_PROJECT_ID`, `GCP_LOCATION` and optionally `GCP_MAX_RATE_LIMIT_RETRIES` (5), `GCP_MAX_OVERLOADED_RETRIES` (5), `GCP_INITIAL_RETRY_INTERVAL_MS` (5000), `GCP_BACKOFF_MULTIPLIER` (2.0), `GCP_MAX_RETRY_INTERVAL_MS` (320_000). |
 | [GitHub Copilot](https://docs.github.com/en/copilot/using-github-copilot/ai-models) | Access to AI models from OpenAI, Anthropic, Google, and other providers through GitHub's Copilot infrastructure. **GitHub account with Copilot access required.** | No manual key. Uses [device flow authentication](#github-copilot-authentication) for both CLI and Desktop. |
+| [Gondola](https://gondola-ai.com/guides)                                     | Pay-per-request inference via Venice AI, settled in USDC on Base. No subscription or minimum. Includes privacy-preserving TEE models (`e2ee-*`). OpenAI-compatible.                                                      | `GONDOLA_API_KEY`, `GONDOLA_HOST` (optional)                                                                                                                                        |
 | [Groq](https://groq.com/)                                                   | High-performance inference hardware and tools for LLMs.                                                                                                                                                                   | `GROQ_API_KEY`                                                                                                                                                                      |
 | [iFlytek Spark](https://www.xfyun.cn/doc/spark/HTTP%E8%B0%83%E7%94%A8%E6%96%87%E6%A1%A3.html) | iFlytek Spark (讯飞星火) models (4.0Ultra, generalv3.5, max-32k) via the OpenAI-compatible HTTP API. Best for chat: Spark needs `tool_calls_switch=true` (not injectable here) to return OpenAI-style tool calls. | `SPARK_API_PASSWORD` |
 | [iFlytek Astron MaaS](https://maas.xfyun.cn/)                               | iFlytek Astron MaaS (讯飞星辰) hosting Spark X2, DeepSeek, GLM, Kimi, MiniMax, Qwen, and Astron coding models via an OpenAI-compatible API. Set `ASTRON_BASE_URL` to switch between the Token Plan and Coding Plan endpoints. | `ASTRON_API_KEY`, `ASTRON_BASE_URL` (optional) |
@@ -169,6 +169,11 @@ prompt: |
  4. Verify MaaS requires its bound model and uses /v1/chat/completions.
  If the source checkout or Rust toolchain is unavailable, mark this validation as skipped rather than failed.

+  ### Gondola Provider Validation
+  When running from a goose source checkout:
+  1. Run `cargo test -p goose test_gondola_provider_registry_wiring` to verify Gondola is registered with the correct name, default model (`deepseek-v4-flash`), and a secret `GONDOLA_API_KEY` config key.
+  If the source checkout or Rust toolchain is unavailable, mark this validation as skipped rather than failed.
+
  Log results to: {{ workspace_dir }}/phase1_basic_tools.md
  {% endif %}