Add Together AI platform + bare-array listing tolerance (provider 9)

The 12th registry row: id "together", TOGETHER_API_KEY > auth.json "together" scope, https://api.together.xyz/v1 with KIGI_TOGETHER_BASE_URL override, Bearer, ChatCompletions + Passthrough, enrichment-backed metadata (models_dev_id togetherai) with the tool-calling listing restriction (Together's listing mixes chat/embedding/rerank/image types). Together's GET /v1/models returns a BARE JSON ARRAY, not the OpenAI {object:list,data:[]} envelope. New shared parser parse_openai_listing tolerates both shapes (sniffs the top-level [ vs { for accurate diagnostics); every OpenAI-listing provider now routes through it, with byte-equivalent envelope behavior (verified: Groq/Google/OpenRouter unchanged) and non-silent errors. Together's org/Model ids match the togetherai snapshot keys exactly (verified), so restrict_to_enriched keeps the ~27 tool-calling models without the id-shape trap. Review: ship-ready, shared-parser change proven strictly-additive and safe, registry/e2e/dialect all correct. Backlog logged: kigi ignores the wire per-model field, so a live Together chat model absent from models.dev is dropped until indexed — a future wire- filter would unlock the fresh full catalog.
2026-07-21 13:16:20 -04:00
parent e7dbb98207
commit 0a147bd8a5
4 changed files with 190 additions and 10 deletions
@@ -603,7 +603,8 @@ mod tests {
                "mistral",
                "fireworks",
                "google",
-                "openrouter"
+                "openrouter",
+                "together"
            ]
        );
        assert_eq!(default_id(&built), Some(XAI_API_KEY_METHOD_ID));
@@ -637,7 +638,8 @@ mod tests {
                "mistral",
                "fireworks",
                "google",
-                "openrouter"
+                "openrouter",
+                "together"
            ]
        );
        assert_eq!(default_id(&built), Some(CACHED_TOKEN_AUTH_METHOD_ID));
@@ -664,7 +666,8 @@ mod tests {
                "mistral",
                "fireworks",
                "google",
-                "openrouter"
+                "openrouter",
+                "together"
            ]
        );
        assert_eq!(default_id(&built), Some(CACHED_TOKEN_AUTH_METHOD_ID));
@@ -694,7 +697,8 @@ mod tests {
                "mistral",
                "fireworks",
                "google",
-                "openrouter"
+                "openrouter",
+                "together"
            ]
        );
        assert_eq!(default_id(&built), None);
@@ -297,7 +297,13 @@ fn fetch_one_platform_models(
        .map(|s| s.to_string());
    let data = match platform.listing() {
        kigi_models::ListingDialect::OpenAi => {
-            response.json::<kigi_models::WireModelsResponse>()?.data
+            // Tolerant of both the {data:[...]} envelope and a bare array
+            // (Together AI serves the bare form).
+            let body = response.text()?;
+            kigi_models::parse_openai_listing(&body).map_err(|e| BackendError::RequestFailed {
+                status: 200,
+                body: format!("openai listing parse failed: {e}"),
+            })?
        }
        kigi_models::ListingDialect::Anthropic => {
            let body = response.text()?;
@@ -1597,6 +1603,98 @@ mod tests {
        );
    }

+    /// Together-cycle e2e: the listing is a BARE JSON ARRAY (no {data:[]}
+    /// envelope) — parse_openai_listing tolerates it. models.dev enrichment
+    /// (matching org/Model keys) supplies context + the tool-calling
+    /// restriction drops non-chat models; Passthrough dialect; slashed id.
+    #[tokio::test(flavor = "multi_thread")]
+    #[serial_test::serial]
+    async fn together_bare_array_listing_enriches_restricts_and_maps_passthrough() {
+        let platform_server = wiremock::MockServer::start().await;
+        wiremock::Mock::given(wiremock::matchers::method("GET"))
+            .and(wiremock::matchers::path("/models"))
+            .and(wiremock::matchers::header("Authorization", "Bearer tg-1"))
+            .respond_with(wiremock::ResponseTemplate::new(200).set_body_json(
+                // BARE ARRAY, not {object:list,data:[]}.
+                serde_json::json!([
+                    { "id": "Qwen/Qwen3-Coder-480B-A35B-Instruct-FP8", "object": "model",
+                      "type": "chat", "context_length": 262144 },
+                    { "id": "togethercomputer/m2-bert-80M-8k-retrieval", "object": "model",
+                      "type": "embedding" }
+                ]),
+            ))
+            .expect(1)
+            .mount(&platform_server)
+            .await;
+        let modelsdev_server = wiremock::MockServer::start().await;
+        wiremock::Mock::given(wiremock::matchers::method("GET"))
+            .and(wiremock::matchers::path("/api.json"))
+            .respond_with(wiremock::ResponseTemplate::new(200).set_body_json(
+                serde_json::json!({ "togetherai": { "models": {
+                    "Qwen/Qwen3-Coder-480B-A35B-Instruct-FP8": {
+                        "limit": {"context": 262144, "output": 32768},
+                        "tool_call": true
+                    },
+                    "togethercomputer/m2-bert-80M-8k-retrieval": { "limit": {"context": 8192} }
+                }}}),
+            ))
+            .expect(1)
+            .mount(&modelsdev_server)
+            .await;
+        let cache_dir = tempfile::tempdir().unwrap();
+        let _base = kigi_test_support::EnvGuard::set(
+            kigi_models::TOGETHER_BASE_URL_ENV,
+            platform_server.uri(),
+        );
+        let _mdev = kigi_test_support::EnvGuard::set(
+            crate::agent::enrichment_fetch::MODELS_DEV_URL_ENV,
+            format!("{}/api.json", modelsdev_server.uri()),
+        );
+        let _mdev_cache = kigi_test_support::EnvGuard::set(
+            crate::agent::enrichment_fetch::MODELS_DEV_CACHE_DIR_ENV,
+            cache_dir.path(),
+        );
+        let endpoints = crate::agent::config::EndpointsConfig::default();
+        let keys = crate::agent::models::PlatformApiKeys::test_single(
+            kigi_models::PlatformId::Together,
+            "tg-1",
+        );
+        let result = tokio::task::spawn_blocking(move || {
+            fetch_platform_models_blocking(&endpoints, None, &keys)
+        })
+        .await
+        .unwrap()
+        .expect("fetch must succeed");
+        assert_eq!(
+            result
+                .models
+                .iter()
+                .map(|m| m.id.as_deref().unwrap_or_default())
+                .collect::<Vec<_>>(),
+            vec!["together/Qwen/Qwen3-Coder-480B-A35B-Instruct-FP8"],
+            "bare array parsed; embedding (not tool-calling) dropped; slashed id in key"
+        );
+        let entry = &result.models[0];
+        assert_eq!(entry.context_window.get(), 262_144);
+        assert_eq!(entry.max_completion_tokens, Some(32_768));
+        assert_eq!(
+            entry.model, "Qwen/Qwen3-Coder-480B-A35B-Instruct-FP8",
+            "the native slashed id rides the wire"
+        );
+        let model_entry = crate::agent::config::ModelEntry::from_config_entry(entry);
+        let creds = crate::agent::config::ResolvedCredentials {
+            api_key: Some("tg-1".into()),
+            base_url: entry.base_url.clone(),
+            auth_type: kigi_chat_state::AuthType::ApiKey,
+            auth_scheme: Default::default(),
+        };
+        let cfg = crate::agent::config::sampling_config_for_model(&model_entry, creds, None);
+        assert_eq!(
+            cfg.chat_compat,
+            kigi_sampling_types::ChatCompat::Passthrough
+        );
+    }
+
    #[test]
    fn get_env_keys_parses_strings_and_rejects_non_strings() {
        use crate::agent::config::EnvKeys;