feat(models): add StepFun provider preset (#7988)

* feat(models): add StepFun provider preset

* fix(models): standardize the StepFun preset label

---------

Co-authored-by: Willem Jiang <willem.jiang@gmail.com>

CHANGELOG.md

@@ -22,6 +22,10 @@ This section accumulates work toward the **2.3.0** milestone
   native subagents pause and resume. Unconfirmed messages can be retried
   after a Gateway restart. ([#7729])
 
+- **models:** Administrators can select StepFun in **Settings → Models** and
+  connect an ordinary StepFun API account with an API key and model ID.
+  ([#7988](https://github.com/bytedance/deer-flow/pull/7988))
+
 ### Fixed
 
 - **community:** A search or fetch API key that ends in a newline no longer

CHANGELOG_zh.md

@@ -21,6 +21,9 @@
   Queue 等当前任务结束后再执行下一条消息；Steer 让主 Agent 先处理新要求，原生子 Agent 可暂停后恢复。
   Gateway 重启后，尚未确认生效的消息可以手动重试。([#7729])
 
+- **模型：** 管理员可在 **设置 → 模型** 中选择 StepFun（阶跃星辰），
+  填写 API Key 和模型 ID 即可接入普通 API。([#7988](https://github.com/bytedance/deer-flow/pull/7988))
+
 ### 修复
 
 - **社区工具：** 以换行结尾的搜索或抓取 API 密钥不再泄露到对话中。GroundRoute

README.md

@@ -335,7 +335,7 @@ fences, including unclosed fences, in complete and streamed replies.
    Administrators can open **Settings → Models** to select a provider and enter
    only its **API Key** and **model ID**. Endpoints, authentication and adapters are
    backend presets for OpenAI, Anthropic, DeepSeek, Alibaba Cloud Bailian/Qwen,
-   Volcengine/Doubao, Moonshot/Kimi, Zhipu/Z.AI and MiniMax. Region and plan-specific
+   Volcengine/Doubao, Moonshot/Kimi, Zhipu/Z.AI, MiniMax and StepFun. Region and plan-specific
    entries are explicitly distinguished. Arbitrary model IDs are accepted; internal
    names are generated automatically and remain stable when a model ID changes.
    Newly saved models are enabled and immediately refresh existing chat and Agent

README_zh.md

@@ -109,7 +109,7 @@ DeerFlow 新近集成了 BytePlus 自研的智能搜索与抓取工具集——[
 
 管理员可在 **设置 → 模型 → 添加模型** 中选择厂商，仅填写 **API Key** 和
 **模型 ID**。后端预置 OpenAI、Anthropic、DeepSeek、百炼/通义千问、火山/豆包、
-Moonshot/Kimi、智谱/Z.AI 和 MiniMax 的端点、认证与适配器，地域和 Coding Plan
+Moonshot/Kimi、智谱/Z.AI、MiniMax 和阶跃星辰（StepFun）的端点、认证与适配器，地域和 Coding Plan
 使用明确区分的选项。模型 ID 不受内置枚举限制；内部名称自动生成且保持稳定。
 新模型默认启用，保存后自动刷新聊天及 Agent 的模型选择器，无需重启 Gateway。
 

backend/docs/CONFIGURATION.md

@@ -334,6 +334,7 @@ are:
 | `zai` | Z.AI, international ordinary API | `https://api.z.ai/api/paas/v4` | Chat Completions / Bearer |
 | `minimax` | MiniMax, international API | `https://api.minimax.io/v1` | Chat Completions / Bearer |
 | `minimax-cn` | MiniMax, China API | `https://api.minimaxi.com/v1` | Chat Completions / Bearer |
+| `stepfun` | StepFun, ordinary API | `https://api.stepfun.com/v1` | Chat Completions / Bearer |
 | `openai-compatible` | Custom OpenAI-compatible, including unauthenticated local endpoints | Administrator-supplied HTTP(S) base URL | Chat Completions / optional key |
 
 Keys for different regions and Coding Plans are not interchangeable. These
@@ -342,6 +343,16 @@ Plan from the model name. Plans/regions without a preset can use the custom
 OpenAI-compatible option if their documented protocol is compatible; native
 protocols need a verified backend preset or operator YAML configuration.
 
+The StepFun preset uses the existing `PatchedChatStepFun` adapter to capture
+`reasoning` / `reasoning_content` in responses and replay reasoning history.
+Both connection probes and chat send the documented `max_tokens` output bound.
+Model IDs remain free-form; this provider registration does not infer thinking,
+vision or context-window capabilities from their names. The preset is for
+ordinary API credentials. Step Plan has a separate `/step_plan/v1` endpoint;
+use the custom OpenAI-compatible option for that channel. See the official
+[OpenAI migration guide](https://platform.stepfun.com/docs/zh/guides/developer/openai)
+and [Chat Completions reference](https://platform.stepfun.com/docs/zh/api-reference/chat/chat-completion-create).
+
 Settings → Models → **Fetch and add models** reads the connection's model
 catalog without changing configuration. Administrators can reuse an existing
 saved connection without exposing its key, search IDs, select visible results,

backend/packages/harness/deerflow/config/managed_model_providers.py

@@ -12,6 +12,7 @@ from urllib.parse import urlsplit
 _OPENAI = "langchain_openai:ChatOpenAI"
 _DEEPSEEK = "deerflow.models.patched_deepseek:PatchedChatDeepSeek"
 _MINIMAX = "deerflow.models.patched_minimax:PatchedChatMiniMax"
+_STEPFUN = "deerflow.models.patched_stepfun:PatchedChatStepFun"
 MOONSHOT_K31_MODELS = frozenset({"kimi-k3.1-pro", "kimi-k3.1-pro-highspeed"})
 OPENAI_GPT6_MODELS = frozenset({"gpt-6-astra", "gpt-6.1-sol", "gpt-6.1-terra", "gpt-6-sol", "gpt-6-luna"})
 
@@ -45,6 +46,7 @@ PROVIDERS = (
     ManagedModelProvider("zai", "Z.AI (International, API)", "https://api.z.ai/api/paas/v4", _DEEPSEEK, endpoint_parameter="api_base"),
     ManagedModelProvider("minimax", "MiniMax (International, API)", "https://api.minimax.io/v1", _MINIMAX, defaults={"temperature": 1.0}),
     ManagedModelProvider("minimax-cn", "MiniMax (China, API)", "https://api.minimaxi.com/v1", _MINIMAX, defaults={"temperature": 1.0}),
+    ManagedModelProvider("stepfun", "StepFun", "https://api.stepfun.com/v1", _STEPFUN),
     ManagedModelProvider("openai-compatible", "Custom OpenAI-compatible", None),
 )
 _PROVIDER_BY_ID = {provider.id: provider for provider in PROVIDERS}

backend/packages/harness/deerflow/models/AGENTS.md

@@ -267,6 +267,8 @@ provider settings/reasoning transforms as chat. Disable optional thinking for
 short probes; required thinking and unknown third-party models use automatic
 tool choice. MiniMax's existing adapter normalizes the OpenAI SDK's
 `max_completion_tokens` into MiniMax's documented `max_tokens` for both paths.
+The `stepfun` preset reuses `PatchedChatStepFun` for reasoning capture/replay and
+the same token-field normalization; its ordinary API route is distinct from Step Plan.
 Tests: `test_managed_models.py`, `test_managed_deepseek.py`,
 `test_managed_provider_presets.py` (real vendor SDKs with HTTP doubles), opt-in
 `test_managed_deepseek_live.py` / `test_managed_presets_live.py`, and

backend/packages/harness/deerflow/models/patched_stepfun.py

@@ -83,6 +83,10 @@ class PatchedChatStepFun(ChatOpenAI):
         """Restore ``reasoning_content`` on historical assistant messages."""
         original_messages = self._convert_input(input_).to_messages()
         payload = super()._get_request_payload(input_, stop=stop, **kwargs)
+        # StepFun documents max_tokens; ChatOpenAI emits the newer OpenAI name.
+        # Preserve the configured output bound on probes and ordinary chat calls.
+        if "max_completion_tokens" in payload:
+            payload["max_tokens"] = payload.pop("max_completion_tokens")
 
         restore_assistant_payloads(
             payload.get("messages", []),

backend/tests/test_managed_provider_presets.py

@@ -35,6 +35,7 @@ def store(tmp_path, monkeypatch):
         ("moonshot-cn", "https://api.moonshot.cn/v1", "deerflow.models.patched_deepseek:PatchedChatDeepSeek"),
         ("zai", "https://api.z.ai/api/paas/v4", "deerflow.models.patched_deepseek:PatchedChatDeepSeek"),
         ("minimax", "https://api.minimax.io/v1", "deerflow.models.patched_minimax:PatchedChatMiniMax"),
+        ("stepfun", "https://api.stepfun.com/v1", "deerflow.models.patched_stepfun:PatchedChatStepFun"),
     ],
 )
 def test_preset_routing_and_conservative_unknown_capabilities(provider, endpoint, adapter):
@@ -150,6 +151,7 @@ def test_preset_overrides_and_injected_config_remain_authoritative(store):
 @pytest.mark.asyncio
 async def test_provider_metadata_requires_admin():
     catalog = await router.list_model_providers(_ADMIN)
+    assert {"id": "stepfun", "display_name": "StepFun", "requires_api_key": True, "custom_endpoint": False} in catalog["providers"]
     assert {p["id"] for p in catalog["providers"]} == {p.id for p in PROVIDERS}
     assert all(set(p) == {"id", "display_name", "requires_api_key", "custom_endpoint"} for p in catalog["providers"])
     with pytest.raises(HTTPException) as exc:
@@ -333,7 +335,8 @@ def test_generated_name_collision_with_yaml_is_retried(store, monkeypatch):
 
 
 @pytest.mark.asyncio
-async def test_preset_api_save_refresh_yaml_conflict_and_disable(store, tmp_path, monkeypatch):
+@pytest.mark.parametrize("provider", ["openai", "stepfun"])
+async def test_preset_api_save_refresh_yaml_conflict_and_disable(store, tmp_path, monkeypatch, provider):
     from unittest.mock import AsyncMock
 
     from app.gateway.routers import models as models_router
@@ -358,7 +361,7 @@ async def test_preset_api_save_refresh_yaml_conflict_and_disable(store, tmp_path
     snapshot = get_app_config()
     try:
         async with httpx.AsyncClient(transport=httpx.ASGITransport(app=app), base_url="http://test") as client:
-            create = await client.put("/api/managed-models", json={"config": {"provider": "openai", "model": "org/新 模型", "api_key": "private-test-key"}})
+            create = await client.put("/api/managed-models", json={"config": {"provider": provider, "model": "org/新 模型", "api_key": "private-test-key"}})
             assert create.status_code == 200
             saved = create.json()
             assert saved["enabled"] and saved["has_api_key"]
@@ -370,7 +373,7 @@ async def test_preset_api_save_refresh_yaml_conflict_and_disable(store, tmp_path
             assert snapshot.models[0].api_key == "yaml-test-key"
             assert len(snapshot.models) == 1
             assert len(next_snapshot.models) == 2
-            edit = {"config": {"name": saved["name"], "provider": "openai", "model": "new-model"}, "expected_revision": saved["revision"]}
+            edit = {"config": {"name": saved["name"], "provider": provider, "model": "new-model"}, "expected_revision": saved["revision"]}
             changed = await client.put("/api/managed-models", json=edit)
             assert changed.status_code == 200
             assert changed.json()["name"] == saved["name"]

backend/tests/test_model_catalog.py

@@ -27,17 +27,18 @@ def connection(**kw):
 
 
 @pytest.mark.asyncio
-async def test_discovery_uses_saved_key_and_never_writes(setup, monkeypatch):
-    saved = setup.save(ManagedModel(provider="openai", model="existing", api_key="catalog-secret"), expected_revision=None)
+@pytest.mark.parametrize("provider,endpoint", [("openai", "https://api.openai.com/v1/models"), ("stepfun", "https://api.stepfun.com/v1/models")])
+async def test_discovery_uses_saved_key_and_never_writes(setup, monkeypatch, provider, endpoint):
+    saved = setup.save(ManagedModel(provider=provider, model="existing", api_key="catalog-secret"), expected_revision=None)
     before = setup.path.read_bytes()
 
     async def send(client, request, **kwargs):
-        assert str(request.url) == "https://api.openai.com/v1/models"
+        assert str(request.url) == endpoint
         assert request.headers["authorization"] == "Bearer catalog-secret"
         return httpx.Response(200, request=request, json={"data": [{"id": "existing"}, {"id": "new"}, {"id": "new"}, {"id": "bad\n"}, {"id": 5}]})
 
     monkeypatch.setattr(httpx.AsyncClient, "send", send)
-    body = router.ModelConnection(provider="openai", saved_model_name=saved.name, expected_revision=saved.revision)
+    body = router.ModelConnection(provider=provider, saved_model_name=saved.name, expected_revision=saved.revision)
     result = await router.discover_models(ADMIN, body)
     assert result == {"models": [{"id": "existing", "configured": True}, {"id": "new", "configured": False}], "truncated": False}
     assert setup.path.read_bytes() == before

backend/tests/test_patched_stepfun.py

@@ -4,6 +4,7 @@ from __future__ import annotations
 
 from unittest.mock import MagicMock, patch
 
+import pytest
 from langchain_core.messages import AIMessage, AIMessageChunk, HumanMessage
 
 
@@ -35,6 +36,21 @@ def test_lc_secrets_contains_stepfun_api_key_mapping():
     assert model.lc_secrets["openai_api_key"] == "STEPFUN_API_KEY"
 
 
+@pytest.mark.parametrize("limit_field", ["max_tokens", "max_completion_tokens"])
+def test_request_uses_documented_max_tokens_for_instance_and_call_limits(limit_field):
+    model = _make_model(max_tokens=128)
+    messages = [HumanMessage(content="Hello")]
+    initial = model._get_request_payload(messages)
+    assert initial["max_tokens"] == 128
+    assert "max_completion_tokens" not in initial
+    overrides = {limit_field: 64}
+    payload = model._get_request_payload(messages, **overrides)
+    assert payload["max_tokens"] == 64
+    assert "max_completion_tokens" not in payload
+    assert overrides == {limit_field: 64}
+    assert model.max_tokens == 128
+
+
 # ---------------------------------------------------------------------------
 # _extract_reasoning helper
 # ---------------------------------------------------------------------------