feat(providers): add DashScope and Bailian Coding providers with reasoning_content support

- Add DashScope (Qwen) native provider with tools+streaming fallback
- Add Bailian Coding provider with hardcoded model list (no /v1/models API)
- Parse reasoning_content in OpenAI-compat streaming/non-streaming responses
- Emit ChatEventThinking events in agent loop for thinking models
- Add vision support for DashScope (qwen3-vl)
- Fix provider form dialog not updating API base URL when switching types
- Update README provider count from 11+ to 13+

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

README.md

@@ -6,7 +6,7 @@
 
 [![Go](https://img.shields.io/badge/Go_1.25-00ADD8?style=flat-square&logo=go&logoColor=white)](https://go.dev/) [![PostgreSQL](https://img.shields.io/badge/PostgreSQL_18-316192?style=flat-square&logo=postgresql&logoColor=white)](https://www.postgresql.org/) [![Docker](https://img.shields.io/badge/Docker-2496ED?style=flat-square&logo=docker&logoColor=white)](https://www.docker.com/) [![WebSocket](https://img.shields.io/badge/WebSocket-010101?style=flat-square&logo=socket.io&logoColor=white)](https://developer.mozilla.org/en-US/docs/Web/API/WebSocket) [![OpenTelemetry](https://img.shields.io/badge/OpenTelemetry-000000?style=flat-square&logo=opentelemetry&logoColor=white)](https://opentelemetry.io/) [![Anthropic](https://img.shields.io/badge/Anthropic-191919?style=flat-square&logo=anthropic&logoColor=white)](https://www.anthropic.com/) [![OpenAI](https://img.shields.io/badge/OpenAI_Compatible-412991?style=flat-square&logo=openai&logoColor=white)](https://openai.com/) [![License: MIT](https://img.shields.io/badge/License-MIT-yellow?style=flat-square)](LICENSE)
 
-**GoClaw** is a multi-agent AI gateway that connects LLMs to your tools, channels, and data — deployed as a single Go binary with zero runtime dependencies. It orchestrates agent teams, inter-agent delegation, and quality-gated workflows across 11+ LLM providers with full multi-tenant isolation.
+**GoClaw** is a multi-agent AI gateway that connects LLMs to your tools, channels, and data — deployed as a single Go binary with zero runtime dependencies. It orchestrates agent teams, inter-agent delegation, and quality-gated workflows across 13+ LLM providers with full multi-tenant isolation.
 
 A Go port of [OpenClaw](https://github.com/openclaw/openclaw) with enhanced security, multi-tenant PostgreSQL, and production-grade observability.
 
@@ -16,7 +16,7 @@ A Go port of [OpenClaw](https://github.com/openclaw/openclaw) with enhanced secu
 - **Multi-Tenant PostgreSQL** — Per-user workspaces, per-user context files, encrypted API keys (AES-256-GCM), isolated sessions — the only Claw project with DB-native multi-tenancy
 - **Single Binary** — ~25 MB static Go binary, no Node.js runtime, <1s startup, runs on a $5 VPS
 - **Production Security** — 5-layer defense: rate limiting, prompt injection detection, SSRF protection, shell deny patterns, AES-256-GCM encryption
-- **11+ LLM Providers** — Anthropic (native HTTP+SSE with prompt caching), OpenAI, OpenRouter, Groq, DeepSeek, Gemini, Mistral, xAI, MiniMax, Cohere, Perplexity
+- **13+ LLM Providers** — Anthropic (native HTTP+SSE with prompt caching), OpenAI, OpenRouter, Groq, DeepSeek, Gemini, Mistral, xAI, MiniMax, Cohere, Perplexity, DashScope (Qwen), Bailian Coding
 - **5 Messaging Channels** — Telegram, Discord, Zalo, Feishu/Lark, WhatsApp with `/stop` and `/stopall` commands
 
 ## Claw Ecosystem
@@ -51,7 +51,7 @@ A Go port of [OpenClaw](https://github.com/openclaw/openclaw) with enhanced secu
 | Messaging channels         | 37+                                  | 15+                                          | 10+                                   | 5+                             |
 | Companion apps             | macOS, iOS, Android                  | Python SDK                                   | —                                     | Web dashboard                  |
 | Live Canvas / Voice        | ✅ (A2UI + TTS/STT)                  | —                                            | Voice transcription                   | TTS (4 providers)              |
-| LLM providers              | 10+                                  | 8 native + 29 compat                         | 13+                                   | **11+**                        |
+| LLM providers              | 10+                                  | 8 native + 29 compat                         | 13+                                   | **13+**                        |
 | Per-user workspaces        | ✅ (file-based)                      | —                                            | —                                     | ✅ (managed mode only)         |
 | Encrypted secrets          | — (env vars only)                    | ✅ ChaCha20-Poly1305                         | — (plaintext JSON)                    | ✅ AES-256-GCM in DB           |
 
@@ -78,7 +78,7 @@ graph TB
         SCHED --> ROUTER["Agent Router"]
         ROUTER --> LOOP["Agent Loop<br/>think → act → observe"]
         LOOP --> TOOLS["Tool Registry<br/>fs · exec · web · memory · delegate · team · mcp · custom"]
-        LOOP --> LLM["LLM Providers<br/>Anthropic (native + prompt caching) · OpenAI-compat (10+)"]
+        LOOP --> LLM["LLM Providers<br/>Anthropic (native + prompt caching) · OpenAI-compat (12+)"]
     end
 
     subgraph Storage
@@ -294,7 +294,7 @@ Quality gates validate agent output before it reaches users. Configured in agent
 ## Features
 
 ### LLM Providers
-- **11+ providers** — OpenRouter, Anthropic, OpenAI, Groq, DeepSeek, Gemini, Mistral, xAI, MiniMax, Cohere, Perplexity, and any OpenAI-compatible endpoint
+- **13+ providers** — OpenRouter, Anthropic, OpenAI, Groq, DeepSeek, Gemini, Mistral, xAI, MiniMax, Cohere, Perplexity, DashScope (Qwen), Bailian Coding, and any OpenAI-compatible endpoint
 - **Anthropic native** — Direct HTTP+SSE integration with prompt caching (`cache_control`) for ~90% cost reduction on repeated prefixes
 - **OpenAI-compatible** — Automatic prompt caching for OpenAI, MiniMax, OpenRouter (cache metrics tracked in traces)
 
@@ -557,6 +557,8 @@ When `GOCLAW_*_API_KEY` environment variables are set, the gateway automatically
 | `GOCLAW_MINIMAX_API_KEY`    | MiniMax                  |
 | `GOCLAW_COHERE_API_KEY`     | Cohere                   |
 | `GOCLAW_PERPLEXITY_API_KEY` | Perplexity               |
+| `GOCLAW_DASHSCOPE_API_KEY`  | DashScope (Qwen)         |
+| `GOCLAW_BAILIAN_API_KEY`    | Bailian Coding           |
 
 </details>
 

cmd/gateway_providers.go

@@ -66,6 +66,20 @@ func registerProviders(registry *providers.Registry, cfg *config.Config) {
 		registry.Register(providers.NewOpenAIProvider("perplexity", cfg.Providers.Perplexity.APIKey, "https://api.perplexity.ai", "sonar-pro"))
 		slog.Info("registered provider", "name", "perplexity")
 	}
+
+	if cfg.Providers.DashScope.APIKey != "" {
+		registry.Register(providers.NewDashScopeProvider(cfg.Providers.DashScope.APIKey, cfg.Providers.DashScope.APIBase, "qwen3-max"))
+		slog.Info("registered provider", "name", "dashscope")
+	}
+
+	if cfg.Providers.Bailian.APIKey != "" {
+		base := cfg.Providers.Bailian.APIBase
+		if base == "" {
+			base = "https://coding-intl.dashscope.aliyuncs.com/v1"
+		}
+		registry.Register(providers.NewOpenAIProvider("bailian", cfg.Providers.Bailian.APIKey, base, "qwen3.5-plus"))
+		slog.Info("registered provider", "name", "bailian")
+	}
 }
 
 // registerProvidersFromDB loads providers from Postgres and registers them.
@@ -84,6 +98,14 @@ func registerProvidersFromDB(registry *providers.Registry, provStore store.Provi
 		if p.ProviderType == store.ProviderAnthropicNative {
 			registry.Register(providers.NewAnthropicProvider(p.APIKey,
 				providers.WithAnthropicBaseURL(p.APIBase)))
+		} else if p.ProviderType == store.ProviderDashScope {
+			registry.Register(providers.NewDashScopeProvider(p.APIKey, p.APIBase, ""))
+		} else if p.ProviderType == store.ProviderBailian {
+			base := p.APIBase
+			if base == "" {
+				base = "https://coding-intl.dashscope.aliyuncs.com/v1"
+			}
+			registry.Register(providers.NewOpenAIProvider(p.Name, p.APIKey, base, "qwen3.5-plus"))
 		} else {
 			prov := providers.NewOpenAIProvider(p.Name, p.APIKey, p.APIBase, "")
 			if p.ProviderType == store.ProviderMiniMax {

internal/agent/loop.go

@@ -553,6 +553,14 @@ func (l *Loop) runLoop(ctx context.Context, req RunRequest) (*RunResult, error)
 
 		if req.Stream {
 			resp, err = l.provider.ChatStream(ctx, chatReq, func(chunk providers.StreamChunk) {
+				if chunk.Thinking != "" {
+					l.emit(AgentEvent{
+						Type:    protocol.ChatEventThinking,
+						AgentID: l.id,
+						RunID:   req.RunID,
+						Payload: map[string]string{"content": chunk.Thinking},
+					})
+				}
 				if chunk.Content != "" {
 					l.emit(AgentEvent{
 						Type:    protocol.ChatEventChunk,

internal/config/config_channels.go

@@ -110,6 +110,8 @@ type ProvidersConfig struct {
 	MiniMax    ProviderConfig `json:"minimax"`
 	Cohere     ProviderConfig `json:"cohere"`
 	Perplexity ProviderConfig `json:"perplexity"`
+	DashScope  ProviderConfig `json:"dashscope"`
+	Bailian    ProviderConfig `json:"bailian"`
 }
 
 type ProviderConfig struct {
@@ -130,7 +132,9 @@ func (c *Config) HasAnyProvider() bool {
 		p.XAI.APIKey != "" ||
 		p.MiniMax.APIKey != "" ||
 		p.Cohere.APIKey != "" ||
-		p.Perplexity.APIKey != ""
+		p.Perplexity.APIKey != "" ||
+		p.DashScope.APIKey != "" ||
+		p.Bailian.APIKey != ""
 }
 
 // GatewayConfig controls the gateway server.

internal/config/config_load.go

@@ -105,6 +105,8 @@ func (c *Config) applyEnvOverrides() {
 	envStr("GOCLAW_MINIMAX_API_KEY", &c.Providers.MiniMax.APIKey)
 	envStr("GOCLAW_COHERE_API_KEY", &c.Providers.Cohere.APIKey)
 	envStr("GOCLAW_PERPLEXITY_API_KEY", &c.Providers.Perplexity.APIKey)
+	envStr("GOCLAW_DASHSCOPE_API_KEY", &c.Providers.DashScope.APIKey)
+	envStr("GOCLAW_BAILIAN_API_KEY", &c.Providers.Bailian.APIKey)
 	envStr("GOCLAW_GATEWAY_TOKEN", &c.Gateway.Token)
 	envStr("GOCLAW_TELEGRAM_TOKEN", &c.Channels.Telegram.Token)
 	envStr("GOCLAW_DISCORD_TOKEN", &c.Channels.Discord.Token)

internal/config/config_secrets.go

@@ -32,6 +32,8 @@ func (c *Config) MaskedCopy() *Config {
 	maskNonEmpty(&cp.Providers.MiniMax.APIKey)
 	maskNonEmpty(&cp.Providers.Cohere.APIKey)
 	maskNonEmpty(&cp.Providers.Perplexity.APIKey)
+	maskNonEmpty(&cp.Providers.DashScope.APIKey)
+	maskNonEmpty(&cp.Providers.Bailian.APIKey)
 
 	// Mask gateway token
 	maskNonEmpty(&cp.Gateway.Token)
@@ -77,6 +79,8 @@ func (c *Config) StripSecrets() {
 	c.Providers.MiniMax.APIKey = ""
 	c.Providers.Cohere.APIKey = ""
 	c.Providers.Perplexity.APIKey = ""
+	c.Providers.DashScope.APIKey = ""
+	c.Providers.Bailian.APIKey = ""
 
 	// Gateway token
 	c.Gateway.Token = ""
@@ -127,6 +131,8 @@ func (c *Config) StripMaskedSecrets() {
 	stripIfMasked(&c.Providers.MiniMax.APIKey)
 	stripIfMasked(&c.Providers.Cohere.APIKey)
 	stripIfMasked(&c.Providers.Perplexity.APIKey)
+	stripIfMasked(&c.Providers.DashScope.APIKey)
+	stripIfMasked(&c.Providers.Bailian.APIKey)
 
 	// Gateway token
 	stripIfMasked(&c.Gateway.Token)

internal/http/provider_models.go

@@ -51,6 +51,8 @@ func (h *ProvidersHandler) handleListProviderModels(w http.ResponseWriter, r *ht
 		models, err = fetchAnthropicModels(ctx, p.APIKey)
 	case "gemini_native":
 		models, err = fetchGeminiModels(ctx, p.APIKey)
+	case "bailian":
+		models = bailianModels()
 	default:
 		// All other types use OpenAI-compatible /models endpoint
 		apiBase := strings.TrimRight(p.APIBase, "/")
@@ -145,6 +147,22 @@ func fetchGeminiModels(ctx context.Context, apiKey string) ([]ModelInfo, error)
 	return models, nil
 }
 
+// bailianModels returns a hardcoded list of models available on the
+// Bailian Coding platform (coding-intl.dashscope.aliyuncs.com).
+// The platform does not expose a /v1/models endpoint.
+func bailianModels() []ModelInfo {
+	return []ModelInfo{
+		{ID: "qwen3.5-plus", Name: "Qwen 3.5 Plus"},
+		{ID: "kimi-k2.5", Name: "Kimi K2.5"},
+		{ID: "GLM-5", Name: "GLM-5"},
+		{ID: "MiniMax-M2.5", Name: "MiniMax M2.5"},
+		{ID: "qwen3-max-2026-01-23", Name: "Qwen 3 Max (2026-01-23)"},
+		{ID: "qwen3-coder-next", Name: "Qwen 3 Coder Next"},
+		{ID: "qwen3-coder-plus", Name: "Qwen 3 Coder Plus"},
+		{ID: "glm-4.7", Name: "GLM 4.7"},
+	}
+}
+
 // fetchOpenAIModels calls an OpenAI-compatible /models endpoint.
 func fetchOpenAIModels(ctx context.Context, apiBase, apiKey string) ([]ModelInfo, error) {
 	req, err := http.NewRequestWithContext(ctx, "GET", apiBase+"/models", nil)
@@ -180,3 +198,4 @@ func fetchOpenAIModels(ctx context.Context, apiBase, apiKey string) ([]ModelInfo
 	}
 	return models, nil
 }
+

internal/http/providers.go

@@ -67,6 +67,14 @@ func (h *ProvidersHandler) registerInMemory(p *store.LLMProviderData) {
 	if p.ProviderType == store.ProviderAnthropicNative {
 		h.providerReg.Register(providers.NewAnthropicProvider(p.APIKey,
 			providers.WithAnthropicBaseURL(p.APIBase)))
+	} else if p.ProviderType == store.ProviderDashScope {
+		h.providerReg.Register(providers.NewDashScopeProvider(p.APIKey, p.APIBase, ""))
+	} else if p.ProviderType == store.ProviderBailian {
+		base := p.APIBase
+		if base == "" {
+			base = "https://coding-intl.dashscope.aliyuncs.com/v1"
+		}
+		h.providerReg.Register(providers.NewOpenAIProvider(p.Name, p.APIKey, base, "qwen3.5-plus"))
 	} else {
 		prov := providers.NewOpenAIProvider(p.Name, p.APIKey, p.APIBase, "")
 		if p.ProviderType == store.ProviderMiniMax {

internal/providers/dashscope.go

@@ -0,0 +1,56 @@
+package providers
+
+import (
+	"context"
+	"log/slog"
+)
+
+const (
+	dashscopeDefaultBase  = "https://dashscope-intl.aliyuncs.com/compatible-mode/v1"
+	dashscopeDefaultModel = "qwen3-max"
+)
+
+// DashScopeProvider wraps OpenAIProvider to handle DashScope-specific behaviors.
+// Critical: DashScope does NOT support tools + streaming simultaneously.
+// When tools are present, ChatStream falls back to non-streaming Chat().
+type DashScopeProvider struct {
+	*OpenAIProvider
+}
+
+func NewDashScopeProvider(apiKey, apiBase, defaultModel string) *DashScopeProvider {
+	if apiBase == "" {
+		apiBase = dashscopeDefaultBase
+	}
+	if defaultModel == "" {
+		defaultModel = dashscopeDefaultModel
+	}
+	return &DashScopeProvider{
+		OpenAIProvider: NewOpenAIProvider("dashscope", apiKey, apiBase, defaultModel),
+	}
+}
+
+func (p *DashScopeProvider) Name() string { return "dashscope" }
+
+// ChatStream handles DashScope's limitation: tools + streaming cannot coexist.
+// When tools are present, falls back to non-streaming Chat() and synthesizes
+// chunk callbacks for the caller.
+func (p *DashScopeProvider) ChatStream(ctx context.Context, req ChatRequest, onChunk func(StreamChunk)) (*ChatResponse, error) {
+	if len(req.Tools) > 0 {
+		slog.Debug("dashscope: tools present, falling back to non-streaming Chat")
+		resp, err := p.Chat(ctx, req)
+		if err != nil {
+			return nil, err
+		}
+		if onChunk != nil {
+			if resp.Thinking != "" {
+				onChunk(StreamChunk{Thinking: resp.Thinking})
+			}
+			if resp.Content != "" {
+				onChunk(StreamChunk{Content: resp.Content})
+			}
+			onChunk(StreamChunk{Done: true})
+		}
+		return resp, nil
+	}
+	return p.OpenAIProvider.ChatStream(ctx, req, onChunk)
+}

internal/providers/openai.go

@@ -122,6 +122,12 @@ func (p *OpenAIProvider) ChatStream(ctx context.Context, req ChatRequest, onChun
 		}
 
 		delta := chunk.Choices[0].Delta
+		if delta.ReasoningContent != "" {
+			result.Thinking += delta.ReasoningContent
+			if onChunk != nil {
+				onChunk(StreamChunk{Thinking: delta.ReasoningContent})
+			}
+		}
 		if delta.Content != "" {
 			result.Content += delta.Content
 			if onChunk != nil {
@@ -326,6 +332,7 @@ func (p *OpenAIProvider) parseResponse(resp *openAIResponse) *ChatResponse {
 	if len(resp.Choices) > 0 {
 		msg := resp.Choices[0].Message
 		result.Content = msg.Content
+		result.Thinking = msg.ReasoningContent
 		result.FinishReason = resp.Choices[0].FinishReason
 
 		for _, tc := range msg.ToolCalls {

internal/providers/openai_types.go

@@ -13,9 +13,10 @@ type openAIChoice struct {
 }
 
 type openAIMessage struct {
-	Role      string           `json:"role"`
-	Content   string           `json:"content"`
-	ToolCalls []openAIToolCall `json:"tool_calls,omitempty"`
+	Role             string           `json:"role"`
+	Content          string           `json:"content"`
+	ReasoningContent string           `json:"reasoning_content,omitempty"`
+	ToolCalls        []openAIToolCall `json:"tool_calls,omitempty"`
 }
 
 type openAIToolCall struct {
@@ -54,8 +55,9 @@ type openAIStreamChoice struct {
 }
 
 type openAIStreamDelta struct {
-	Content   string                 `json:"content,omitempty"`
-	ToolCalls []openAIStreamToolCall `json:"tool_calls,omitempty"`
+	Content          string                 `json:"content,omitempty"`
+	ReasoningContent string                 `json:"reasoning_content,omitempty"`
+	ToolCalls        []openAIStreamToolCall `json:"tool_calls,omitempty"`
 }
 
 type openAIStreamToolCall struct {

internal/providers/types.go

@@ -30,6 +30,7 @@ type ChatRequest struct {
 // ChatResponse is the result from an LLM call.
 type ChatResponse struct {
 	Content      string     `json:"content"`
+	Thinking     string     `json:"thinking,omitempty"`
 	ToolCalls    []ToolCall `json:"tool_calls,omitempty"`
 	FinishReason string     `json:"finish_reason"` // "stop", "tool_calls", "length"
 	Usage        *Usage     `json:"usage,omitempty"`

internal/store/provider_store.go

@@ -19,6 +19,8 @@ const (
 	ProviderMiniMax         = "minimax_native"
 	ProviderCohere          = "cohere"
 	ProviderPerplexity      = "perplexity"
+	ProviderDashScope       = "dashscope"
+	ProviderBailian         = "bailian"
 )
 
 // ValidProviderTypes lists all accepted provider_type values.
@@ -34,6 +36,8 @@ var ValidProviderTypes = map[string]bool{
 	ProviderMiniMax:         true,
 	ProviderCohere:          true,
 	ProviderPerplexity:      true,
+	ProviderDashScope:       true,
+	ProviderBailian:         true,
 }
 
 // LLMProviderData represents an LLM provider configuration.

internal/tools/read_image.go

@@ -27,12 +27,13 @@ func MediaImagesFromCtx(ctx context.Context) []providers.ImageContent {
 // --- ReadImageTool ---
 
 // visionProviderPriority is the order in which providers are tried for vision.
-var visionProviderPriority = []string{"openrouter", "gemini", "anthropic"}
+var visionProviderPriority = []string{"openrouter", "gemini", "anthropic", "dashscope"}
 
 // visionModelOverrides maps provider names to preferred vision models.
 // Providers not listed here use their default model.
 var visionModelOverrides = map[string]string{
 	"openrouter": "google/gemini-2.5-flash-image",
+	"dashscope":  "qwen3-vl",
 }
 
 // ReadImageTool uses a vision-capable provider to describe images attached to the current message.

ui/web/src/pages/providers/provider-form-dialog.tsx

@@ -40,6 +40,8 @@ const PROVIDER_TYPES = [
   { value: "minimax_native", label: "MiniMax (Native)", apiBase: "https://api.minimax.io/v1", placeholder: "" },
   { value: "cohere", label: "Cohere", apiBase: "https://api.cohere.ai/compatibility/v1", placeholder: "" },
   { value: "perplexity", label: "Perplexity", apiBase: "https://api.perplexity.ai", placeholder: "" },
+  { value: "dashscope", label: "DashScope (Qwen)", apiBase: "https://dashscope-intl.aliyuncs.com/compatible-mode/v1", placeholder: "" },
+  { value: "bailian", label: "Bailian Coding", apiBase: "https://coding-intl.dashscope.aliyuncs.com/v1", placeholder: "" },
 ];
 
 export function ProviderFormDialog({ open, onOpenChange, provider, onSubmit }: ProviderFormDialogProps) {
@@ -139,7 +141,7 @@ export function ProviderFormDialog({ open, onOpenChange, provider, onSubmit }: P
                 setProviderType(v);
                 if (!isEdit) {
                   const preset = PROVIDER_TYPES.find((t) => t.value === v);
-                  if (preset?.apiBase) setApiBase(preset.apiBase);
+                  setApiBase(preset?.apiBase || "");
                 }
               }}
             >