@arikusi/deepseek-mcp-server 1.7.0 → 2.0.0

package/CHANGELOG.md

@@ -7,6 +7,44 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+## [2.0.0] - 2026-06-22
+
+### Changed
+- **DeepSeek V4 migration.** `deepseek-v4-flash` (new default) and `deepseek-v4-pro` are now the primary models, both with 1M context and up to 384K output tokens. `deepseek-chat` and `deepseek-reasoner` are kept as backward-compatible aliases that resolve to `deepseek-v4-flash` (chat maps to non-thinking, reasoner to thinking). The DeepSeek API retires those two names on 2026-07-24, so the server translates them before sending the request.
+- **Default model is now `deepseek-v4-flash`** (was `deepseek-chat`). Direct v4 calls default to non-thinking for fast responses; enable reasoning with `thinking: {type: "enabled"}` or the `deepseek-reasoner` alias. The V4 API defaults thinking to enabled, so the server now always sends an explicit thinking flag.
+- V4 pricing per 1M tokens: v4-flash `$0.0028` cache hit / `$0.14` cache miss / `$0.28` output; v4-pro `$0.003625` / `$0.435` / `$0.87`.
+- `max_tokens` upper bound raised to 384000.
+- Model fallback now pairs `deepseek-v4-flash` with `deepseek-v4-pro`.
+- The hosted worker endpoint (`deepseek-mcp.tahirl.com`) was migrated to V4 as well.
+
+### Added
+- `reasoning_effort` parameter (`high` / `max`) for thinking mode.
+
+### Fixed
+- Declared `cost_usd` and `routed_from` in the `deepseek_chat` output schema. Strict MCP clients rejected every response under the SDK 1.29 structured-content validation because these fields were returned but not declared.
+
+### Removed
+- Stopped sending `frequency_penalty` / `presence_penalty`, which the V4 API deprecated and ignores.
+
+## [1.8.0] - 2026-06-14
+
+### Security
+- **Missing authentication on the self-hosted HTTP endpoint.** In HTTP transport mode the server holds your `DEEPSEEK_API_KEY` and uses it for every `deepseek_chat` call, yet `POST /mcp` had no authentication and the server bound to `0.0.0.0`, so any client that could reach the port could initialize a session, enumerate tools, and invoke them. The defaults now bind to loopback and an optional bearer token guards the endpoint. Reported independently; advisory and CVE coordination in progress.
+
+### Changed
+- HTTP transport now binds to `127.0.0.1` by default (configurable via `HTTP_HOST`). The SDK's DNS rebinding protection is active on loopback. Binding to `0.0.0.0` without a token prints a startup security warning.
+- `docker-compose.yml` publishes the port to `127.0.0.1` only, and the README's `docker run` example does the same.
+- **Minimum Node.js is now 20.** Node 18 reached end of life in April 2025 and the test toolchain (vitest 4) no longer runs on it. The published package follows suit (`engines.node` is `>=20.0.0`); CI tests on Node 20, 22, and 24.
+
+### Added
+- `HTTP_AUTH_TOKEN`: when set, `POST/GET/DELETE /mcp` require `Authorization: Bearer <token>` (constant-time comparison). `/health` stays open for probes.
+- `HTTP_ALLOWED_HOSTS`: comma-separated allowed `Host` headers, keeping DNS rebinding protection when binding to `0.0.0.0`.
+- `SECURITY.md` with the disclosure policy and self-hosted HTTP hardening guidance.
+- Auth and host-binding tests (`src/transport-auth.test.ts`).
+
+### Fixed
+- Bumped `@modelcontextprotocol/sdk` to 1.29.0 and `vitest`/`@vitest/coverage-v8` to 4.1.8, clearing all transitive `npm audit` advisories (13 to 0).
+
 ## [1.7.0] - 2026-04-22
 
 ### Security

package/README.md

@@ -5,7 +5,7 @@
 <h1 align="center">DeepSeek MCP Server</h1>
 
 <p align="center">
-  MCP server for DeepSeek AI with chat, reasoning, multi-turn sessions, function calling, thinking mode, and cost tracking.
+  MCP server for DeepSeek V4 (v4-flash and v4-pro, 1M context) with multi-turn sessions, function calling, thinking mode, and cost tracking.
 </p>
 
 <p align="center">
@@ -13,7 +13,8 @@
   <a href="https://www.npmjs.com/package/@arikusi/deepseek-mcp-server"><img src="https://img.shields.io/npm/dm/@arikusi/deepseek-mcp-server.svg" alt="npm downloads" /></a>
   <a href="https://opensource.org/licenses/MIT"><img src="https://img.shields.io/badge/License-MIT-yellow.svg" alt="License: MIT" /></a>
   <a href="https://nodejs.org/"><img src="https://img.shields.io/node/v/@arikusi/deepseek-mcp-server.svg" alt="Node.js Version" /></a>
-  <a href="https://www.typescriptlang.org/"><img src="https://img.shields.io/badge/TypeScript-5.7-blue.svg" alt="TypeScript" /></a>
+  <a href="https://www.typescriptlang.org/"><img src="https://img.shields.io/badge/TypeScript-6.0-blue.svg" alt="TypeScript" /></a>
+  <a href="https://api-docs.deepseek.com"><img src="https://img.shields.io/badge/DeepSeek-V4-7c3aed.svg" alt="DeepSeek V4" /></a>
   <a href="https://github.com/arikusi/deepseek-mcp-server/actions"><img src="https://github.com/arikusi/deepseek-mcp-server/workflows/CI/badge.svg" alt="Build Status" /></a>
 </p>
 
@@ -34,6 +35,8 @@
   </a>
 </p>
 
+> **v2.0.0 runs on DeepSeek V4.** Two models, `deepseek-v4-flash` (fast and economical) and `deepseek-v4-pro` (top capability), both with a 1M-token context window and optional chain-of-thought thinking. Existing `deepseek-chat` and `deepseek-reasoner` setups keep working as aliases, so upgrading is drop-in.
+
 ## Quick Start
 
 ### Remote (No Install)
@@ -84,7 +87,7 @@ gemini mcp add deepseek npx @arikusi/deepseek-mcp-server -e DEEPSEEK_API_KEY=you
 
 ## Features
 
-- **DeepSeek V3.2**: Both models now run DeepSeek-V3.2 (since Sept 2025)
+- **DeepSeek V4**: `deepseek-v4-flash` and `deepseek-v4-pro`, both with 1M context and optional chain-of-thought thinking mode
 - **Multi-Turn Sessions**: Conversation context preserved across requests via `session_id` parameter
 - **Model Fallback & Circuit Breaker**: Automatic fallback between models with circuit breaker protection against cascading failures
 - **MCP Resources**: `deepseek://models`, `deepseek://config`, `deepseek://usage` — query model info, config, and usage stats
@@ -100,7 +103,7 @@ gemini mcp add deepseek npx @arikusi/deepseek-mcp-server -e DEEPSEEK_API_KEY=you
 - **Remote Endpoint**: Hosted at `deepseek-mcp.tahirl.com/mcp` — BYOK (Bring Your Own Key), no install needed
 - **HTTP Transport**: Self-hosted remote access via Streamable HTTP with `TRANSPORT=http`
 - **Docker Ready**: Multi-stage Dockerfile with health checks for containerized deployment
-- **Tested**: 265 tests, ~89% line coverage
+- **Tested**: 280 tests, ~89% line coverage
 - **Type-Safe**: Full TypeScript implementation
 - **MCP Compatible**: Works with any MCP-compatible CLI (Claude Code, Gemini CLI, etc.)
 
@@ -108,7 +111,7 @@ gemini mcp add deepseek npx @arikusi/deepseek-mcp-server -e DEEPSEEK_API_KEY=you
 
 ### Prerequisites
 
-- Node.js 18+
+- Node.js 20+
 - A DeepSeek API key (get one at [https://platform.deepseek.com](https://platform.deepseek.com))
 
 ### Manual Installation
@@ -186,13 +189,14 @@ Chat with DeepSeek AI models with automatic cost tracking and function calling s
   - `role`: "system" | "user" | "assistant" | "tool"
   - `content`: Message text
   - `tool_call_id` (optional): Required for tool role messages
-- `model` (optional): "deepseek-chat" (default) or "deepseek-reasoner"
+- `model` (optional): "deepseek-v4-flash" (default) or "deepseek-v4-pro". "deepseek-chat" and "deepseek-reasoner" are accepted as aliases that resolve to v4-flash (non-thinking / thinking).
 - `temperature` (optional): 0-2, controls randomness (default: 1.0). Ignored when thinking mode is enabled.
-- `max_tokens` (optional): Maximum tokens to generate (deepseek-chat: max 8192, deepseek-reasoner: max 65536)
+- `max_tokens` (optional): Maximum tokens to generate (V4 models support up to 384000)
 - `stream` (optional): Enable streaming mode (default: false)
 - `tools` (optional): Array of tool definitions for function calling (max 128)
 - `tool_choice` (optional): "auto" | "none" | "required" | `{type: "function", function: {name: "..."}}`
-- `thinking` (optional): Enable thinking mode `{type: "enabled"}`
+- `thinking` (optional): Toggle thinking mode, `{type: "enabled"}` to reason or `{type: "disabled"}` for a fast answer (non-thinking is the default)
+- `reasoning_effort` (optional): "high" (default) or "max", applies only while thinking mode is active
 - `json_mode` (optional): Enable JSON output mode (supported by both models)
 - `session_id` (optional): Session ID for multi-turn conversations. Previous context is automatically prepended.
 
@@ -212,13 +216,13 @@ Chat with DeepSeek AI models with automatic cost tracking and function calling s
       "content": "Explain the theory of relativity in simple terms"
     }
   ],
-  "model": "deepseek-chat",
+  "model": "deepseek-v4-flash",
   "temperature": 0.7,
   "max_tokens": 1000
 }
 ```
 
-**DeepSeek Reasoner Example:**
+**Reasoning Example (`deepseek-reasoner` alias, routes to v4-flash + thinking):**
 
 ```json
 {
@@ -232,7 +236,22 @@ Chat with DeepSeek AI models with automatic cost tracking and function calling s
 }
 ```
 
-The reasoner model will show its thinking process in `<thinking>` tags followed by the final answer.
+Thinking mode returns the chain-of-thought in `<thinking>` tags followed by the final answer.
+
+**DeepSeek V4 Pro Example (hardest tasks):**
+
+```json
+{
+  "messages": [
+    {
+      "role": "user",
+      "content": "Prove that the square root of 2 is irrational."
+    }
+  ],
+  "model": "deepseek-v4-pro",
+  "thinking": { "type": "enabled" }
+}
+```
 
 **Function Calling Example:**
 
@@ -279,12 +298,12 @@ When the model decides to call a function, the response includes `tool_calls` wi
       "content": "Analyze the time complexity of quicksort"
     }
   ],
-  "model": "deepseek-chat",
+  "model": "deepseek-v4-flash",
   "thinking": { "type": "enabled" }
 }
 ```
 
-When thinking mode is enabled, `temperature`, `top_p`, `frequency_penalty`, and `presence_penalty` are automatically ignored.
+When thinking mode is enabled, `temperature` and `top_p` are automatically ignored.
 
 **JSON Output Mode Example:**
 
@@ -296,12 +315,12 @@ When thinking mode is enabled, `temperature`, `top_p`, `frequency_penalty`, and
       "content": "Return a json object with name, age, and city fields for a sample user"
     }
   ],
-  "model": "deepseek-chat",
+  "model": "deepseek-v4-flash",
   "json_mode": true
 }
 ```
 
-JSON mode ensures the model outputs valid JSON. Include the word "json" in your prompt for best results. Supported by both `deepseek-chat` and `deepseek-reasoner`.
+JSON mode ensures the model outputs valid JSON. Include the word "json" in your prompt for best results. Supported by all models.
 
 **Multi-Turn Session Example:**
 
@@ -383,28 +402,27 @@ Each prompt is optimized for the DeepSeek Reasoner model to provide detailed rea
 
 ## Models
 
-Both models run **DeepSeek-V3.2** with unified pricing.
+Both V4 models have a 1M-token context window, up to 384K output tokens, and support function calling, JSON mode, and optional chain-of-thought thinking. They are non-thinking by default here for fast responses; enable reasoning with `thinking: {type: "enabled"}` (or the `deepseek-reasoner` alias).
+
+### deepseek-v4-flash (default)
+
+- **Best for**: General conversations, coding, content generation, agent loops
+- **Speed**: Fast and economical
+- **Context**: 1M tokens
+- **Max Output**: 384K tokens
+- **Pricing**: $0.0028/1M cache hit, $0.14/1M cache miss, $0.28/1M output
 
-### deepseek-chat
+### deepseek-v4-pro
 
-- **Best for**: General conversations, coding, content generation
-- **Speed**: Fast
-- **Context**: 128K tokens
-- **Max Output**: 8K tokens (default 4K)
-- **Mode**: Non-thinking (can enable thinking via parameter)
-- **Features**: Thinking mode, JSON mode, function calling, FIM completion
-- **Pricing**: $0.028/1M cache hit, $0.28/1M cache miss, $0.42/1M output
+- **Best for**: Complex reasoning, math, hard multi-step tasks, top-quality output
+- **Speed**: Slower than flash, highest capability
+- **Context**: 1M tokens
+- **Max Output**: 384K tokens
+- **Pricing**: $0.003625/1M cache hit, $0.435/1M cache miss, $0.87/1M output
 
-### deepseek-reasoner
+### Compatibility aliases
 
-- **Best for**: Complex reasoning, math, logic problems, multi-step tasks
-- **Speed**: Slower (shows thinking process)
-- **Context**: 128K tokens
-- **Max Output**: 64K tokens (default 32K)
-- **Mode**: Thinking (always active, chain-of-thought reasoning)
-- **Features**: JSON mode, function calling
-- **Output**: Both reasoning process and final answer
-- **Pricing**: $0.028/1M cache hit, $0.28/1M cache miss, $0.42/1M output
+`deepseek-chat` and `deepseek-reasoner` are still accepted and resolve to `deepseek-v4-flash` (chat = non-thinking, reasoner = thinking), so existing configs keep working. The DeepSeek API retires those two names on **2026-07-24**; this server translates them to V4 for you.
 
 ## Configuration
 
@@ -414,7 +432,7 @@ The server is configured via environment variables. All settings except `DEEPSEE
 |----------|---------|-------------|
 | `DEEPSEEK_API_KEY` | (required) | Your DeepSeek API key |
 | `DEEPSEEK_BASE_URL` | `https://api.deepseek.com` | Custom API endpoint |
-| `DEFAULT_MODEL` | `deepseek-chat` | Default model for requests |
+| `DEFAULT_MODEL` | `deepseek-v4-flash` | Default model for requests |
 | `SHOW_COST_INFO` | `true` | Show cost info in responses |
 | `REQUEST_TIMEOUT` | `60000` | Request timeout in milliseconds |
 | `MAX_RETRIES` | `2` | Maximum retry count for failed requests |
@@ -429,6 +447,9 @@ The server is configured via environment variables. All settings except `DEEPSEE
 | `ENABLE_MULTIMODAL` | `false` | Enable multimodal (image) input support |
 | `TRANSPORT` | `stdio` | Transport mode: `stdio` or `http` |
 | `HTTP_PORT` | `3000` | HTTP server port (when TRANSPORT=http) |
+| `HTTP_HOST` | `127.0.0.1` | Bind address for HTTP transport. Loopback by default so a fresh run is not exposed. Set to `0.0.0.0` to accept remote connections (do this only with auth or a proxy in front) |
+| `HTTP_AUTH_TOKEN` | _(unset)_ | When set, `POST /mcp` requires `Authorization: Bearer <token>`. `/health` stays open. Strongly recommended whenever the port is reachable beyond localhost |
+| `HTTP_ALLOWED_HOSTS` | _(unset)_ | Comma-separated list of allowed `Host` headers for DNS rebinding protection when binding to `0.0.0.0` (e.g. `mcp.example.com,localhost`) |
 
 **Example with custom config:**
 ```bash
@@ -559,10 +580,39 @@ curl http://localhost:3000/health
 
 The MCP endpoint is available at `POST /mcp` (Streamable HTTP protocol).
 
+**Securing the endpoint (read before exposing it).** In self-hosted HTTP mode the
+server holds your `DEEPSEEK_API_KEY` and uses it for every `deepseek_chat` call.
+Anyone who can reach `POST /mcp` can invoke tools and spend that key, so the
+endpoint must not sit open on a public interface. The defaults are built around
+this:
+
+1. `HTTP_HOST` defaults to `127.0.0.1`, so a plain run only listens on loopback and the SDK's DNS rebinding protection is active. Nothing off the machine can reach it.
+2. To accept remote connections, set `HTTP_HOST=0.0.0.0`, but then set `HTTP_AUTH_TOKEN` as well so `/mcp` requires `Authorization: Bearer <token>`. If you bind to `0.0.0.0` without a token, the server prints a loud warning on startup.
+3. For an internet-facing deployment, put an authenticating reverse proxy with TLS in front and set `HTTP_ALLOWED_HOSTS` to your real hostname(s).
+
+```bash
+# Exposed deployment with a bearer token
+TRANSPORT=http HTTP_HOST=0.0.0.0 HTTP_PORT=3000 \
+  HTTP_AUTH_TOKEN=$(openssl rand -hex 32) \
+  HTTP_ALLOWED_HOSTS=mcp.example.com \
+  DEEPSEEK_API_KEY=your-key node dist/index.js
+
+# Calling it
+curl -X POST http://mcp.example.com:3000/mcp \
+  -H "Authorization: Bearer YOUR_TOKEN" \
+  -H "Content-Type: application/json" \
+  -H "Accept: application/json, text/event-stream" \
+  -d '{"jsonrpc":"2.0","method":"initialize","params":{"capabilities":{}},"id":1}'
+```
+
+`HTTP_AUTH_TOKEN` is a static gateway token for the self-hosted endpoint and is
+unrelated to your DeepSeek key. It is separate from the hosted BYOK endpoint
+above, where clients pass their own DeepSeek key as the bearer.
+
 **Session isolation (1.7.0+):** In HTTP transport each connected MCP session
 gets its own `McpServer` instance and its own `SessionStore`. Conversation
 history, session listings, and deletions are scoped to the MCP session that
-created them — one client cannot read, enumerate, or wipe another client's
+created them, so one client cannot read, enumerate, or wipe another client's
 sessions. STDIO transport is single-tenant by nature and unaffected.
 
 ### Docker
@@ -571,14 +621,21 @@ sessions. STDIO transport is single-tenant by nature and unaffected.
 # Build
 docker build -t deepseek-mcp-server .
 
-# Run
-docker run -d -p 3000:3000 -e DEEPSEEK_API_KEY=your-key deepseek-mcp-server
+# Run, reachable only from the host's loopback, with a bearer token
+docker run -d -p 127.0.0.1:3000:3000 \
+  -e DEEPSEEK_API_KEY=your-key \
+  -e HTTP_AUTH_TOKEN=your-token \
+  deepseek-mcp-server
 
 # Or use docker-compose
-DEEPSEEK_API_KEY=your-key docker compose up -d
+DEEPSEEK_API_KEY=your-key HTTP_AUTH_TOKEN=your-token docker compose up -d
 ```
 
-The Docker image defaults to HTTP transport on port 3000 with a built-in health check.
+The image runs HTTP transport on port 3000 with a health check. Inside the
+container it binds `0.0.0.0` (required for the port mapping to work), so control
+exposure at the publish layer: the example above and the bundled
+`docker-compose.yml` publish to `127.0.0.1` only. If you publish the port on a
+public interface, set `HTTP_AUTH_TOKEN`.
 
 ## Troubleshooting
 

package/dist/config.d.ts

@@ -24,6 +24,9 @@ declare const ConfigSchema: z.ZodObject<{
         http: "http";
     }>>;
     httpPort: z.ZodDefault<z.ZodNumber>;
+    httpHost: z.ZodDefault<z.ZodString>;
+    httpAuthToken: z.ZodOptional<z.ZodString>;
+    httpAllowedHosts: z.ZodOptional<z.ZodArray<z.ZodString>>;
 }, z.core.$strip>;
 export type Config = z.infer<typeof ConfigSchema>;
 /**

package/dist/config.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"config.d.ts","sourceRoot":"","sources":["../src/config.ts"],"names":[],"mappings":"AAAA;;;GAGG;AAEH,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAC;AAGxB,QAAA,MAAM,YAAY;;;;;;;;;;;;;;;;;;;;;iBAkBhB,CAAC;AAEH,MAAM,MAAM,MAAM,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,YAAY,CAAC,CAAC;AAIlD;;;;GAIG;AACH,wBAAgB,UAAU,IAAI,MAAM,CAyDnC;AAED;;;GAGG;AACH,wBAAgB,SAAS,IAAI,MAAM,CAKlC;AAED;;GAEG;AACH,wBAAgB,WAAW,IAAI,IAAI,CAElC"}
+{"version":3,"file":"config.d.ts","sourceRoot":"","sources":["../src/config.ts"],"names":[],"mappings":"AAAA;;;GAGG;AAEH,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAC;AAGxB,QAAA,MAAM,YAAY;;;;;;;;;;;;;;;;;;;;;;;;iBAqBhB,CAAC;AAEH,MAAM,MAAM,MAAM,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,YAAY,CAAC,CAAC;AAIlD;;;;GAIG;AACH,wBAAgB,UAAU,IAAI,MAAM,CA8DnC;AAED;;;GAGG;AACH,wBAAgB,SAAS,IAAI,MAAM,CAKlC;AAED;;GAEG;AACH,wBAAgB,WAAW,IAAI,IAAI,CAElC"}

package/dist/config.js

@@ -15,13 +15,16 @@ const ConfigSchema = z.object({
     sessionTtlMinutes: z.number().positive().default(30),
     maxSessions: z.number().positive().default(100),
     fallbackEnabled: z.boolean().default(true),
-    defaultModel: z.string().default('deepseek-chat'),
+    defaultModel: z.string().default('deepseek-v4-flash'),
     circuitBreakerThreshold: z.number().positive().default(5),
     circuitBreakerResetTimeout: z.number().positive().default(30000),
     maxSessionMessages: z.number().positive().default(200),
     enableMultimodal: z.boolean().default(false),
     transport: z.enum(['stdio', 'http']).default('stdio'),
     httpPort: z.number().positive().default(3000),
+    httpHost: z.string().min(1).default('127.0.0.1'),
+    httpAuthToken: z.string().min(1).optional(),
+    httpAllowedHosts: z.array(z.string().min(1)).optional(),
 });
 let cachedConfig = null;
 /**
@@ -51,7 +54,7 @@ export function loadConfig() {
             ? parseInt(process.env.MAX_SESSIONS, 10)
             : 100,
         fallbackEnabled: process.env.FALLBACK_ENABLED !== 'false',
-        defaultModel: process.env.DEFAULT_MODEL || 'deepseek-chat',
+        defaultModel: process.env.DEFAULT_MODEL || 'deepseek-v4-flash',
         circuitBreakerThreshold: process.env.CIRCUIT_BREAKER_THRESHOLD
             ? parseInt(process.env.CIRCUIT_BREAKER_THRESHOLD, 10)
             : 5,
@@ -64,6 +67,11 @@ export function loadConfig() {
         enableMultimodal: process.env.ENABLE_MULTIMODAL === 'true',
         transport: (process.env.TRANSPORT || 'stdio'),
         httpPort: process.env.HTTP_PORT ? parseInt(process.env.HTTP_PORT, 10) : 3000,
+        httpHost: process.env.HTTP_HOST || '127.0.0.1',
+        httpAuthToken: process.env.HTTP_AUTH_TOKEN || undefined,
+        httpAllowedHosts: process.env.HTTP_ALLOWED_HOSTS
+            ? process.env.HTTP_ALLOWED_HOSTS.split(',').map((h) => h.trim()).filter(Boolean)
+            : undefined,
     };
     const result = ConfigSchema.safeParse(raw);
     if (!result.success) {

package/dist/config.js.map

@@ -1 +1 @@
-{"version":3,"file":"config.js","sourceRoot":"","sources":["../src/config.ts"],"names":[],"mappings":"AAAA;;;GAGG;AAEH,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAC;AACxB,OAAO,EAAE,WAAW,EAAE,MAAM,aAAa,CAAC;AAE1C,MAAM,YAAY,GAAG,CAAC,CAAC,MAAM,CAAC;IAC5B,MAAM,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,GAAG,CAAC,CAAC,EAAE,8BAA8B,CAAC;IACzD,OAAO,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,GAAG,EAAE,CAAC,OAAO,CAAC,0BAA0B,CAAC;IAC7D,YAAY,EAAE,CAAC,CAAC,OAAO,EAAE,CAAC,OAAO,CAAC,IAAI,CAAC;IACvC,cAAc,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE,CAAC,OAAO,CAAC,KAAK,CAAC;IACpD,UAAU,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC;IAChD,kBAAkB,EAAE,CAAC,CAAC,OAAO,EAAE,CAAC,OAAO,CAAC,KAAK,CAAC;IAC9C,gBAAgB,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE,CAAC,OAAO,CAAC,OAAO,CAAC;IACxD,iBAAiB,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE,CAAC,OAAO,CAAC,EAAE,CAAC;IACpD,WAAW,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE,CAAC,OAAO,CAAC,GAAG,CAAC;IAC/C,eAAe,EAAE,CAAC,CAAC,OAAO,EAAE,CAAC,OAAO,CAAC,IAAI,CAAC;IAC1C,YAAY,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,OAAO,CAAC,eAAe,CAAC;IACjD,uBAAuB,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE,CAAC,OAAO,CAAC,CAAC,CAAC;IACzD,0BAA0B,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE,CAAC,OAAO,CAAC,KAAK,CAAC;IAChE,kBAAkB,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE,CAAC,OAAO,CAAC,GAAG,CAAC;IACtD,gBAAgB,EAAE,CAAC,CAAC,OAAO,EAAE,CAAC,OAAO,CAAC,KAAK,CAAC;IAC5C,SAAS,EAAE,CAAC,CAAC,IAAI,CAAC,CAAC,OAAO,EAAE,MAAM,CAAC,CAAC,CAAC,OAAO,CAAC,OAAO,CAAC;IACrD,QAAQ,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE,CAAC,OAAO,CAAC,IAAI,CAAC;CAC9C,CAAC,CAAC;AAIH,IAAI,YAAY,GAAkB,IAAI,CAAC;AAEvC;;;;GAIG;AACH,MAAM,UAAU,UAAU;IACxB,MAAM,GAAG,GAAG;QACV,MAAM,EAAE,OAAO,CAAC,GAAG,CAAC,gBAAgB,IAAI,EAAE;QAC1C,OAAO,EAAE,OAAO,CAAC,GAAG,CAAC,iBAAiB,IAAI,0BAA0B;QACpE,YAAY,EAAE,OAAO,CAAC,GAAG,CAAC,cAAc,KAAK,OAAO;QACpD,cAAc,EAAE,OAAO,CAAC,GAAG,CAAC,eAAe;YACzC,CAAC,CAAC,QAAQ,CAAC,OAAO,CAAC,GAAG,CAAC,eAAe,EAAE,EAAE,CAAC;YAC3C,CAAC,CAAC,KAAK;QACT,UAAU,EAAE,OAAO,CAAC,GAAG,CAAC,WAAW;YACjC,CAAC,CAAC,QAAQ,CAAC,OAAO,CAAC,GAAG,CAAC,WAAW,EAAE,EAAE,CAAC;YACvC,CAAC,CAAC,CAAC;QACL,kBAAkB,EAAE,OAAO,CAAC,GAAG,CAAC,oBAAoB,KAAK,MAAM;QAC/D,gBAAgB,EAAE,OAAO,CAAC,GAAG,CAAC,kBAAkB;YAC9C,CAAC,CAAC,QAAQ,CAAC,OAAO,CAAC,GAAG,CAAC,kBAAkB,EAAE,EAAE,CAAC;YAC9C,CAAC,CAAC,OAAO;QACX,iBAAiB,EAAE,OAAO,CAAC,GAAG,CAAC,mBAAmB;YAChD,CAAC,CAAC,QAAQ,CAAC,OAAO,CAAC,GAAG,CAAC,mBAAmB,EAAE,EAAE,CAAC;YAC/C,CAAC,CAAC,EAAE;QACN,WAAW,EAAE,OAAO,CAAC,GAAG,CAAC,YAAY;YACnC,CAAC,CAAC,QAAQ,CAAC,OAAO,CAAC,GAAG,CAAC,YAAY,EAAE,EAAE,CAAC;YACxC,CAAC,CAAC,GAAG;QACP,eAAe,EAAE,OAAO,CAAC,GAAG,CAAC,gBAAgB,KAAK,OAAO;QACzD,YAAY,EAAE,OAAO,CAAC,GAAG,CAAC,aAAa,IAAI,eAAe;QAC1D,uBAAuB,EAAE,OAAO,CAAC,GAAG,CAAC,yBAAyB;YAC5D,CAAC,CAAC,QAAQ,CAAC,OAAO,CAAC,GAAG,CAAC,yBAAyB,EAAE,EAAE,CAAC;YACrD,CAAC,CAAC,CAAC;QACL,0BAA0B,EAAE,OAAO,CAAC,GAAG,CAAC,6BAA6B;YACnE,CAAC,CAAC,QAAQ,CAAC,OAAO,CAAC,GAAG,CAAC,6BAA6B,EAAE,EAAE,CAAC;YACzD,CAAC,CAAC,KAAK;QACT,kBAAkB,EAAE,OAAO,CAAC,GAAG,CAAC,oBAAoB;YAClD,CAAC,CAAC,QAAQ,CAAC,OAAO,CAAC,GAAG,CAAC,oBAAoB,EAAE,EAAE,CAAC;YAChD,CAAC,CAAC,GAAG;QACP,gBAAgB,EAAE,OAAO,CAAC,GAAG,CAAC,iBAAiB,KAAK,MAAM;QAC1D,SAAS,EAAE,CAAC,OAAO,CAAC,GAAG,CAAC,SAAS,IAAI,OAAO,CAAqB;QACjE,QAAQ,EAAE,OAAO,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC,CAAC,QAAQ,CAAC,OAAO,CAAC,GAAG,CAAC,SAAS,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,IAAI;KAC7E,CAAC;IAEF,MAAM,MAAM,GAAG,YAAY,CAAC,SAAS,CAAC,GAAG,CAAC,CAAC;IAE3C,IAAI,CAAC,MAAM,CAAC,OAAO,EAAE,CAAC;QACpB,MAAM,MAAM,GAAG,MAAM,CAAC,KAAK,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC;YACjD,IAAI,EAAE,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC;YAC1B,OAAO,EAAE,KAAK,CAAC,OAAO;SACvB,CAAC,CAAC,CAAC;QAEJ,MAAM,IAAI,GAAG,CAAC,GAAG,CAAC,MAAM;YACtB,CAAC,CAAC,oFAAoF;YACtF,CAAC,CAAC,EAAE,CAAC;QAEP,MAAM,IAAI,WAAW,CACnB,kCAAkC,IAAI,EAAE,EACxC,MAAM,CACP,CAAC;IACJ,CAAC;IAED,YAAY,GAAG,MAAM,CAAC,IAAI,CAAC;IAC3B,OAAO,YAAY,CAAC;AACtB,CAAC;AAED;;;GAGG;AACH,MAAM,UAAU,SAAS;IACvB,IAAI,CAAC,YAAY,EAAE,CAAC;QAClB,MAAM,IAAI,KAAK,CAAC,6CAA6C,CAAC,CAAC;IACjE,CAAC;IACD,OAAO,YAAY,CAAC;AACtB,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,WAAW;IACzB,YAAY,GAAG,IAAI,CAAC;AACtB,CAAC"}
+{"version":3,"file":"config.js","sourceRoot":"","sources":["../src/config.ts"],"names":[],"mappings":"AAAA;;;GAGG;AAEH,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAC;AACxB,OAAO,EAAE,WAAW,EAAE,MAAM,aAAa,CAAC;AAE1C,MAAM,YAAY,GAAG,CAAC,CAAC,MAAM,CAAC;IAC5B,MAAM,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,GAAG,CAAC,CAAC,EAAE,8BAA8B,CAAC;IACzD,OAAO,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,GAAG,EAAE,CAAC,OAAO,CAAC,0BAA0B,CAAC;IAC7D,YAAY,EAAE,CAAC,CAAC,OAAO,EAAE,CAAC,OAAO,CAAC,IAAI,CAAC;IACvC,cAAc,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE,CAAC,OAAO,CAAC,KAAK,CAAC;IACpD,UAAU,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC;IAChD,kBAAkB,EAAE,CAAC,CAAC,OAAO,EAAE,CAAC,OAAO,CAAC,KAAK,CAAC;IAC9C,gBAAgB,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE,CAAC,OAAO,CAAC,OAAO,CAAC;IACxD,iBAAiB,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE,CAAC,OAAO,CAAC,EAAE,CAAC;IACpD,WAAW,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE,CAAC,OAAO,CAAC,GAAG,CAAC;IAC/C,eAAe,EAAE,CAAC,CAAC,OAAO,EAAE,CAAC,OAAO,CAAC,IAAI,CAAC;IAC1C,YAAY,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,OAAO,CAAC,mBAAmB,CAAC;IACrD,uBAAuB,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE,CAAC,OAAO,CAAC,CAAC,CAAC;IACzD,0BAA0B,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE,CAAC,OAAO,CAAC,KAAK,CAAC;IAChE,kBAAkB,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE,CAAC,OAAO,CAAC,GAAG,CAAC;IACtD,gBAAgB,EAAE,CAAC,CAAC,OAAO,EAAE,CAAC,OAAO,CAAC,KAAK,CAAC;IAC5C,SAAS,EAAE,CAAC,CAAC,IAAI,CAAC,CAAC,OAAO,EAAE,MAAM,CAAC,CAAC,CAAC,OAAO,CAAC,OAAO,CAAC;IACrD,QAAQ,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE,CAAC,OAAO,CAAC,IAAI,CAAC;IAC7C,QAAQ,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,WAAW,CAAC;IAChD,aAAa,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,QAAQ,EAAE;IAC3C,gBAAgB,EAAE,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,MAAM,EAAE,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,QAAQ,EAAE;CACxD,CAAC,CAAC;AAIH,IAAI,YAAY,GAAkB,IAAI,CAAC;AAEvC;;;;GAIG;AACH,MAAM,UAAU,UAAU;IACxB,MAAM,GAAG,GAAG;QACV,MAAM,EAAE,OAAO,CAAC,GAAG,CAAC,gBAAgB,IAAI,EAAE;QAC1C,OAAO,EAAE,OAAO,CAAC,GAAG,CAAC,iBAAiB,IAAI,0BAA0B;QACpE,YAAY,EAAE,OAAO,CAAC,GAAG,CAAC,cAAc,KAAK,OAAO;QACpD,cAAc,EAAE,OAAO,CAAC,GAAG,CAAC,eAAe;YACzC,CAAC,CAAC,QAAQ,CAAC,OAAO,CAAC,GAAG,CAAC,eAAe,EAAE,EAAE,CAAC;YAC3C,CAAC,CAAC,KAAK;QACT,UAAU,EAAE,OAAO,CAAC,GAAG,CAAC,WAAW;YACjC,CAAC,CAAC,QAAQ,CAAC,OAAO,CAAC,GAAG,CAAC,WAAW,EAAE,EAAE,CAAC;YACvC,CAAC,CAAC,CAAC;QACL,kBAAkB,EAAE,OAAO,CAAC,GAAG,CAAC,oBAAoB,KAAK,MAAM;QAC/D,gBAAgB,EAAE,OAAO,CAAC,GAAG,CAAC,kBAAkB;YAC9C,CAAC,CAAC,QAAQ,CAAC,OAAO,CAAC,GAAG,CAAC,kBAAkB,EAAE,EAAE,CAAC;YAC9C,CAAC,CAAC,OAAO;QACX,iBAAiB,EAAE,OAAO,CAAC,GAAG,CAAC,mBAAmB;YAChD,CAAC,CAAC,QAAQ,CAAC,OAAO,CAAC,GAAG,CAAC,mBAAmB,EAAE,EAAE,CAAC;YAC/C,CAAC,CAAC,EAAE;QACN,WAAW,EAAE,OAAO,CAAC,GAAG,CAAC,YAAY;YACnC,CAAC,CAAC,QAAQ,CAAC,OAAO,CAAC,GAAG,CAAC,YAAY,EAAE,EAAE,CAAC;YACxC,CAAC,CAAC,GAAG;QACP,eAAe,EAAE,OAAO,CAAC,GAAG,CAAC,gBAAgB,KAAK,OAAO;QACzD,YAAY,EAAE,OAAO,CAAC,GAAG,CAAC,aAAa,IAAI,mBAAmB;QAC9D,uBAAuB,EAAE,OAAO,CAAC,GAAG,CAAC,yBAAyB;YAC5D,CAAC,CAAC,QAAQ,CAAC,OAAO,CAAC,GAAG,CAAC,yBAAyB,EAAE,EAAE,CAAC;YACrD,CAAC,CAAC,CAAC;QACL,0BAA0B,EAAE,OAAO,CAAC,GAAG,CAAC,6BAA6B;YACnE,CAAC,CAAC,QAAQ,CAAC,OAAO,CAAC,GAAG,CAAC,6BAA6B,EAAE,EAAE,CAAC;YACzD,CAAC,CAAC,KAAK;QACT,kBAAkB,EAAE,OAAO,CAAC,GAAG,CAAC,oBAAoB;YAClD,CAAC,CAAC,QAAQ,CAAC,OAAO,CAAC,GAAG,CAAC,oBAAoB,EAAE,EAAE,CAAC;YAChD,CAAC,CAAC,GAAG;QACP,gBAAgB,EAAE,OAAO,CAAC,GAAG,CAAC,iBAAiB,KAAK,MAAM;QAC1D,SAAS,EAAE,CAAC,OAAO,CAAC,GAAG,CAAC,SAAS,IAAI,OAAO,CAAqB;QACjE,QAAQ,EAAE,OAAO,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC,CAAC,QAAQ,CAAC,OAAO,CAAC,GAAG,CAAC,SAAS,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,IAAI;QAC5E,QAAQ,EAAE,OAAO,CAAC,GAAG,CAAC,SAAS,IAAI,WAAW;QAC9C,aAAa,EAAE,OAAO,CAAC,GAAG,CAAC,eAAe,IAAI,SAAS;QACvD,gBAAgB,EAAE,OAAO,CAAC,GAAG,CAAC,kBAAkB;YAC9C,CAAC,CAAC,OAAO,CAAC,GAAG,CAAC,kBAAkB,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,MAAM,CAAC,OAAO,CAAC;YAChF,CAAC,CAAC,SAAS;KACd,CAAC;IAEF,MAAM,MAAM,GAAG,YAAY,CAAC,SAAS,CAAC,GAAG,CAAC,CAAC;IAE3C,IAAI,CAAC,MAAM,CAAC,OAAO,EAAE,CAAC;QACpB,MAAM,MAAM,GAAG,MAAM,CAAC,KAAK,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC;YACjD,IAAI,EAAE,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC;YAC1B,OAAO,EAAE,KAAK,CAAC,OAAO;SACvB,CAAC,CAAC,CAAC;QAEJ,MAAM,IAAI,GAAG,CAAC,GAAG,CAAC,MAAM;YACtB,CAAC,CAAC,oFAAoF;YACtF,CAAC,CAAC,EAAE,CAAC;QAEP,MAAM,IAAI,WAAW,CACnB,kCAAkC,IAAI,EAAE,EACxC,MAAM,CACP,CAAC;IACJ,CAAC;IAED,YAAY,GAAG,MAAM,CAAC,IAAI,CAAC;IAC3B,OAAO,YAAY,CAAC;AACtB,CAAC;AAED;;;GAGG;AACH,MAAM,UAAU,SAAS;IACvB,IAAI,CAAC,YAAY,EAAE,CAAC;QAClB,MAAM,IAAI,KAAK,CAAC,6CAA6C,CAAC,CAAC;IACjE,CAAC;IACD,OAAO,YAAY,CAAC;AACtB,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,WAAW;IACzB,YAAY,GAAG,IAAI,CAAC;AACtB,CAAC"}

package/dist/cost.d.ts

@@ -2,9 +2,11 @@
  * Cost Calculation Module
  * Handles pricing and cost formatting for DeepSeek API requests
  *
- * Model-aware pricing per 1M tokens (USD):
- * - deepseek-chat / deepseek-reasoner (V3.2): cache hit $0.028, cache miss $0.28, output $0.42
- * - New models can be added to MODEL_PRICING as they become available
+ * Model-aware pricing per 1M tokens (USD), from the DeepSeek V4 pricing page:
+ * - deepseek-v4-flash: cache hit $0.0028, cache miss $0.14, output $0.28
+ * - deepseek-v4-pro:   cache hit $0.003625, cache miss $0.435, output $0.87
+ * The deepseek-chat / deepseek-reasoner aliases resolve to deepseek-v4-flash,
+ * so they carry v4-flash pricing. New models can be added to MODEL_PRICING.
  */
 /**
  * Pricing structure for a model
@@ -14,7 +16,7 @@ export interface ModelPricing {
     cache_miss: number;
     output: number;
 }
-/** Default pricing (V3.2 unified) — used for unknown models */
+/** v4-flash pricing — also the default for unknown models (cheapest, most common) */
 export declare const DEFAULT_PRICING: ModelPricing;
 /** Backward-compatible alias */
 export declare const PRICING: ModelPricing;
@@ -36,7 +38,7 @@ export interface CostBreakdown {
 }
 /**
  * Calculate cost for a request based on token usage.
- * Supports V3.2 cache hit/miss pricing. If cache fields are absent,
+ * Supports cache hit/miss pricing. If cache fields are absent,
  * treats all input tokens as cache miss (backward compatible).
  */
 export declare function calculateCost(usage: {

package/dist/cost.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"cost.d.ts","sourceRoot":"","sources":["../src/cost.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AAEH;;GAEG;AACH,MAAM,WAAW,YAAY;IAC3B,SAAS,EAAE,MAAM,CAAC;IAClB,UAAU,EAAE,MAAM,CAAC;IACnB,MAAM,EAAE,MAAM,CAAC;CAChB;AAED,+DAA+D;AAC/D,eAAO,MAAM,eAAe,EAAE,YAI7B,CAAC;AAEF,gCAAgC;AAChC,eAAO,MAAM,OAAO,cAAkB,CAAC;AAEvC,2EAA2E;AAC3E,eAAO,MAAM,aAAa,EAAE,MAAM,CAAC,MAAM,EAAE,YAAY,CAGtD,CAAC;AAEF;;GAEG;AACH,wBAAgB,UAAU,CAAC,KAAK,CAAC,EAAE,MAAM,GAAG,YAAY,CAKvD;AAED;;GAEG;AACH,MAAM,WAAW,aAAa;IAC5B,SAAS,EAAE,MAAM,CAAC;IAClB,UAAU,EAAE,MAAM,CAAC;IACnB,SAAS,EAAE,MAAM,CAAC;IAClB,aAAa,CAAC,EAAE,MAAM,CAAC;IACvB,YAAY,CAAC,EAAE,MAAM,CAAC;CACvB;AAED;;;;GAIG;AACH,wBAAgB,aAAa,CAAC,KAAK,EAAE;IACnC,aAAa,EAAE,MAAM,CAAC;IACtB,iBAAiB,EAAE,MAAM,CAAC;IAC1B,uBAAuB,CAAC,EAAE,MAAM,CAAC;IACjC,wBAAwB,CAAC,EAAE,MAAM,CAAC;CACnC,EAAE,KAAK,CAAC,EAAE,MAAM,GAAG,aAAa,CA8ChC;AAED;;GAEG;AACH,wBAAgB,UAAU,CAAC,SAAS,EAAE,aAAa,GAAG,MAAM,CAqB3D"}
+{"version":3,"file":"cost.d.ts","sourceRoot":"","sources":["../src/cost.ts"],"names":[],"mappings":"AAAA;;;;;;;;;GASG;AAEH;;GAEG;AACH,MAAM,WAAW,YAAY;IAC3B,SAAS,EAAE,MAAM,CAAC;IAClB,UAAU,EAAE,MAAM,CAAC;IACnB,MAAM,EAAE,MAAM,CAAC;CAChB;AAED,qFAAqF;AACrF,eAAO,MAAM,eAAe,EAAE,YAI7B,CAAC;AAEF,gCAAgC;AAChC,eAAO,MAAM,OAAO,cAAkB,CAAC;AAEvC,2EAA2E;AAC3E,eAAO,MAAM,aAAa,EAAE,MAAM,CAAC,MAAM,EAAE,YAAY,CAMtD,CAAC;AAEF;;GAEG;AACH,wBAAgB,UAAU,CAAC,KAAK,CAAC,EAAE,MAAM,GAAG,YAAY,CAKvD;AAED;;GAEG;AACH,MAAM,WAAW,aAAa;IAC5B,SAAS,EAAE,MAAM,CAAC;IAClB,UAAU,EAAE,MAAM,CAAC;IACnB,SAAS,EAAE,MAAM,CAAC;IAClB,aAAa,CAAC,EAAE,MAAM,CAAC;IACvB,YAAY,CAAC,EAAE,MAAM,CAAC;CACvB;AAED;;;;GAIG;AACH,wBAAgB,aAAa,CAAC,KAAK,EAAE;IACnC,aAAa,EAAE,MAAM,CAAC;IACtB,iBAAiB,EAAE,MAAM,CAAC;IAC1B,uBAAuB,CAAC,EAAE,MAAM,CAAC;IACjC,wBAAwB,CAAC,EAAE,MAAM,CAAC;CACnC,EAAE,KAAK,CAAC,EAAE,MAAM,GAAG,aAAa,CA8ChC;AAED;;GAEG;AACH,wBAAgB,UAAU,CAAC,SAAS,EAAE,aAAa,GAAG,MAAM,CAqB3D"}

package/dist/cost.js

@@ -2,22 +2,27 @@
  * Cost Calculation Module
  * Handles pricing and cost formatting for DeepSeek API requests
  *
- * Model-aware pricing per 1M tokens (USD):
- * - deepseek-chat / deepseek-reasoner (V3.2): cache hit $0.028, cache miss $0.28, output $0.42
- * - New models can be added to MODEL_PRICING as they become available
+ * Model-aware pricing per 1M tokens (USD), from the DeepSeek V4 pricing page:
+ * - deepseek-v4-flash: cache hit $0.0028, cache miss $0.14, output $0.28
+ * - deepseek-v4-pro:   cache hit $0.003625, cache miss $0.435, output $0.87
+ * The deepseek-chat / deepseek-reasoner aliases resolve to deepseek-v4-flash,
+ * so they carry v4-flash pricing. New models can be added to MODEL_PRICING.
  */
-/** Default pricing (V3.2 unified) — used for unknown models */
+/** v4-flash pricing — also the default for unknown models (cheapest, most common) */
 export const DEFAULT_PRICING = {
-    cache_hit: 0.028,
-    cache_miss: 0.28,
-    output: 0.42,
+    cache_hit: 0.0028,
+    cache_miss: 0.14,
+    output: 0.28,
 };
 /** Backward-compatible alias */
 export const PRICING = DEFAULT_PRICING;
 /** Per-model pricing map. Add new models here as they become available. */
 export const MODEL_PRICING = {
-    'deepseek-chat': { cache_hit: 0.028, cache_miss: 0.28, output: 0.42 },
-    'deepseek-reasoner': { cache_hit: 0.028, cache_miss: 0.28, output: 0.42 },
+    'deepseek-v4-flash': { cache_hit: 0.0028, cache_miss: 0.14, output: 0.28 },
+    'deepseek-v4-pro': { cache_hit: 0.003625, cache_miss: 0.435, output: 0.87 },
+    // Compatibility aliases (resolve to v4-flash on the wire)
+    'deepseek-chat': { cache_hit: 0.0028, cache_miss: 0.14, output: 0.28 },
+    'deepseek-reasoner': { cache_hit: 0.0028, cache_miss: 0.14, output: 0.28 },
 };
 /**
  * Get pricing for a specific model. Falls back to DEFAULT_PRICING for unknown models.
@@ -30,7 +35,7 @@ export function getPricing(model) {
 }
 /**
  * Calculate cost for a request based on token usage.
- * Supports V3.2 cache hit/miss pricing. If cache fields are absent,
+ * Supports cache hit/miss pricing. If cache fields are absent,
  * treats all input tokens as cache miss (backward compatible).
  */
 export function calculateCost(usage, model) {

package/dist/cost.js.map

@@ -1 +1 @@
-{"version":3,"file":"cost.js","sourceRoot":"","sources":["../src/cost.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AAWH,+DAA+D;AAC/D,MAAM,CAAC,MAAM,eAAe,GAAiB;IAC3C,SAAS,EAAE,KAAK;IAChB,UAAU,EAAE,IAAI;IAChB,MAAM,EAAE,IAAI;CACb,CAAC;AAEF,gCAAgC;AAChC,MAAM,CAAC,MAAM,OAAO,GAAG,eAAe,CAAC;AAEvC,2EAA2E;AAC3E,MAAM,CAAC,MAAM,aAAa,GAAiC;IACzD,eAAe,EAAE,EAAE,SAAS,EAAE,KAAK,EAAE,UAAU,EAAE,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE;IACrE,mBAAmB,EAAE,EAAE,SAAS,EAAE,KAAK,EAAE,UAAU,EAAE,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE;CAC1E,CAAC;AAEF;;GAEG;AACH,MAAM,UAAU,UAAU,CAAC,KAAc;IACvC,IAAI,KAAK,IAAI,KAAK,IAAI,aAAa,EAAE,CAAC;QACpC,OAAO,aAAa,CAAC,KAAK,CAAC,CAAC;IAC9B,CAAC;IACD,OAAO,eAAe,CAAC;AACzB,CAAC;AAaD;;;;GAIG;AACH,MAAM,UAAU,aAAa,CAAC,KAK7B,EAAE,KAAc;IACf,MAAM,EACJ,aAAa,EACb,iBAAiB,EACjB,uBAAuB,EACvB,wBAAwB,GACzB,GAAG,KAAK,CAAC;IAEV,MAAM,OAAO,GAAG,UAAU,CAAC,KAAK,CAAC,CAAC;IAElC,IAAI,SAAiB,CAAC;IACtB,IAAI,aAAiC,CAAC;IACtC,IAAI,YAAgC,CAAC;IAErC,IACE,uBAAuB,KAAK,SAAS;QACrC,wBAAwB,KAAK,SAAS,EACtC,CAAC;QACD,sBAAsB;QACtB,MAAM,OAAO,GACX,CAAC,uBAAuB,GAAG,SAAS,CAAC,GAAG,OAAO,CAAC,SAAS,CAAC;QAC5D,MAAM,QAAQ,GACZ,CAAC,wBAAwB,GAAG,SAAS,CAAC,GAAG,OAAO,CAAC,UAAU,CAAC;QAC9D,SAAS,GAAG,OAAO,GAAG,QAAQ,CAAC;QAE/B,IAAI,aAAa,GAAG,CAAC,EAAE,CAAC;YACtB,aAAa,GAAG,uBAAuB,GAAG,aAAa,CAAC;QAC1D,CAAC;QAED,uDAAuD;QACvD,MAAM,WAAW,GAAG,CAAC,aAAa,GAAG,SAAS,CAAC,GAAG,OAAO,CAAC,UAAU,CAAC;QACrE,YAAY,GAAG,WAAW,GAAG,SAAS,CAAC;IACzC,CAAC;SAAM,CAAC;QACN,qDAAqD;QACrD,SAAS,GAAG,CAAC,aAAa,GAAG,SAAS,CAAC,GAAG,OAAO,CAAC,UAAU,CAAC;IAC/D,CAAC;IAED,MAAM,UAAU,GAAG,CAAC,iBAAiB,GAAG,SAAS,CAAC,GAAG,OAAO,CAAC,MAAM,CAAC;IAEpE,OAAO;QACL,SAAS;QACT,UAAU;QACV,SAAS,EAAE,SAAS,GAAG,UAAU;QACjC,aAAa;QACb,YAAY;KACb,CAAC;AACJ,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,UAAU,CAAC,SAAwB;IACjD,MAAM,IAAI,GAAG,SAAS,CAAC,SAAS,CAAC;IACjC,IAAI,SAAiB,CAAC;IAEtB,IAAI,IAAI,GAAG,IAAI,EAAE,CAAC;QAChB,SAAS,GAAG,IAAI,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,CAAC;IACpC,CAAC;SAAM,CAAC;QACN,SAAS,GAAG,IAAI,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,CAAC;IACpC,CAAC;IAED,IACE,SAAS,CAAC,aAAa,KAAK,SAAS;QACrC,SAAS,CAAC,aAAa,GAAG,CAAC;QAC3B,SAAS,CAAC,YAAY,KAAK,SAAS;QACpC,SAAS,CAAC,YAAY,GAAG,CAAC,EAC1B,CAAC;QACD,MAAM,GAAG,GAAG,IAAI,CAAC,KAAK,CAAC,SAAS,CAAC,aAAa,GAAG,GAAG,CAAC,CAAC;QACtD,SAAS,IAAI,gBAAgB,GAAG,cAAc,SAAS,CAAC,YAAY,CAAC,OAAO,CAAC,CAAC,CAAC,GAAG,CAAC;IACrF,CAAC;IAED,OAAO,SAAS,CAAC;AACnB,CAAC"}
+{"version":3,"file":"cost.js","sourceRoot":"","sources":["../src/cost.ts"],"names":[],"mappings":"AAAA;;;;;;;;;GASG;AAWH,qFAAqF;AACrF,MAAM,CAAC,MAAM,eAAe,GAAiB;IAC3C,SAAS,EAAE,MAAM;IACjB,UAAU,EAAE,IAAI;IAChB,MAAM,EAAE,IAAI;CACb,CAAC;AAEF,gCAAgC;AAChC,MAAM,CAAC,MAAM,OAAO,GAAG,eAAe,CAAC;AAEvC,2EAA2E;AAC3E,MAAM,CAAC,MAAM,aAAa,GAAiC;IACzD,mBAAmB,EAAE,EAAE,SAAS,EAAE,MAAM,EAAE,UAAU,EAAE,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE;IAC1E,iBAAiB,EAAE,EAAE,SAAS,EAAE,QAAQ,EAAE,UAAU,EAAE,KAAK,EAAE,MAAM,EAAE,IAAI,EAAE;IAC3E,0DAA0D;IAC1D,eAAe,EAAE,EAAE,SAAS,EAAE,MAAM,EAAE,UAAU,EAAE,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE;IACtE,mBAAmB,EAAE,EAAE,SAAS,EAAE,MAAM,EAAE,UAAU,EAAE,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE;CAC3E,CAAC;AAEF;;GAEG;AACH,MAAM,UAAU,UAAU,CAAC,KAAc;IACvC,IAAI,KAAK,IAAI,KAAK,IAAI,aAAa,EAAE,CAAC;QACpC,OAAO,aAAa,CAAC,KAAK,CAAC,CAAC;IAC9B,CAAC;IACD,OAAO,eAAe,CAAC;AACzB,CAAC;AAaD;;;;GAIG;AACH,MAAM,UAAU,aAAa,CAAC,KAK7B,EAAE,KAAc;IACf,MAAM,EACJ,aAAa,EACb,iBAAiB,EACjB,uBAAuB,EACvB,wBAAwB,GACzB,GAAG,KAAK,CAAC;IAEV,MAAM,OAAO,GAAG,UAAU,CAAC,KAAK,CAAC,CAAC;IAElC,IAAI,SAAiB,CAAC;IACtB,IAAI,aAAiC,CAAC;IACtC,IAAI,YAAgC,CAAC;IAErC,IACE,uBAAuB,KAAK,SAAS;QACrC,wBAAwB,KAAK,SAAS,EACtC,CAAC;QACD,sBAAsB;QACtB,MAAM,OAAO,GACX,CAAC,uBAAuB,GAAG,SAAS,CAAC,GAAG,OAAO,CAAC,SAAS,CAAC;QAC5D,MAAM,QAAQ,GACZ,CAAC,wBAAwB,GAAG,SAAS,CAAC,GAAG,OAAO,CAAC,UAAU,CAAC;QAC9D,SAAS,GAAG,OAAO,GAAG,QAAQ,CAAC;QAE/B,IAAI,aAAa,GAAG,CAAC,EAAE,CAAC;YACtB,aAAa,GAAG,uBAAuB,GAAG,aAAa,CAAC;QAC1D,CAAC;QAED,uDAAuD;QACvD,MAAM,WAAW,GAAG,CAAC,aAAa,GAAG,SAAS,CAAC,GAAG,OAAO,CAAC,UAAU,CAAC;QACrE,YAAY,GAAG,WAAW,GAAG,SAAS,CAAC;IACzC,CAAC;SAAM,CAAC;QACN,qDAAqD;QACrD,SAAS,GAAG,CAAC,aAAa,GAAG,SAAS,CAAC,GAAG,OAAO,CAAC,UAAU,CAAC;IAC/D,CAAC;IAED,MAAM,UAAU,GAAG,CAAC,iBAAiB,GAAG,SAAS,CAAC,GAAG,OAAO,CAAC,MAAM,CAAC;IAEpE,OAAO;QACL,SAAS;QACT,UAAU;QACV,SAAS,EAAE,SAAS,GAAG,UAAU;QACjC,aAAa;QACb,YAAY;KACb,CAAC;AACJ,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,UAAU,CAAC,SAAwB;IACjD,MAAM,IAAI,GAAG,SAAS,CAAC,SAAS,CAAC;IACjC,IAAI,SAAiB,CAAC;IAEtB,IAAI,IAAI,GAAG,IAAI,EAAE,CAAC;QAChB,SAAS,GAAG,IAAI,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,CAAC;IACpC,CAAC;SAAM,CAAC;QACN,SAAS,GAAG,IAAI,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,CAAC;IACpC,CAAC;IAED,IACE,SAAS,CAAC,aAAa,KAAK,SAAS;QACrC,SAAS,CAAC,aAAa,GAAG,CAAC;QAC3B,SAAS,CAAC,YAAY,KAAK,SAAS;QACpC,SAAS,CAAC,YAAY,GAAG,CAAC,EAC1B,CAAC;QACD,MAAM,GAAG,GAAG,IAAI,CAAC,KAAK,CAAC,SAAS,CAAC,aAAa,GAAG,GAAG,CAAC,CAAC;QACtD,SAAS,IAAI,gBAAgB,GAAG,cAAc,SAAS,CAAC,YAAY,CAAC,OAAO,CAAC,CAAC,CAAC,GAAG,CAAC;IACrF,CAAC;IAED,OAAO,SAAS,CAAC;AACnB,CAAC"}

package/dist/deepseek-client.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"deepseek-client.d.ts","sourceRoot":"","sources":["../src/deepseek-client.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAUH,OAAO,KAAK,EACV,oBAAoB,EACpB,sBAAsB,EACtB,oBAAoB,EAGpB,YAAY,EAEb,MAAM,YAAY,CAAC;AAqBpB,oDAAoD;AACpD,MAAM,WAAW,kCAAmC,SAAQ,sBAAsB;IAChF,QAAQ,CAAC,EAAE,YAAY,CAAC;CACzB;AAwBD,qBAAa,cAAc;IACzB,OAAO,CAAC,MAAM,CAAS;IACvB,OAAO,CAAC,eAAe,CAAqC;IAC5D,OAAO,CAAC,WAAW,CAAS;IAC5B,OAAO,CAAC,cAAc,CAAS;;IAgB/B;;OAEG;IACH,OAAO,CAAC,iBAAiB;IASzB;;OAEG;IACH,uBAAuB;IAQvB;;OAEG;IACH,OAAO,CAAC,kBAAkB;IAgE1B;;OAEG;IACH,OAAO,CAAC,SAAS;IASjB;;OAEG;IACH,OAAO,CAAC,aAAa;IAoCrB;;OAEG;IACG,oBAAoB,CACxB,MAAM,EAAE,oBAAoB,GAC3B,OAAO,CAAC,kCAAkC,CAAC;IAqD9C;;;OAGG;IACG,6BAA6B,CACjC,MAAM,EAAE,oBAAoB,GAC3B,OAAO,CAAC,kCAAkC,CAAC;IAyC9C;;OAEG;YACW,cAAc;IA4F5B;;OAEG;IACG,cAAc,IAAI,OAAO,CAAC,OAAO,CAAC;CAazC"}
+{"version":3,"file":"deepseek-client.d.ts","sourceRoot":"","sources":["../src/deepseek-client.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAUH,OAAO,KAAK,EACV,oBAAoB,EACpB,sBAAsB,EACtB,oBAAoB,EAGpB,YAAY,EAEb,MAAM,YAAY,CAAC;AAmBpB,oDAAoD;AACpD,MAAM,WAAW,kCAAmC,SAAQ,sBAAsB;IAChF,QAAQ,CAAC,EAAE,YAAY,CAAC;CACzB;AAwBD,qBAAa,cAAc;IACzB,OAAO,CAAC,MAAM,CAAS;IACvB,OAAO,CAAC,eAAe,CAAqC;IAC5D,OAAO,CAAC,WAAW,CAAS;IAC5B,OAAO,CAAC,cAAc,CAAS;;IAgB/B;;OAEG;IACH,OAAO,CAAC,iBAAiB;IASzB;;OAEG;IACH,uBAAuB;IAQvB;;OAEG;IACH,OAAO,CAAC,kBAAkB;IAiF1B;;OAEG;IACH,OAAO,CAAC,SAAS;IASjB;;OAEG;IACH,OAAO,CAAC,aAAa;IAoCrB;;OAEG;IACG,oBAAoB,CACxB,MAAM,EAAE,oBAAoB,GAC3B,OAAO,CAAC,kCAAkC,CAAC;IAqD9C;;;OAGG;IACG,6BAA6B,CACjC,MAAM,EAAE,oBAAoB,GAC3B,OAAO,CAAC,kCAAkC,CAAC;IAyC9C;;OAEG;YACW,cAAc;IA4F5B;;OAEG;IACG,cAAc,IAAI,OAAO,CAAC,OAAO,CAAC;CAazC"}

package/dist/deepseek-client.js

@@ -8,21 +8,19 @@ import { getConfig } from './config.js';
 import { ApiError, CircuitBreakerOpenError, FallbackExhaustedError, } from './errors.js';
 import { CircuitBreaker } from './circuit-breaker.js';
 import { hasReasoningContent, getErrorMessage } from './types.js';
-/** Parameters that are incompatible with thinking mode */
-const THINKING_INCOMPATIBLE_PARAMS = [
-    'temperature',
-    'top_p',
-    'frequency_penalty',
-    'presence_penalty',
-];
+/** Parameters that the API ignores while thinking mode is active */
+const THINKING_INCOMPATIBLE_PARAMS = ['temperature', 'top_p'];
 /**
  * Fallback order per model. Each model lists fallback candidates in priority order.
  * When a model fails with a retryable error, the first available fallback is tried.
- * Add new models here as they become available in the API.
+ * v4-flash and v4-pro are genuinely different models, so they back each other up.
+ * The chat/reasoner aliases (which resolve to v4-flash) fall back to v4-pro.
  */
 const FALLBACK_ORDER = {
-    'deepseek-chat': ['deepseek-reasoner'],
-    'deepseek-reasoner': ['deepseek-chat'],
+    'deepseek-v4-flash': ['deepseek-v4-pro'],
+    'deepseek-v4-pro': ['deepseek-v4-flash'],
+    'deepseek-chat': ['deepseek-v4-pro'],
+    'deepseek-reasoner': ['deepseek-v4-pro'],
 };
 /**
  * Check if an error is retryable / should trigger fallback
@@ -93,25 +91,36 @@ export class DeepSeekClient {
      * Build request params shared between streaming and non-streaming
      */
     buildRequestParams(params, stream) {
-        // Transparently convert reasoner to chat + thinking
-        // deepseek-reasoner = deepseek-chat with thinking always enabled
-        // chat + thinking supports function calling, which reasoner alone does not
-        let effectiveModel = params.model;
-        let effectiveThinking = params.thinking;
-        if (params.model === 'deepseek-reasoner') {
-            effectiveModel = 'deepseek-chat';
-            effectiveThinking = { type: 'enabled' };
-            console.error('[DeepSeek MCP] Routing: reasoner -> chat + thinking');
+        // Resolve the user-facing model + thinking flag to what the API expects.
+        // v4-flash / v4-pro are the live API models. deepseek-chat and
+        // deepseek-reasoner are compatibility aliases (the API retires those names
+        // on 2026-07-24), so they are translated to v4-flash here. The API defaults
+        // thinking to ENABLED, so we always send an explicit flag — including
+        // disabled — to keep the historical fast (non-thinking) default.
+        let effectiveModel;
+        let effectiveThinking;
+        switch (params.model) {
+            case 'deepseek-reasoner':
+                // "thinking" alias: always reason
+                effectiveModel = 'deepseek-v4-flash';
+                effectiveThinking = { type: 'enabled' };
+                console.error('[DeepSeek MCP] Routing: deepseek-reasoner -> deepseek-v4-flash + thinking');
+                break;
+            case 'deepseek-chat':
+                // "non-thinking" alias: fast by default, but honour an explicit thinking:enabled
+                effectiveModel = 'deepseek-v4-flash';
+                effectiveThinking = params.thinking ?? { type: 'disabled' };
+                console.error('[DeepSeek MCP] Routing: deepseek-chat -> deepseek-v4-flash');
+                break;
+            default:
+                // deepseek-v4-flash / deepseek-v4-pro: pass through, default to non-thinking
+                effectiveModel = params.model;
+                effectiveThinking = params.thinking ?? { type: 'disabled' };
         }
-        const isThinkingEnabled = effectiveThinking?.type === 'enabled';
-        // Filter incompatible params when thinking mode is active
+        const isThinkingEnabled = effectiveThinking.type === 'enabled';
+        // Warn when caller-supplied sampling params will be ignored under thinking mode
         if (isThinkingEnabled) {
-            const filtered = [];
-            for (const key of THINKING_INCOMPATIBLE_PARAMS) {
-                if (params[key] !== undefined) {
-                    filtered.push(key);
-                }
-            }
+            const filtered = THINKING_INCOMPATIBLE_PARAMS.filter((key) => params[key] !== undefined);
             if (filtered.length > 0) {
                 console.error(`[DeepSeek MCP] Warning: Thinking mode active, ignoring incompatible params: ${filtered.join(', ')}`);
             }
@@ -119,17 +128,21 @@ export class DeepSeekClient {
         const requestParams = {
             model: effectiveModel,
             messages: params.messages,
-            temperature: isThinkingEnabled ? undefined : (params.temperature ?? 1.0),
             max_tokens: params.max_tokens,
-            top_p: isThinkingEnabled ? undefined : params.top_p,
-            frequency_penalty: isThinkingEnabled ? undefined : params.frequency_penalty,
-            presence_penalty: isThinkingEnabled ? undefined : params.presence_penalty,
             stop: params.stop,
             stream,
+            // Always explicit: the API's thinking default is enabled
+            thinking: effectiveThinking,
         };
-        // Pass thinking as top-level param (OpenAI SDK v6 passes unknown props to the API body)
-        if (effectiveThinking?.type === 'enabled') {
-            requestParams.thinking = effectiveThinking;
+        // Sampling params only take effect in non-thinking mode
+        if (!isThinkingEnabled) {
+            requestParams.temperature = params.temperature ?? 1.0;
+            if (params.top_p !== undefined)
+                requestParams.top_p = params.top_p;
+        }
+        // reasoning_effort only applies while thinking
+        if (isThinkingEnabled && params.reasoning_effort) {
+            requestParams.reasoning_effort = params.reasoning_effort;
         }
         // Pass response_format for JSON mode
         if (params.response_format) {
@@ -360,7 +373,7 @@ export class DeepSeekClient {
     async testConnection() {
         try {
             const response = await this.createChatCompletion({
-                model: 'deepseek-chat',
+                model: 'deepseek-v4-flash',
                 messages: [{ role: 'user', content: 'Hi' }],
                 max_tokens: 10,
             });