Claude visitors: new accounts receive $1.00 in bonus API credit within 30 days. One per person; usage credits only.
Use DeepSeek in WorkBuddy.
DeepSeek V4 is a reasoning-capable LLM available through the RunAPI MCP server — Flash for speed, Pro for quality, with built-in chain-of-thought that activates via a request parameter. Add RunAPI as an MCP server in WorkBuddy once, and the workbench agent can call DeepSeek at 50% of official DeepSeek rates.
Using the runapi MCP server, ask DeepSeek to help me with a coding task.
Requirements:
- Call the chat tool exposed by the runapi MCP server.
- Set model to "deepseek-v4-pro".
- Enable reasoning by setting thinking.type to "enabled".
- Task: write a Python function that finds the longest palindromic
substring in O(n) time using Manacher's algorithm, then explain
the key insight in two sentences.
- Show me the reply and the token usage.
{
"mcp_servers": {
"runapi": {
"url": "https://runapi.ai/mcp",
"transport": "streamable_http",
"headers": {
"Authorization": "Bearer $RUNAPI_API_KEY"
}
}
}
}
{
"model": "deepseek-v4-pro",
"choices": [
{
"message": {
"role": "assistant",
"content": "Here is Manacher's algorithm implementation..."
},
"finish_reason": "stop"
}
],
"usage": {
"prompt_tokens": 42,
"completion_tokens": 356,
"total_tokens": 398
}
}
Use DeepSeek in WorkBuddy in three steps
Add RunAPI as an MCP server
Set RUNAPI_API_KEY in the environment that launches WorkBuddy, then register an MCP server named runapi pointing at https://runapi.ai/mcp over streamable HTTP. The Authorization header carries your key — DeepSeek shows up as a callable tool alongside every other RunAPI model.
export RUNAPI_API_KEY=runapi_xxx
# MCP server: runapi → https://runapi.ai/mcp
Prompt the workbench agent
Type a natural-language request into WorkBuddy and name the model. Ask for deepseek-v4-pro for quality or deepseek-v4-flash for speed, and mention thinking.type set to enabled when you want visible step-by-step reasoning — no separate model slug needed.
model: deepseek-v4-pro
Read the reply in the workbench
WorkBuddy calls the runapi MCP tool and the reply lands back in the conversation, with token usage reported for billing transparency. The same flow works whether the agent calls DeepSeek, GPT, or Claude through RunAPI — only the model name changes.
"total_tokens": 398
DeepSeek chat parameters
| Parameter | Type | Description |
|---|---|---|
model |
string |
Required. deepseek-v4-pro or deepseek-v4-flash. |
messages |
array |
Required. Array of message objects with role (system, user, assistant) and content fields. |
stream |
boolean |
Optional. Set to true for streamed output. Default false. |
max_tokens |
integer |
Optional. Maximum number of tokens to generate. |
temperature |
number |
Optional. Sampling temperature between 0 and 2. Lower values are more deterministic. |
thinking |
object |
Optional. Set type to "enabled" to activate DeepSeek's built-in reasoning mode. |
What is DeepSeek on WorkBuddy?
DeepSeek V4 is a reasoning-capable LLM that WorkBuddy reaches through the RunAPI MCP server — no per-provider signup, no extra keys. It supports built-in chain-of-thought reasoning via a request parameter (no separate model slug needed), and RunAPI prices it at 50% of the already-cheap official DeepSeek rates, making it the lowest per-token option for a reasoning-capable model your workbench agent can call.
DeepSeek use cases
Budget-friendly workbench tasks at high volume
Route everyday drafting, summarizing, and batch translation from WorkBuddy to DeepSeek instead of a premium model. At 50% of DeepSeek's already low rates through RunAPI, token costs stay small even when the workbench runs thousands of requests.
Reasoning-heavy tasks with visible chain-of-thought
Enable thinking mode for math, logic, code debugging, or planning tasks where step-by-step reasoning improves accuracy. The thinking tokens come back in the tool result, so you can see how the model reached its answer inside WorkBuddy.
Code generation and debugging
Ask the workbench agent to send code generation, algorithm implementation, and debugging work to DeepSeek V4 Pro. It scores well on coding benchmarks and costs a fraction of GPT-5.5 per token.
DeepSeek + WorkBuddy questions
No. WorkBuddy speaks MCP natively, and the runapi MCP server exposes DeepSeek as a standard tool. Register the server once with your RUNAPI_API_KEY, then name deepseek-v4-pro or deepseek-v4-flash in your prompt — no DeepSeek account or separate credentials needed.
When you enable reasoning (thinking.type set to enabled), DeepSeek generates additional "thinking" tokens that show its chain-of-thought process. These tokens count as output tokens and are billed accordingly. They improve accuracy on complex tasks but increase total token usage.
Flash prioritizes speed with lower latency and cost. Pro prioritizes quality with better performance on complex reasoning, coding, and multi-step tasks. Both are reachable through the same runapi MCP server and accept the same parameters.
Yes. Every model behind the runapi MCP server is a tool the workbench agent can pick per request. Ask for DeepSeek on a budget-sensitive step and a premium model on a hard one — same key, same server, no config change.
RunAPI prices DeepSeek tokens at 50% of official DeepSeek rates. DeepSeek is already one of the lowest-cost LLM providers, so through RunAPI you get the lowest per-token price available for a reasoning-capable model. Billing is per-token with no subscription or minimum spend.
WorkBuddy general setup
Not connected yet? Start with the RunAPI MCP setup guide for WorkBuddy.
WorkBuddy setup guide →Try DeepSeek in WorkBuddy today.
Get a free RunAPI key, add the MCP server to WorkBuddy, and ask the workbench agent for deepseek-v4-pro — reasoning included, at the lowest token price available.