Install
Enable server tools
Server tools are in early access. Ask Baseten to enable them for your workspace, then send thex-baseten-server-tools: true header with each request.
Name the tools in a request
Add the two tool types to thetools array of a Messages request:
type. Baseten holds the schemas, so there’s no Keenable JSON schema to paste and no arguments to declare.
Or use an SDK
The endpoint implements the Anthropic Messages API, so an Anthropic SDK client can reach it throughbase_url. The SDK sends the key as x-api-key, which Baseten doesn’t read, so set both headers yourself:
/v1/chat/completions and on /v1/responses. Prefer Messages. On Messages it’s straightforward to send the assistant, tool-call and tool-result blocks back byte for byte, which keeps the KV cache warm across turns.
The two tools
They’re the MCP server’s two tools, with the same names and the same arguments. The
baseten__keenable__ prefix is Baseten’s namespace for hosted tools. The model fills in every argument it sees.
What comes back
The response contains the whole loop. Each tool round is atool_use block, whose name is the full tool type and whose input holds the arguments the model wrote. The tool_result block with Keenable’s answer follows it. The last text block is the final answer. Baseten runs tool calls that the model emits together at the same time, so independent lookups overlap instead of running one after another.
Handle these three cases:
- The iteration cap. Baseten limits how many tool rounds one request may run. If a turn hits the cap, the response has
stop_reason: "pause_turn"and can end on a tool result with no answer. To continue with a fresh budget, send the message history back in a new request. - Failed tool calls. Baseten marks a failed call as an error for the model, which usually retries or works around it. Failed calls aren’t billed.
max_tokens. The budget covers the reasoning and commentary between the tool rounds, not the answer alone, so a grounded answer needs more of it than the same question asked without tools. If the value is too low, the response hasstop_reason: "max_tokens"and the answer stops mid-word.
Read your usage
Alongside the content, the response carries abaseten object with one entry per tool call: the provider and tool name, its status, whether it was billable, and the Keenable SKU and quantity it was metered as. Failed calls show billable: false and no SKU. Use it to attribute cost per call. Treat the shape as informational: it’s Baseten’s extension field, not a stable protocol.
Models
Baseten enables server tools per model. Check Baseten’s model list for the current set, which includes DeepSeek V4.1 Flash, DeepSeek V4 Pro, GLM 5.3 and 5.3 Fast, and Kimi K3.Billing
No Keenable account is involved, and Keenable doesn’t bill you directly. For each successful call, Keenable reports one of the four SKUs and an amount, and Baseten puts that usage on your Baseten invoice. Failed calls are not billed. Per-model tool usage is at the bottom of the model’s metrics page in the Baseten console.When to call Keenable directly
The hosted tools trade control for having nothing to run. Bring your own key and use the MCP server or the REST API instead when you want to pick the search mode, filter by site or date, search the index as it stood at a past instant, or extract a page with a prompt.Related pages
OpenRouter
Web search and page fetch for any OpenRouter model, as tools it calls itself: from Python, the AI SDK, or the Agent SDK over MCP. No Keenable key needed.
Tool calling
Give an LLM web search and page fetch as tools. Pass the SDK’s TOOLS to any OpenAI-compatible chat completions client and run each call with run_tool_call.
Overview
Install the official Keenable SDK for Python, TypeScript, Go or Rust with one command, search with no key, then add fetch, tool calling and Monitors.
OmniRoute
Route web search and page fetch through Keenable in the OmniRoute AI gateway with provider keenable-search. No key needed, in OmniRoute 3.8.54 and later.