Parallel API| Tool | Model calls | Price | Billed |
|---|---|---|---|
| parallel:search | parallel_search(objective, search_queries) | $1 / 1K searches | Each search, up to 10 results |
| parallel:extract | parallel_extract(urls, objective) | $1 / 1K URLs | Each URL read; unread URLs are free |
Parallel's list price, USD, no markup. Model tokens billed separately. Docs
| Model | Lab | Context | Input / 1M | Cached / 1M | Output / 1M | Providers |
|---|---|---|---|---|---|---|
| Parallel Decisionsparallel/decisions | Parallel | 32K | – | – | – | 1 |
Parallel's price, USD, as of November 8, 2026. Live prices
curl https://api.routerplus.com/v1/chat/completions \
-H "Authorization: Bearer $ROUTERPLUS_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "claude-haiku-4-5",
"messages": [{ "role": "user", "content": "Which RFC defines HTTP semantics?" }],
"tools": [
{ "type": "parallel:search", "parameters": { "max_uses": 3 } },
{ "type": "parallel:extract", "parameters": { "max_urls": 2 } }
],
"tool_choice": { "type": "parallel:search" }
}'curl https://api.routerplus.com/v1/chat/completions \
-H "Authorization: Bearer $ROUTERPLUS_API_KEY" \
-H "Content-Type: application/json" \
-H "x-tm-tools: parallel:search, parallel:extract" \
-d '{
"model": "claude-haiku-4-5",
"messages": [{ "role": "user", "content": "Which RFC defines HTTP semantics?" }]
}'"env": { "ANTHROPIC_CUSTOM_HEADERS": "x-tm-tools: parallel:search,parallel:extract" }[model_providers.routerplus]
http_headers = { "x-tm-tools" = "parallel:search,parallel:extract" }| Tool | Field | Values | Default |
|---|---|---|---|
| parallel:search | max_uses | 1 to 10 | 5 |
| parallel:search | mode | "turbo" or "fast" | "fast" |
| parallel:search | max_results | 1 to 10 | 5 |
| parallel:extract | max_uses | 1 to 10 | 3 |
| parallel:extract | max_urls | 1 to 5 | 3 |
| Endpoint | Each call returns as |
|---|---|
| POST /v1/chat/completions | chat.completion, counts in usage.server_tool_use |
| POST /v1/messages | server_tool_use + parallel_search_tool_result blocks |
| POST /v1/responses | parallel_search_call / parallel_extract_call items |
Organization keys, catalog models. Server tools reference
Add {"type": "parallel:search"} and {"type": "parallel:extract"} to tools, or send the header x-tm-tools: parallel:search, parallel:extract. Both work on /v1/chat/completions, /v1/messages and /v1/responses with an organization key, for catalog models. The model decides when to call them and you get one finished answer.
$1 per 1,000 searches ($0.001 a search) and $1 per 1,000 URLs read ($0.001 a URL), Parallel's list price with no markup. A URL Parallel could not read is not billed, and neither is a call that fails. The model's turns are billed as ordinary chat requests.
The Parallel tools are turned on per gateway, separately from Tako search. A 400 server_tool_unsupported that says "not enabled on this gateway" means they are off: ask support to turn them on.
Yes, in one request. x-tm-tools: tako:search, parallel:search, parallel:extract turns on all three server tools.
The queries, objectives and URLs the model writes, sent with RouterPlus's key. It never gets your messages as such, your key or your organization. Parallel is in the United States. Under its standard terms it may use inputs and outputs, not linked to any individual, to improve its models; assume it does until RouterPlus's account terms with Parallel are settled.
max_uses calls of each tool per request, 60 Parallel calls a minute per organization, 15 seconds per search, 30 seconds per extract and 150 seconds for the whole request. A call that fails or times out is not billed, and the model goes on without it.
Yes, with the x-tm-tools header. Claude Code keeps earlier calls in its history. Codex does not, so it may search again on each request; a lower max_uses in a tools entry bounds the calls.