Bluwing
Sign inStart free
01Overview02Why it's cheaper03Features04How it works05Promises06Pricing07Get started
BlogSign in

Bluwing: the tool-calling engine for AI agents. The right tool and real arguments, in milliseconds.

Scroll

Thetool-callingengineforAIagents.Therighttool,realarguments,inmilliseconds.

The tool-calling engine for AI agents

Tool calls in milliseconds.Without the guessing.

Bluwing decides which of your tools to call and fills every argument from what the user actually said. It speaks OpenAI’s format, answers in about 60 ms, and costs a fraction of a cent a turn.

Start freeSee what you save

200 free credits. No card needed.

  • About 60 ms a turn
  • From $0.00086 a call
  • OpenAI and Anthropic formats
POST /v1/chat/completions0ms
User

2,000tools
  • Email
  • Calendar
  • Dining
  • Payments
  • Files
  • Support
  • find_restaurant
  • book_table
  • cancel_booking
  • list_bookings
Responsetool_calls
Waiting for the request
{  "status": "tool_calls",  "tool_calls": [{    "name": "book_table",    "arguments": {      "restaurant": "Nok",      "party_size": 4,      "datetime": "2026-11-07T19:00"    }  }]}
  1. 1Read
  2. 2Narrow
  3. 3Fill
  4. 4Return

Example 1 of 4: “Book Nok for four people tomorrow at 7pm”. Bluwing calls book_table with restaurant Nok, party size 4 and tomorrow at 19:00, each taken from the sentence.

Slowed down so you can follow it. The clock shows the engine’s target time.

Why it's cheaper

Stop paying for tokens just to pick a tool.

An LLM rereads your whole tool list on every turn, and you pay for each of those tokens. Bluwing charges one credit per turn, however many tools you have.

Your monthly bill, both ways
Turns a month, on one scale. Pick a tile to flip the board.
Same scale
$6k$4.5k$3k$1.5k
1k2.5k5k10k20k35k50k75k100k125k150k200k

One turn costs

LLM tool calling
$0.0215
Bluwing
$0.00086
  • Tokens billed8,1500
  • 1,000 turns$21.50$0.86
  • More tools, higher billYesNo

25×less per turn on Plan S+

LLM side: 8,000 input and 150 output tokens a turn, at $2.50 and $10.00 per million. Bluwing side: plan prices, top-ups included.

WhatLLM tool callingBluwing
Time per turn0.5 to 3 secondsAbout 60 ms
What you pay forEvery token, every turn1 credit per turn
2,000 toolsAccuracy drops as the list growsNarrowed to a dozen first
Argument valuesCan be made upTaken from the chat or your facts
When nothing fitsOften calls something anywayCalls nothing
Same request twiceCan differSame result, every time

Times are engineering targets on 8 vCPUs with 2,000 tools, not independent benchmarks. Bluwing chooses tools and fills arguments; your app or LLM still writes the reply.

Features

Everything a tool call needs. Nothing it doesn't.

Bluwing does one job: decide which tool to call and with what. These tiles are it doing that job. Try the big one.

POST /v1/chat/completionssupport-desk · 2,000 tools
User

Book Nok for four tomorrow at 7pm

tool_calls44 ms
status: "tool_calls"tool: "book_table" restaurant: "Nok" party_size: 4 datetime: "2026-11-07T19:00"

Try a request

Pick something a user might say. Bluwing returns the tool, the arguments, and how long it took.

  1. Bluwing0 ms
  2. LLM0 ms
Running the same request…

Back before a spinner shows

About 60 ms a turn, against one to three seconds for an LLM.

Book Nok for four people tomorrow at 7pm

  • restaurant"Nok"
  • party_size4
  • datetime"Sat 19:00"

Every argument has a source

Values come from what the user said or facts you pass in. Nothing is made up.

  1. Book Nok tomorrow at 7pm

  2. Sure. For how many people?

    needs_input40 ms
  3. Four

  4. Booked Nok for 4, tomorrow at 19:00.

    book_table44 ms

It asks instead of guessing

A missing detail comes back by name, so your app can ask.

“What’s the capital of Peru?”

book_tablerefund_ordergmail_searchmove_eventsend_invoicecreate_ticket

tool_calls: []7 ms

Knows when to do nothing

Small talk and questions get no tool call, so nothing fires by accident.

2,000tools in your toolset

Thousands of tools, one decision

A tree narrows 2,000 tools to a dozen before it chooses, so more tools don't mean worse picks.

 from openai import OpenAI -client = OpenAI()+client = OpenAI(base_url="https://api.bluwing.si/v1")  res = client.chat.completions.create(-    model="gpt-4o",+    model="brick:support-desk@live",     messages=messages, tools=tools, )

Two lines in your code

Keep your OpenAI client. Change the base URL and the model name.

  • @stagingv8
  • @livev7

Change tools without breaking live

Test a new version on staging, then move live to it. Rolling back is one call.

How it works

Live in an afternoon. Two of the four steps you already do.

Upload the tool definitions you already have. From then on, every request is a normal tool-calling request.

1{2  "name": "support-desk",3  "tools": [{4    "type": "function",5    "function": {6      "name": "gmail_search",7      "description": "Search Gmail by text, sender and date.",8      "parameters": {9        "type": "object",10        "properties": {11          "query":  { "type": "string" },12          "sender": { "type": "string" }13        },14        "required": ["query"]15      }16    }17  }]18}
Valid JSON SchemaStep 1 of 4

What happens in those 60 ms

Your request is checked, 2,000 tools are narrowed to the one that fits, and each argument is filled from the conversation. No LLM is called on the way.

Inside one requestYour app sends a request, which is checked first. Bluwing then narrows 2,000 tools to one and fills the arguments. The tool call goes back to your app.Your appSendCheckNarrowFillbook_table1Checks the key and the input1 ms22,000 tools down to 16 ms3Fills each argument from the chat42 ms
ReadyPress Send to run it again.

Promises

Rules it never bends. Here they are, holding.

An agent that calls tools has to be predictable. Bluwing is built around nine rules, and each one is checked by its own tests.

Rule 01 of 9

No made-up arguments

Every value comes from the request, the chat, your facts or an earlier tool result.

testThe order number was never said
inuser: "Refund my last order"outstatus: "needs_input"tool: "refund_order"needs_input: ["order_id"]
Heldorder_id is asked for, not invented

What it leaves to you: writing the reply to the user, and running the tool itself. Running tools on our servers is planned for v1.0.

Pricing

Pay per turn. Not per token.

Every plan runs the full engine. Move the ruler to your volume: each plan shows what that month would cost, and the cheapest one lights up.

Calls a month

0

Cheapest: Plan A at $0.00 a month. An LLM would cost about $0.00.

1001k10k200k

You keep $0.00 a month against LLM tool calling.

F
Lowest at your volume

FreeStart here

Try the full engine, then pay as you go.

5k calls a month

$0.00/ mo

All top-ups, after 200 free credits
Plan
$0
Credits
200
Per 1k
$1.50
  • $0.015 / 1M per call
  • Standard execution speed
  • Standard queue
  • 10 RPS limit · 4 concurrent
Start free
A
Lowest at your volume

APlan ABuilders

For side projects and prototypes.

5k calls a month

$0.00/ mo

0% of the 500M included credits
Plan
$4.50
Credits
500M
Per 1k
$1.50
  • $0.015 / 1M per call
  • Accelerated execution speed
  • Fast-path lane
  • 50 RPS limit · 16 concurrent
Choose Plan A
S
Lowest at your volume

SPlan SMost picked

For startups with agents in production.

5k calls a month

$0.00/ mo

0% of the 1B included credits
Plan
$9.00
Credits
1B
Per 1k
$1.50
  • $0.015 / 1M per call
  • Ultra-fast execution speed
  • High priority lane
  • 200 RPS limit · 64 concurrent
Choose Plan S
S+
Lowest at your volume

S+Plan S+Scale

For platforms running fleets of agents.

5k calls a month

$0.00/ mo

0% of the 2.2B included credits
Plan
$19.00
Credits
2.2B
Per 1k
$1.50
  • $0.015 / 1M per call
  • Real-time execution speed
  • Dedicated zero-wait lane
  • 1000 RPS limit · 256 concurrent
Choose Plan S+
LimitFreePlan APlan SPlan S+
Monthly price$0$4.50$9.00$19.00
Tokens included5M at signup500M / mo1B / mo2.2B / mo
Cost per 1M tokens$0.015 / 1M$0.015 / 1M$0.015 / 1M$0.015 / 1M
Top-up per 100M tokens$1.50$1.50$1.50$1.50
Execution speedStandardAcceleratedUltra-fastReal-time
Priority queueStandard queueFast-path laneHigh priority laneDedicated zero-wait lane
Rate limit10 RPS50 RPS200 RPS1000 RPS
Requests in flight4 concurrent16 concurrent64 concurrent256 concurrent
Active toolsetsUnlimitedUnlimitedUnlimitedUnlimited
Edits and versionsInstant live rolloutsInstant live rolloutsInstant live rolloutsInstant live rollouts
  • Book Nok at 71 cr
    Token-based pricing$0.015 per 1,000,000 input tokens. Output tokens are 100% free.
  • 20 + 1 each
    Multimodal equivalenceStandardized token equivalence for text, images, and audio.
  • JFMAMJJASOND
    Top-ups never expireFrom $1.50 per 100M tokens. Plan tokens roll over for 60 days.
  • no cap
    No rate limitsNo per-second or monthly caps. Sub-millisecond execution engine.

Prices in USD. Dodo Payments handles checkout and collects sales tax and VAT. Estimates leave out toolset builds and round top-ups up to the next 100 credits. The LLM figure uses the assumptions shown above.

Your agent’s next tool callcould take 60 ms.

Start on the free credits. Pay only when the volume is real.

Start freeTalk to the founders
  1. 1Create an account200 credits, no card
  2. 2Upload your toolsThe JSON Schema you have
  3. 3Change one lineYour client, our base URL
Start freeBluwing logoBluwing

Explore

  • Why it's cheaper
  • Features
  • How it works
  • Pricing
  • Documentation
  • API Reference
  • Comparisons
  • Blog & Guides
  • Privacy Policy
  • Terms of Service
  • System Status

Talk to us

© 2026 Bluwing · bluwing.si. The tool-calling engine for AI agents.

  • Blog→
  • Product Hunt
  • LinkedIn