Spend by model
Every model and provider, side by side.
AI Hub. Know what every AI task costs.
Your AI bills show tokens. AI Hub shows what you got for them: the cost of every solved task, by team, user and repository, and where to save. Claude, ChatGPT, Cursor, Bedrock and more, in one place.
16 AI connectors · No SDK changes
AI cost by task · last 30 days
$1.31 per solved task
11,059 solved tasks32% one-shot rateRetries add 12%
TaskRunsSuccessPer taskPer solved taskSpend
CodingWriting or generating code3,99191%$0.99▼ 8%
$3.6K
AgentsAgentic workflows2,84179%$2.90▲ 21%
$6.5K
Code reviewDebugging or reviewing code1,32988%$1.11▲ 5%
$1.3K
Data analysisData analysis1,16490%$1.19▼ 3%
$1.3K
ResearchResearch99393%$0.87▼ 12%
$0.8K
taskonomics
Tokens tell you what you spent. Tasks tell you what you got.
A token count can't tell you whether the work got done. Taskonomics groups AI usage into tasks, like writing code, reviewing PRs or running agents, and measures what each one costs to actually solve, by department, repository and initiative.
See your cost per taskAI cost by task · last 30 days
$14.5K across 11,059 solved tasks
TaskSuccessPer solved taskSpend
Writing or generating code91%$0.99▼ 8%
$3.6K
Agentic workflows79%$2.90▲ 21%
$6.5K
Debugging or reviewing code88%$1.11▲ 5%
$1.3K
Data analysis90%$1.19▼ 3%
$1.3K
Research93%$0.87▼ 12%
$0.8K
Summarizing content98%$0.51▼ 4%
$0.1K
ai integrations
Every AI tool your teams use, in one view.
Model providers, coding agents, gateways, clouds and data platforms. Connect them once and see every AI dollar side by side.
24 shown · all integrations
tokenomics and governance
One place for every AI bill.
See spend by model and identity, then set the guardrails that keep it in check.
Human vs agent spend
Split each user's spend between their own work and their agents.
Spending limits
Monthly limits per person, team, repo or initiative, with approvals.
Model governance
Allow or block models and providers for any group.
One AI gateway
Route AI traffic through one endpoint. No SDK changes.
Tokens per feature
Input, output and cached tokens, by feature and customer.
AI Hub
Every AI dollar, from model to task
Tokenomics shows spend by provider, model and user. Taskonomics shows what that spend got done, and what each solved task cost.
Total AI spend
$53.1K
last 30 days
Cost per solved task
$1.31
11,059 solved tasks
Active models
36
7 model developers
Agentic spend
22%
of AI spend
Daily AI spend
Spend by provider
Amazon Bedrock
$24.4K
Cursor
$16.5K
OpenAI
$7.4K
Anthropic
$4.8K
Tokenomics
Spend by model and user.
Taskonomics
Cost per task, team and repo.
Insights
Savings per task, ready to act on.
Governance
Limits and model rules.
how it works
From AI bill to cost per task
1
Connect your AI tools.
Add Claude, ChatGPT Enterprise, Cursor, Bedrock and more with connectors, or route traffic through the AI gateway.
2
Every request gets context.
Spend is tied to the user, team, repository and task behind it, so every AI dollar has an owner.
3
Act on cost per task.
Insights, spending limits and model rules keep AI spend where it pays off.
Frequently asked
questions
What is Taskonomics?
Taskonomics measures AI cost by the work it does. It groups usage into tasks such as writing code, debugging, agentic workflows or research, then shows runs, success rate, retries and the cost per solved task, by department, repository and initiative.
How is AI Hub different from a provider dashboard?
Provider dashboards show tokens and spend for one vendor. AI Hub brings every provider and AI tool into one view and adds context: who used it, for which task, and whether the task got solved.
Which AI tools does AI Hub support?
Sixteen connectors today, including Claude, ChatGPT Enterprise, Codex, Cursor, OpenAI, Amazon Bedrock, xAI, Databricks, Snowflake and MCP servers. You can also route traffic through the AI gateway.
Do I need to change my code?
No. Connectors read spend and usage from the tools you already use, and the AI gateway works without SDK changes.
You know what AI costs. Now know what it's worth.
See AI spend by team, user and task, down to every solved task.