CostHawk tracks AI adoption and cost across engineering teams and providers. Plans start at $19/month with a 30-day free trial. Read the full review.
Researched and fact-checked by Bartu Cavusoglu · Updated October 26, 2026
CostHawk is an AI adoption monitoring platform for engineering teams. It tracks who on a team is using AI tools, what that usage costs, and how adoption is trending across providers such as OpenAI, Anthropic, and Google. The homepage frames the core question as whether AI spending is paying off, and the product answers it with one view instead of several separate billing dashboards.
The product is made by Outperform Ventures LLC, which does business as TrainMyAgent. Founder Chase Dillingham built it after his engineering team received AI API bills two or three times higher than expected, with no breakdown by model or project. CostHawk is aimed at engineering leads, CTOs, CPOs, and CFOs who need adoption and cost figures they can report to others.
CostHawk supports three integration methods. MCP telemetry captures local developer activity from tools such as Claude Code, Codex, Cursor, and Gemini CLI. Admin API sync pulls organization-wide cost data from provider accounts. Wrapped proxy keys sit between code and the provider, so real provider keys never touch the codebase or client devices.
The company describes its data handling as metadata only. It stores token counts, model names, timestamps, and hashed project IDs, and says it never stores prompts, responses, or code. MCP telemetry is computed locally before upload, dry-run syncs preview the payload, and automatic uploads stay off until enabled with COSTHAWK_AUTO_SYNC=true. Admin API keys are encrypted at rest with AES-256-GCM.
The adoption side includes an adoption heatmap, a per-engineer view, model preference breakdowns, and week-over-week consistency tracking. The sample dashboard on the homepage shows adoption rate, active engineers, average sessions per day, and team cost over the last 30 days. Team benchmarks rank groups on activity, consistency, cache efficiency, and breadth of adoption, and stay private to the organization.
The cost side includes budget alerts, a cost optimizer that looks for cheaper models, a savings engine that compares subscription and pay-per-use plans, and exports in CSV, JSON, or formatted reports. A REST API exposes adoption and cost data programmatically. Access to rankings is controlled by role-based access, and the public leaderboard uses aliases rather than real names, account names, or emails.
CostHawk is an AI adoption monitoring platform for engineering teams. It tracks who is using AI tools, how much they cost, and how adoption is trending across providers such as OpenAI, Anthropic, and Google.
Individual costs $19 per month for one person, and Team costs $99 per month for 2-100 people. Enterprise pricing is custom for organizations with 101 or more people. Billing is per seat, not per token or session.
Yes. Individual and Team include a 30-day free trial. A card is required at signup and is not charged until the trial ends, and users can cancel from the billing portal.
Claude Code, OpenAI Codex CLI, Cursor, and Gemini CLI connect through the CostHawk MCP. Providers include OpenAI, Anthropic, Google, xAI, Mistral, Cohere, DeepSeek, Groq, Meta, Perplexity, Together, AWS Bedrock, and Azure OpenAI.
According to the company, no. It stores usage metadata such as token counts, model names, timestamps, and hashed project IDs, and never stores prompts, responses, or code. MCP telemetry is computed locally before upload.
Benchmarks rank teams on activity, consistency, cache efficiency, and adoption breadth. They are scoped to the organization and protected by role-based access control. The public leaderboard uses aliases only, never real names, account names, or emails.
The site says setup takes under five minutes. Users sign up, invite the team, and connect through MCP telemetry or admin API keys, with no infrastructure changes.
It suits engineering leads, CTOs, CPOs, and CFOs who need one view of AI adoption and cost across teams and providers. Individual developers can also use the $19 plan for a personal dashboard.