$0.60 per session, just on orientation. Here is what my AI agent was doing before writing any code.
agents autonomous claude openai
| Source: Dev.to | Original article
A recent experiment by developer Glincker shows that the “orientation” phase of an AI‑driven coding assistant can already cost around $0.60 per session, even before a single line of code is produced. Using the open‑source StackLit framework (github.com/glincker/stacklit) and the one‑liner `npx stacklit init`, the author logged token consumption from the underlying language model and translated those figures into real‑world electricity usage and cloud‑provider pricing. The data reveal that the initial prompt‑parsing, environment‑setup, and context‑gathering steps consume roughly 1.2 kWh of compute, translating to the quoted per‑session fee.
Why this matters is twofold. First, developers and enterprises are increasingly relying on autonomous coding agents to accelerate software delivery, yet most pricing models only expose the cost of the final output. Hidden “orientation” expenses can quickly erode the economic advantage, especially at scale. Second, the energy footprint of these preparatory steps adds to the growing carbon impact of AI‑augmented development, a concern echoed in recent analyses of AI coding agents’ electricity use. By quantifying the pre‑coding overhead, Glincker’s work pushes the conversation beyond headline‑grabbing token counts toward a fuller accounting of both financial and environmental costs.
What to watch next is how platform providers respond. OpenAI’s Agents SDK and similar toolkits are already being packaged with more granular metering, and competitors may introduce tiered pricing that separates orientation from execution. At the same time, the community is likely to see optimisation efforts—lighter prompt engineering, caching of environment data, and hybrid on‑device inference—to shave off wasted compute. Regulators and sustainability auditors may soon demand transparent reporting of AI‑agent energy use, turning these early cost disclosures into a baseline for industry standards.
Sources
Back to AIPULSEN