Context regrowth

LLM APIs are stateless: each call pays for every token of history again. A conversation that adds a constant amount per turn therefore costs the sum of an arithmetic series — turn twenty re-sends the previous nineteen. Agents make this worse than chat, because tool loops add turns nobody reads.

As arithmetic, not a benchmark: a session where each turn adds about 500 tokens has re-sent roughly 105,000 input tokens by turn twenty, for about 10,000 tokens of actual new content — a ten-fold amplification that grows with every further turn.

By conversation turn counts and context-window utilisation per call — both of which Capsera's prompt analysis extracts without storing content. A cost-per-turn curve that slopes upward on a flat workload is the signature.

What is context regrowth?

Context regrowth is the compounding cost of re-sending an entire conversation history with every turn, so total input tokens grow roughly with the square of the conversation length. LLM APIs are stateless, so each call pays for every token of history again. Turn twenty re-sends the previous nineteen.

• Tool result bloat — Tool result bloat is the cost incurred when large tool outputs — full API payloads, whole files, verbose logs — are inserted into an agent's context and then re-sent on every subsequent turn.
• Prompt prefix stability — Prompt prefix stability is the property that the opening bytes of a prompt are identical across calls, which is what allows a provider's prompt cache to match and discount repeated input.
• Multi-agent amplification — Multi-agent amplification is the compounding of LLM costs that occurs when agents invoke other agents, so one task multiplies into a tree of calls whose total is far larger than any node suggests.