Coral Bricks

Coral Bricks serves GLM, Kimi, gpt-oss, and DeepSeek behind an OpenAI-compatible API at $1.12 per 1M input tokens on GLM 5.3, with free cached reads and 340 tok/s decode against 93 tok/s for Fireworks. For agencies running coding or research agents for clients, that throughput and cache economics turn repeated context into a fixed cost instead of a per-call tax. The catch is structural: this is developer infrastructure, not a white-label product, so the margin only lands when the client is building or operating their own agent system.

Coral Bricks prices GLM 5.3 at $1.12 per 1M input tokens and $1.68 per 1M cache write tokens, but cached reads are free. That asymmetry is the whole margin model for an agency running a client agent loop. A coding or research agent that re-sends a 200k-token repository context on every tool call pays full input price each turn without caching; with cache writes, the first pass costs $1.68 per 1M and every subsequent read costs nothing. Take the Coral Bricks Starter Agent at $2,500 with 20h setup: if the client's agent fires 40 turns per session over a stable context, the retainer only holds if you configure caching before launch. Model choice compounds it, since GLM Flash and DeepSeek Flash slugs carry different per-token rates. Audit token flow in week one, or the second month of delivery eats the first.

Inference Cost Pass-Through Ceiling is the point at which an agency can no longer absorb a model provider's price or latency change inside a fixed retainer, so the cost has to move to the client or the work has to shrink. The framework asks three questions per client engagement: what share of delivery cost is metered inference, how fast can that share be re-routed to a cheaper model, and what contract language lets you reprice. Forrester's 2027 predictions flag AI growth colliding with energy and infrastructure limits, which converts compute scarcity into API price movement on agency tools. A concrete case: an agency running document analysis on a frontier API can shift bulk classification to a smaller open-weight model served through Ollama or a gateway like Helicone, keeping the frontier model only for reasoning steps. That split is the ceiling defense.

Provider Substitution Window is the interval during which an agency can move a client workload from one model provider to another without rewriting prompts, evals, or integration code. The window is widest at the orchestration layer and narrowest at the fine-tuned weights layer: a gateway swap takes hours, a retrained model takes a quarter. Agencies that measure this window per client account know exactly when they hold pricing leverage and when a vendor holds it. Forrester's 2027 predictions flag compute and energy constraints pushing API pricing upward, which turns a wide substitution window into a margin defense rather than an engineering nicety. A concrete case: an agency routing Claude and GPT traffic through a gateway such as Helicone or Portkey can shift a client's summarization workload in an afternoon when one provider raises rates, while a competitor with hardcoded SDK calls absorbs the increase on a fixed retainer.

Adopt Coral Bricks only when the client owns an agent loop you can repoint at an OpenAI-compatible endpoint, and price the engagement on setup plus usage rather than a flat retainer.

Map every AI dependency in the delivery stack to a named provider, a fallback route, and a pass-through cost clause before quoting fixed-fee client work.

IF your agency is shipping or operating coding and research agents that need long contexts up to 1M tokens and tool calls, THEN Coral Bricks is worth adopting because its OpenAI-compatible API drops into OpenCode, Codex CLI, GitHub Copilot, and Cursor with minimal configuration, and free cached reads cut the cost of repeated context. IF your revenue depends on reselling a white-label client product, THEN skip it: Coral Bricks is an inference API, not a client-facing SaaS platform, and the $0 'Own your AI' tier is a contact-sales VPC commitment priced by resources rather than tokens.

A five to seven day playbook for agencies to stand up a client coding or research agent on Coral Bricks' OpenAI-compatible endpoint, using free cached reads and pay-per-token pricing to keep delivery costs predictable.