GPT 6 Luna: the cheapest smart paid model

Introduction

GPT 6 Luna is the small member of the GPT 6 family (Astra, Sol, Luna) and the cheapest model in that line that still works as a full agent. At roughly ten cents per 1M input tokens it follows tools, keeps context, and survives long refactors — enough for overnight runs and CI jobs where every cent counts. It is the usual paid step-up from a free model: use it when a free OpenCode model is rate-limited, and reserve Claude Opus 5.5 for the hard final pass.

Pricing and ids

Endpoint family:

https://opencode.ai/inference/v1/responses

Context: Luna is cheaper than most Flash, Nano and Mini models at comparable quality as an agent. Verify current prices in the Console models docs before budgeting.

Where it runs

OpenCode Console serves Luna under the id above, with pay-as-you-go credits, and that is the route these guides use. It is also sold directly by its maker and by aggregators such as OpenRouter, each under its own id and price, so compare the row for the route you actually use. Prices, rate limits and retention terms differ per route.

Use it in OpenCode

Same Console key as the other Console models, then pin Luna as default or per agent in opencode.jsonc at the project root:

vi opencode.jsonc

{ "$schema": "https://opencode.ai/config.json", "model": "opencode/gpt-6-luna" }

One-shot review from CI with no TUI:

opencode run --standalone --model opencode/gpt-6-luna "Review this repository for correctness and summarize issues"

Skills and agents fit

Luna likes checklist skills: preconditions, diff-before-write, exact test command, stop conditions. Best matches from Five complete examples: db-migrate and git-release. In the aredel repo it owns page plus blocks plus verify. Agent side, it runs builder: edits allowed, tests allowed, push denied. Pair it with @test-writer for coverage on every feature.

Example workflow

Routine paid build after free triage:

With builder and @test-writer, add cursor pagination to orders with three tests, show git diff --stat and stop

Use the reviewer subagent with @api-review on git diff --cached

Luna does the implementation steps, then hands the staged diff to Claude Opus 5.5 for the verdict. For background work set the review command to subagent: true and keep typing in the parent.

Privacy and limits

Paid Console requests follow the provider data policy with limited retention for safety, unlike a free contributor tier that may train on prompts. Luna is therefore the default for customer code that the free OpenCode model must not see. Set monthly workspace limits and review opencode stats --cost --models weekly — Luna is cheap per token but long sessions still add up.

Troubleshooting

• Over 272K pricing jump: long sessions cross the threshold — enable compaction with a keep.tokens budget and commit mid-way.
• Weak planning: switch to Plan mode first, then back to Build; do not let Luna edit on the first prompt for hard features.
• CI cost creep: pin Luna for CI, keep Opus for manual final review only.

Official sources:

• opencode.ai/v2/docs
• opencode.ai/docs/console/models

Article author: Arthur Isaev