Model Comparison

GPT-5.6 Sol vs Terra vs Luna: Which Tier for Which Work

Three tiers, one generation, and price gaps that are much larger than the benchmark gaps. Here is the verified data on OpenAI's GPT-5.6 family, and a decision rule you can apply per task instead of per project.

By Kylian Migot · Updated July 2026 · 7 min read

Quick answer

Default to Terra ($2.5 in / $15 out per MTok), the workhorse with roughly GPT-5.5 performance at about half the price. Escalate to Sol ($5 in / $30 out per MTok) for hard problems and long runs, and when you want its exclusive max reasoning effort and ultra mode. Drop to Luna ($1 in / $6 out per MTok) for cheap, mechanical, high-volume work.
GPT-5.6 Sol price
$5 in / $30 out per MTok
GPT-5.6 Terra price
$2.5 in / $15 out per MTok
GPT-5.6 Luna price
$1 in / $6 out per MTok
Terminal-Bench 2.1
Sol 88.8% › Terra 87.4% › Luna 84.7% (OpenAI launch materials)
01

First, the Naming Scheme

OpenAI's GPT-5.6 launch on July 9, 2026 came with a naming scheme that finally makes the lineup legible: the number is the generation, the name is the capability tier. GPT-5.6 is the generation; Sol, Terra, and Luna are the tiers within it, flagship, workhorse, and budget/fast respectively. The API identifiers follow the same pattern: gpt-5.6-sol, gpt-5.6-terra, gpt-5.6-luna.

That means this is not a comparison of three different models so much as three price points on one generation. All three went generally available the same day and all three are selectable in Codex. The question is not “which is best”, Sol is, by construction, but which one each unit of work deserves. Single-model deep dives: Sol, Terra, and Luna.

02

Head-to-Head: The Verified Data

GPT-5.6 SolGPT-5.6 TerraGPT-5.6 Luna
RoleFlagshipWorkhorseBudget / fast
Price (per MTok)$5 in / $30 out per MTok$2.5 in / $15 out per MTok$1 in / $6 out per MTok
Terminal-Bench 2.188.8% (91.9% in ultra mode)87.4%84.7%
Max reasoning effortYesNoNo
Ultra mode (parallel subagents)YesNoNo
ReleasedJuly 9, 2026July 9, 2026July 9, 2026

Data verified July 18, 2026 against OpenAI's published pricing and launch materials. Terminal-Bench 2.1 is the one benchmark OpenAI published across all three tiers, which makes it the family's only like-for-like yardstick, and worth reading carefully, next.

03

Price vs Benchmark: Where the Gaps Actually Are

Put the two ladders side by side and the family's economics jump out. From Terra to Sol, the benchmark gap is 1.4 points (87.4% to 88.8%) and the price gap is 2x ($2.5 to $5 per MTok in, $15 to $30 out). From Luna to Terra, the benchmark gap is 2.7 points (84.7% to 87.4%) for a 2.5x price gap ($1 to $2.5 in, $6 to $15 out).

Read that way, the Luna-to-Terra step buys nearly twice the benchmark improvement of the Terra-to-Sol step. That is why Terra is the default: it sits at the top of the steep part of the curve, where each dollar still buys measurable capability, while Sol's premium pays for the last 1.4 points plus its exclusive features. How this slots into the wider Codex lineup, including GPT-5.5, is covered in the best Codex model guide.

Ship your first agent today

Download AIDEN free and point it at your existing Claude Code or Codex setup. No credit card, running in minutes.

Download AIDEN free

Free to start · macOS 12+ · No credit card required

04

What Only Sol Can Do

The benchmark gap understates the tier gap in one respect: two capabilities exist only on Sol, at any price. Max reasoning effort is the family's highest thinking budget, for problems where you want the model to grind. Ultra mode runs native parallel subagents, Sol fans a task out to multiple copies of itself working concurrently, and per OpenAI's launch materials it lifts Sol's Terminal-Bench 2.1 score from 88.8% to 91.9% in ultra mode.

If your workload needs either, the tier question answers itself, Terra and Luna are not discounted versions of these features, they simply do not have them. If your workload does not need them, the 1.4-point benchmark gap is the honest measure of what Sol's 2x premium buys.

05

The Decision Rule, Step by Step

  1. 1

    Start every task on Terra

    The workhorse default: roughly GPT-5.5 performance at about half the price ($2.5 in / $15 out per MTok), and within 1.4 points of the flagship on the family's only shared benchmark. Most Codex sessions should begin and end here.
  2. 2

    Drop to Luna when the task is mechanical

    Lint fixes, renames, formulaic edits, high-volume batch work: if the spec fully determines the diff, Luna's $1 in / $6 out per MTok does the job at 2.5x below Terra's price. Never send it work hard enough to fail, a failed cheap run plus a re-run costs more than Terra succeeding once.
  3. 3

    Escalate to Sol when failure is expensive

    Hard cross-cutting problems, long agentic runs, terminal-heavy chains, and anything that wants max reasoning effort or ultra mode. At $5 in / $30 out per MTok the premium is real, and worth it exactly when a Terra failure would cost more than the price difference.
  4. 4

    Let outcomes tune the routing

    Track which tier each kind of task actually succeeds on. If Terra keeps failing a category, that category is Sol work; if Luna never fails one, Terra was overspend. The rule improves with every merged diff.
06

Routing Per Story, Not Per Project

The tier decision is cheapest when it is made per unit of work rather than once per project, and that is an orchestration problem more than a model problem. AIDEN's approach: your Codex CLI runs on a kanban board where every story is a card with its own model, its own branch, and its own diff, so the mechanical cards run Luna, the defaults run Terra, the hard ones run Sol, and a wrong routing costs you one card, not a sprint. The same board also runs Claude Code side by side, if the answer to “which tier” turns out to be “a different vendor for this story”, the full field is at the models hub.

FAQ

Which GPT-5.6 model should I use?
Default to Terra ($2.5 in / $15 out per MTok): roughly GPT-5.5 performance at about half the price, and only 1.4 points behind Sol on Terminal-Bench 2.1. Escalate to Sol ($5 in / $30 out per MTok) for hard problems, long agentic runs, and when you want max reasoning effort or ultra mode, both Sol-only. Drop to Luna ($1 in / $6 out per MTok) for cheap, mechanical, high-volume work.
Is Sol worth 2x Terra's price?
On the published benchmark alone, usually not: Sol's Terminal-Bench 2.1 lead over Terra is 1.4 points (88.8% vs 87.4%) for a 2x price gap. What can justify the premium is the exclusives, max reasoning effort and ultra mode, and the economics of failure: on tasks hard enough that Terra might fail, one Sol run succeeding is cheaper than a failed Terra run plus a re-run.
What is ultra mode?
Ultra mode is a Sol-exclusive execution mode that runs native parallel subagents: Sol fans a task out to multiple copies of itself working concurrently. Per OpenAI's launch materials, it lifts Sol's Terminal-Bench 2.1 score from 88.8% to 91.9% in ultra mode. Terra and Luna do not have it at any price.
Can I switch tiers mid-project?
Yes. All three tiers are selectable in Codex within the current generation, and nothing binds a project to one tier; the practical pattern is choosing per task rather than per project. AIDEN makes that routing explicit: each story on its kanban board runs its own model on its own branch, so mechanical cards go to Luna, default work to Terra, and hard stories to Sol within the same project.
Do the GPT-5.6 tiers share a context window?
We don't know, honestly: OpenAI's context-window specs for the GPT-5.6 tiers are not in our verified data, and we do not publish numbers we have not verified. What we can say is that all three tiers are selectable in Codex as the current generation. If context size is the deciding factor for your workload, check OpenAI's current model documentation directly.

Keep reading

Route every story to the right tier.

AIDEN puts your Codex CLI on a kanban board: Luna on the mechanical cards, Terra on the defaults, Sol on the hard ones, each story on its own branch. Free for one project.

macOS 12+ · Bring your own Claude Code or Codex · Your code stays local