GPT-5.6Updated 2026-08

GPT-5.6 API: Sol vs Terra vs Luna — Pricing and How to Choose (2026)

GPT-5.6 ships as three tiers. All at 90% of list on cocodot, metered, Alipay or card. Real prices, the surprising Sol-vs-5.5 comparison, and a tiering strategy that halves bills.

TL;DR: GPT-5.6 is a family of three tiers: Sol (flagship reasoning/code/research), Terra (balanced daily driver), Luna (high-throughput, low-latency, cheapest). cocodot serves all three at 90% of list: Sol $3.6/$18 per M tokens (inputs >272K: $7.2/$27), Terra $1.8/$10.8, Luna $0.18/$1.08 — the cheapest GPT on the menu. One fact worth knowing: after OpenAI's Aug-25 price cut, Sol's output price is below last-gen GPT-5.5 ($18 vs $27) — the newer flagship is the cheaper one. Tier your workload: Luna for volume chores, Terra for everyday work, Sol for the hard 20%. Same key, switch by model name.

One generation, three compute tiers

GPT-5.6 is a family, not a single model. Sol targets the hardest reasoning, coding and research work; Terra balances capability and cost for everyday use; Luna is built for speed and volume. All three carry 1050K context and tool calling. The right question is not which is strongest — it is which tasks deserve which tier's price.

The counterintuitive fact: the new flagship is cheaper

After OpenAI's Aug 25 repricing, Sol lists at $4/$20 and cocodot serves it at $3.6/$18 — output at two-thirds of last-gen GPT-5.5's $27. Unless you need version pinning, switching mog-6 to mog-8-s is free money. 5.5 stays available; no forced migration.

Luna, the underrated volume engine

At $0.18 per million input tokens, Luna is the cheapest GPT here. Batch content processing, classification, summarization, support bots, the chore nodes of an agent graph. The $0.5 signup credit runs roughly 1.2M tokens of mixed traffic on Luna — enough to trial your whole pipeline before paying.

Two-minute setup

OpenAI-compatible: base_url https://cocodot.co/api/ai/v1, model mog-8-s / mog-8-t / mog-8-l (official names auto-route). Alipay or card, metered, no monthly fee. Long-context tiers are fully listed above — no hidden surcharge past 272K.

GPT-5.6 tiers + GPT-5.5 on cocodot (USD per M tokens, 90% of list)

Model	ID	In / Out	Role
GPT-5.6 Sol	mog-8-s	$3.6 / $18 (>272K: $7.2/$27)	Flagship: reasoning, code, research
GPT-5.6 Terra	mog-8-t	$1.8 / $10.8 (>272K: $3.6/$16.2)	Balanced daily driver
GPT-5.6 Luna	mog-8-l	$0.18 / $1.08 (>272K: $0.36/$1.62)	Volume: high-frequency, cost-sensitive
GPT-5.5	mog-6	$4.5 / $27 (>272K: $9/$40.5)	Previous flagship, still available

FAQ

Real models or clones?

Official relay via licensed cloud vendors. Verify with our open-source checker: probe.cocodot.co — including against us.

What's missing versus OpenAI direct?

Honestly: Prompt caching does work here — cached input is billed at half the input rate, and usage returns the cache token counts so you can verify it yourself; hit rate depends on how stable your prefix is, and we do not promise a number. Plain text/code calls are indistinguishable.

Do prices track official?

Anchored at 90% of list; when OpenAI moves, we move (we matched the Aug-25 cut same day). Live prices at /pricing.

Can I call GPT-6.1 Sol through this endpoint?

Not at the moment — this relay does not carry the GPT-6 family yet (Astra, Sol, 6.1 Sol, Luna). GPT-6.1 Sol, released by OpenAI on September 29, 2026, can be called through OpenAI's own API as gpt-6.1-sol; official pricing and how to pay from China are at /hub/gpt-61-sol-api-china, and the main DevDay 2026 announcements are summarised at /hub/openai-devday-2026.