Price tracker · as of 2026-08-02

GPT-5.6 Luna Price Tracker

OpenAI cut Luna 80% and Terra 20% on 2026-07-30. This page tracks official short-context API rates and what changed for Codex usage counting.

#Luna $0.20/$1.20 #Terra $2/$12 #Sol $5/$30 #OpenAI API

Current short-context API rates (OpenAI direct)

As of 2026-08-02: Luna $0.20 input / $1.20 output per 1M tokens; Terra $2 / $12; Sol $5 / $30. Cached input is typically 10% of uncached input. Prices can change — re-check OpenAI pricing before large runs.

Last reviewed 2026-08-02 against OpenAI docs and LiteLLM openai keys.

Family price table

Short-context standard rates per 1M tokens. Long-context requests bill higher (Luna long-context about $0.40 / $1.80).

Tier	Input	Output	Cached input	Notes
gpt-5.6-luna	$0.20	$1.20	$0.02	80% cut on 2026-07-30 (was $1 / $6)
gpt-5.6-terra	$2.00	$12.00	$0.20	20% cut on 2026-07-30 (was $2.50 / $15)
gpt-5.6-sol	$5.00	$30.00	$0.50	List price unchanged; Fast mode about 2.5x speed at 2x price

Luna details worth budgeting for

Context and long prompts

API context window is about 1.05M tokens. Prompts over 272K input tokens are billed at higher long-context rates for the full request (Luna about $0.40 input / $1.80 output per 1M).

Subscriptions and Codex

Plan prices and overall quotas did not drop. OpenAI states the lower Luna and Terra rates are reflected in how usage is counted against paid ChatGPT Work and Codex subscriptions — the same quota can cover more Luna/Terra work.

LiteLLM note

LiteLLM openai keys for gpt-5.6-luna already show $0.20/$1.20 (checked 2026-08-02). Some azure/* entries may still list older rates — this page uses OpenAI API direct rates.

Routing suggestion (cost, not a quality ranking)

Match tier to the job. These are practical defaults, not benchmark claims.

Luna — high-volume execution

Bulk tool calls, well-specified edits, tests, classification, and cheap agent steps after a plan exists.

Terra — everyday coding default

Balanced day-to-day coding and agent work when you want better quality than Luna without Sol spend.

Sol — planning and hard cases

Ambiguous design, hard bugs, and high-stakes plans. Use Sol Fast when latency matters and 2x list price is acceptable.

Price timeline

Filled events first. Empty slots stay hidden until new i18n keys are added.

2026-07-09

GPT-5.6 family general availability

Sol / Terra / Luna GA. Launch short-context list prices: Sol $5/$30, Terra $2.50/$15, Luna $1/$6 per 1M tokens.

2026-07-30

Luna −80%, Terra −20%

OpenAI announced lower Luna and Terra prices effective that day: Luna $0.20/$1.20, Terra $2/$12. Sol list price unchanged; Sol Fast mode introduced in the API.

2026-08-02

Tracker page published on QCode

Rates above re-checked against OpenAI model/pricing docs and LiteLLM openai keys. Future cuts can be appended as new timeline slots without a template change.

Primary sources

OpenAI blog Advancing the price-performance frontier with GPT-5.6 (2026-07-30); developers.openai.com pricing and gpt-5.6-luna model card; LiteLLM model_prices_and_context_window.json (openai keys, 2026-08-02).

FAQ

What is the current GPT-5.6 Luna price?

As of 2026-08-02, OpenAI short-context API rates list Luna at $0.20 per million input tokens and $1.20 per million output tokens, with cached input about $0.02.

Did Sol get cheaper too?

No. Sol stayed at $5/$30 list. OpenAI introduced Fast mode for Sol at about 2x price for up to about 2.5x speed, without claiming higher intelligence.

Did Codex subscription prices fall?

OpenAI did not lower subscription sticker prices in the July 30 note. It said lower Luna and Terra rates are reflected in how usage is counted against paid Codex and ChatGPT Work plans.

Can I call Luna through QCode?

Yes. QCode exposes OpenAI-compatible access so you can select gpt-5.6-luna alongside Terra and Sol on one key. Always confirm the live rate on the models page before large jobs.

Create a QCode account

Plans from ¥60 / $8.57. One key across Claude, GPT, and Gemini models.

Create account Browse models View pricing