OpenAI API Pricing (August 2026)

The headline rates apply to short-context requests. OpenAI publishes a separate long-context meter for the current flagship tiers: GPT-5.6 Sol rises from $5/$30 to $10/$45 per million input/output tokens, Terra from $2/$12 to $4/$18, and Luna from $0.20/$1.20 to $0.40/$1.80. Cached input doubles with the input meter, while the output uplift is 50%.

GPT-5.5 uses the same $10/$45 long-context rate as Sol. GPT-5.4 rises from $2.50/$15 to $5/$22.50. The calculator above uses short-context rates, so long prompts need the higher meter applied before a budget is approved. OpenAI also lists a 10% regional-processing uplift for eligible models released on or after March 5, 2026.

Cached input is automatic: repeated prompt prefixes (system prompts, tool definitions, shared context) bill at 10% of the standard input rate — $0.50/M instead of $5/M on GPT-5.6 Sol. On GPT-5.6 the accounting got more predictable: cache writes cost 1.25x standard input and cached content persists for at least 30 minutes, so any prefix reused more than once nets out cheaper.

The Batch API takes a flat 50% off input and output for asynchronous jobs, and GPT-5.5 additionally offers Flex at half rate and Priority at 2.5x. Stacked, a cached and batched Sol workload can run at roughly a quarter of list price. Model your own mix with the LLM API pricing calculator or the estimator above.

An OpenAI API key has no monthly subscription fee. Usage is billed from the token, tool, storage, and processing meters attached to the API account. ChatGPT Free, Go, Plus, Pro, Business, and Enterprise are separate products with per-user or plan pricing; paying for ChatGPT does not create a pool of API credits.

Use this page for application and agent workloads billed through an API key. For the consumer-plan decision, see the current ChatGPT Plus analysis. For hosted OpenAI models with Azure routing and residency controls, compare the Azure LLM pricing table.

Start cheap and escalate on measured quality gaps. GPT-5.6 Luna ($0.20/$1.20) and the GPT-5.4 nano/mini tiers cover classification, extraction, and routing. GPT-5.6 Terra ($2/$12) is the production default, with the newer family's capabilities. GPT-5.6 Sol and GPT-5.5 ($5/$30) are the flagship tier for frontier coding, reasoning, and agentic work. GPT-5.5 Pro ($30/$180) is a 6x jump that only makes sense when you have evaluation results proving the flagship tier falls short.

Older families stay useful: OpenAI keeps GPT-5.1 ($1.25/$10) and the GPT-5.4 family on the live price sheet, and if your evals pass on them there is no reason to pay 2026-flagship rates.

The flagship tier has converged: GPT-5.6 Sol and GPT-5.5 at $5/$30 sit beside Claude Opus 4.8 at $5/$25 — identical input, Anthropic 17% cheaper on output. Mid-tier, GPT-5.6 Terra ($2/$12) undercuts Claude Sonnet 5's standard $3/$15 (though Sonnet's $2/$10 intro pricing is cheaper on output through August 31). Gemini 3.1 Pro now matches Terra at $2/$12, and at the very top Claude Fable 5 ($10/$50) costs a third of GPT-5.5 Pro ($30/$180).

With prices this close, benchmarks decide it: see the table above next to the Claude API pricing hub, or the head-to-head Opus 4.8 vs GPT-5.6 Sol comparison on the compare hub.