Service Tiers: Standard, Batch, Flex & Fast

GPT-5.6 Sol	Standard $5.00 · Batch $2.50 · Fast $10.00 /MTok input
GPT-5.6 Terra	Standard $2.00 · Batch $1.00 · Fast $4.00 /MTok input
GPT-5.6 Luna	Standard $0.200 · Batch $0.100 · Fast $0.400 /MTok input
GPT-5.5	Standard $5.00 · Batch $2.50 · Fast $12.50 /MTok input
GPT-5.5 Pro	Standard $30.00 · Batch $15.00 · Fast not offered /MTok input
GPT-5.4	Standard $2.50 · Batch $1.25 · Fast $5.00 /MTok input
GPT-5.4 Mini	Standard $0.750 · Batch $0.375 · Fast $1.50 /MTok input
GPT-5.4 Nano	Standard $0.200 · Batch $0.100 · Fast not offered /MTok input
GPT-5.4 Pro	Standard $30.00 · Batch $15.00 · Fast not offered /MTok input
GPT-5.3 Codex	Standard $1.75 · Batch — · Fast $3.50 /MTok input

OpenAI publishes four separate price tables per model, not one: Standard, Batch, Flex, and Fast. They aren't four unrelated products — they're the same underlying model, priced differently for different delivery guarantees.

The shape of it

Standard is the default synchronous API call, at the reference price every other tier is measured against. Batch and Flex are priced identically to each other for every text model that offers both — genuinely the same price, published as two separate tables, which is itself worth knowing rather than assuming a typo. What differs between them is delivery semantics, not cost: Batch guarantees a fixed completion window and runs against its own separate rate-limit pool with no output-token limit; Flex doesn't carry that same guarantee. Fast mode (renamed from "Priority" processing) costs a real premium over Standard in exchange for lower latency, and — this is the detail worth checking before you plan around it — Fast isn't offered for every model at all.

Picking a tier

If your workload can tolerate a delay in exchange for a lower price and doesn't need a response in the same request cycle, Batch is usually the right default — the discount is real and the fixed window is generous for most asynchronous work. If you need a synchronous response but latency isn't the binding constraint, Standard is the right default. Fast mode earns its premium specifically when latency itself has a cost — a user waiting on a response, or a pipeline step that's on the critical path.

Batch versus Flex, specifically

Because the two are priced identically, the choice between them is entirely about the guarantee, not the bill. Batch commits to a fixed completion window and draws from its own separate rate-limit pool, which is exactly what you want for a large, predictable, off-peak job — a nightly backfill, a scheduled report generation run, anything where "done by tomorrow morning" is an acceptable and useful promise. Flex doesn't carry that same fixed-window guarantee, which makes it a better fit when you want asynchronous, discounted pricing but can't commit to a rigid completion deadline up front. Neither is more "correct" than the other; they're the same price for two different operational shapes.

What the table below shows

Every model this site prices, with its Standard, Batch and Fast rates side by side (Flex omitted from the table since it matches Batch exactly for every model listed) — computed directly from the facts module, not retyped by hand.