GPT-5.6 Cyber
Last verified September 5, 2026 · OpenAI pricing ↗$12.50/1M input · $75.00/1M output · 400K context · OpenAI
Count Tokens
—
Tokens
—
Input Cost
—
Output Cost
Estimate Monthly Cost
Monthly Cost Estimator
Quick:
<$0.0001/mo
Pick a preset above or enter custom usage
Alternatives to GPT-5.6 Cyber
Pricing Details
OpenAISecurity-specialised tier of the GPT-5.6 family, announced 2026-08-10 alongside OpenAI's Daybreak program. Gated: requires enrollment in Trusted Access for Cyber, and on AWS a second access request through your account team. Priced at exactly 2.5x GPT-5.6 Sol on every line: $12.50 input, $1.25 cached input, $75.00 output, and a 30-minute cache write at $15.625 (1.25x input, the same ratio Sol and the rest of the family use). STANDARD TIER ONLY: OpenAI publishes no Batch, Flex or Fast rows for this model, which makes it the only GPT-5.6 tier with no cheaper service tier available. CONTEXT WINDOW IS CONTESTED between two primary sources: OpenAI's model page states 400,000 tokens with 128,000 max output (the figure stored here, since this entry's sourceUrl is OpenAI's own page), while the AWS Bedrock model card states 272K and files its single pricing table under a heading reading Short Context Window (272K). 272,000 is the long-context billing threshold on the rest of the GPT-5.6 family rather than a capacity limit, so AWS appears to have printed a billing boundary in a capability field; AWS publishes no long-context rate for Cyber at all. On Amazon Bedrock this model is In-Region only in us-east-2, served on the bedrock-mantle endpoint under model ID openai.gpt-5.6-cyber, launched 2026-08-12, at $13.75 / $1.375 / $17.1875 / $82.50, an exact 1.1000x on all four lines with Geo and Global cross-region inference both marked unsupported. No Bedrock batch or provisioned throughput exists for any GPT-5.6 model. Daybreak Blue is a separate Bedrock card (openai.gpt-daybreak-blue-5.6-sol) carrying GPT-5.6 Sol with relaxed cyber guardrails at Sol's In-Region rate and no Global row. Reported capability gap, secondhand via press rather than OpenAI's own page (which returned 403 on 2026-08-18): 95% completion on a sensitive-request set against 1.5% for general-access Sol. Rates confirmed on OpenAI's own pricing page (verified 2026-08-18). DAYBREAK ALIASES, verified 2026-09-05: OpenAI's docs state that gpt-daybreak-blue-latest and gpt-daybreak-red-latest 'are aliases that currently point to gpt-5.6-sol and gpt-5.6-cyber, respectively'. So Daybreak Red IS this model, not a distinct model sharing a price, and Daybreak Blue IS gpt-5.6-sol ($4.00/$0.40/$20.00, 1,050,000 context). Both Daybreak model pages are live but publish NO rate card - only per-tier rate limits (500 RPM / 500K TPM at Tier 1 up to 15,000 RPM / 40M TPM at Tier 5) and a link to central pricing; searching the pricing page for 'daybreak' returns no row. The un-prefixed URLs without 'gpt-' 404. OpenAI adds that 'as new models are released through the Daybreak program, these aliases will be updated to point to the latest models, with pricing adjusted to match each underlying model', so a Daybreak model string is stable while its rate is not. NO DISCOUNT TIERS AT ALL: this model supports only the Responses API (v1/responses); Chat Completions, Realtime, Assistants and Batch are unavailable, and OpenAI publishes no batch, flex or fast rate for it. That makes it the only gated model in the September 2026 cyber cohort that loses its discounts as well as its availability - Anthropic's gated claude-mythos-5-1 keeps the full 50% batch discount. On a 3:1 input:output blend Red is $28.125/M against Mythos 5.1's $20.00/M standard (1.406x) but $10.00/M batched (2.813x), so the missing tier is worth more than the headline rate gap. Knowledge cutoff 2026-02-16. OpenAI's 2026-09-03 announcement commits $1B in SUBSIDISED ACCESS (its words; not 'credits' - the only credits named are a separate up-to-$1M no-cost API credit offer to states and utilities after the US water system attacks) covering access, training, technical support and partnerships, starting with the United States and 'targeting it to be consumed over the next six months', with thousands of defenders across 2,000 approved organisations already using Daybreak. At a 3:1 blend that $1B converts to about 125 trillion tokens on Blue and 35.6 trillion on Red, a 3.516x swing decided purely by which alias an organisation is approved for. Daybreak does not include GPT-6 Astra: the aliases resolve to 5.6-generation models. Not on OpenRouter (431-model catalogue pulled 2026-09-05 has zero rows matching cyber or daybreak).
Input / 1M tokens
$12.5
Output / 1M tokens
$75
Context Window
400K
Max Output
128K
Price History
No price change recorded since 2026-08-18, when this site started tracking this model; it launched on 2026-08-10.
Frequently Asked Questions
Common questions about GPT-5.6 Cyber pricing and usage
Read More About GPT-5.6 Cyber
OpenAI put a price on GPT-Rosalind for the first time: $5, $0.50 and $25, which is Claude Opus 5's card to the cent and $5 under the GPT-5.5 it is built on. The meter does not start until October 5, and there is still no page saying how much context the money buys.
September 15, 2026 · 12 min read
Three labs shipped a security-gated model in three days. Anthropic published a full rate card for one its own cyber programme cannot sell you yet, Google published no number at all, and OpenAI published a pointer.
September 5, 2026 · 12 min read