Skip to content

OpenAI's Codex Security CLI and TypeScript SDK for finding, validating, and fixing security vulnerabilities. npm: https://www.npmjs.com/package/@openai/codex-security

fix(sdk): price OpenAI models by Bedrock region and inference route (#2362)

Michael D'Angelo31a42e9ba74a
diff --git a/docs/bedrock.md b/docs/bedrock.md
index 57432c90e..5d52a3cb8 100644
--- a/docs/bedrock.md
+++ b/docs/bedrock.md
@@ -93,7 +93,8 @@ Use the exact Bedrock identifier for the model your account can invoke:
| Daybreak Blue | `openai.gpt-daybreak-blue-5.6-sol` | [Blue](https://docs.aws.amazon.com/en_en/bedrock/latest/userguide/model-card-openai-gpt-daybreak-blue-56-sol.html) |
| Daybreak Red | `openai.gpt-5.6-cyber` | [Red](https://docs.aws.amazon.com/en_en/bedrock/latest/userguide/model-card-openai-gpt-56-cyber.html) |
-AWS lists `us-east-2` and the Bedrock Mantle Responses API for both models.
+AWS lists the Bedrock Mantle Responses API in `us-east-2` for both models.
+Red is also available in `us-gov-east-1`.
The native provider handles that endpoint; no custom base URL or connector is
needed. Access requires OpenAI Trusted Access for Cyber enrollment and approval,
followed by AWS provisioning through your account team. Red requires separate
@@ -117,12 +118,47 @@ npx @openai/codex-security scan /path/to/repository \
--provider amazon-bedrock --model openai.gpt-5.6-cyber --effort high
```
-Cost estimates and `--max-cost` recognize both exact Bedrock IDs using their
-AWS model cards' Standard commercial in-region rates, including the AWS fee.
-Blue reports a short/long-context range because aggregate usage does not identify
-each request's context tier. Red's card lists a 272K context window and only
-short-context pricing, so its upper estimate remains unavailable. Spending limits
-use the short-context baseline. AWS bills the actual usage.
+## Scan cost estimates
+
+Estimates and `--max-cost` use the published Standard token rates for the exact
+Bedrock model ID and inference route. For example, `openai.gpt-6.1-sol` and
+`us.openai.gpt-6.1-sol` cost $2.20 per million short-context input tokens.
+`global.openai.gpt-6.1-sol` costs $2.00. All three prices already include any
+applicable AWS fee. See the [AWS model cards](https://docs.aws.amazon.com/bedrock/latest/userguide/model-cards-openai.html).
+
+Live scans use the region in an explicit AWS Bedrock endpoint. Without an
+endpoint override, they use the selected provider's `aws.region`, then
+`AWS_REGION`, then `AWS_DEFAULT_REGION`. If the native provider obtains
+its region from an AWS profile or managed credentials, set one of those values
+to that same region for cost estimates and `--max-cost`. For a Red scan with
+`AWS_REGION=us-gov-east-1`, the short-context input rate is $16.875 per million
+tokens. The same model in Ohio costs $13.75. Open-weight GPT models can also
+have different prices between commercial regions. Each estimate records the
+price source and the rates used. When estimating from token counts in the SDK,
+pass the region as the third argument, for example
+`estimateScanCost("openai.gpt-5.6-cyber", usage, "us-gov-east-1")`.
+With only a frontier model ID and usage, the SDK uses the commercial
+in-region price for an unprefixed route.
+
+The frontier model cards and the AWS Price List give different GovCloud West
+prices for GPT-5.4, GPT-5.6 Terra, and GPT-5.6 Luna. Estimates use the
+model-card rates for those three. The GPT-5.4 GovCloud East estimate comes
+from the [regional price list](https://pricing.us-east-1.amazonaws.com/offers/v1.0/aws/AmazonBedrock/current/us-gov-east-1/index.json).
+
+For the open-weight models, estimates use regional Standard SKUs where available.
+For example, the GPT OSS Safeguard 120B output SKU in Tokyo and São Paulo is
+$0.73 per million tokens. The [Bedrock pricing page](https://aws.amazon.com/bedrock/pricing/#OpenAI)
+groups both regions at $0.71. AWS bills actual usage.
+
+Usage is reported across the scan, so the estimate gives a range where AWS
+publishes both short- and long-context rates. Spending limits use the
+short-context baseline. For Red, AWS publishes only short-context rates and
+the upper estimate stays unavailable. A model or route without published rates
+has no cost estimate and cannot be used with `--max-cost`. For example, AWS
+lists global inference profiles for GPT-5.4 and GPT-5.5 but gives no exact
+Standard rate for either global profile. The GPT OSS and Safeguard cards also
+list Melbourne, Seoul, and Singapore. The regional price list has no Standard
+token rates for those models in those three regions.
<a id="repeatable-live-smoke-test"></a>
diff --git a/sdk/typescript/docs/cli.md b/sdk/typescript/docs/cli.md
index 5e0288ee1..64c7361a9 100644
--- a/sdk/typescript/docs/cli.md
+++ b/sdk/typescript/docs/cli.md
@@ -1140,10 +1140,12 @@ their own timestamps, window durations, and reset times. These are historical
observations, not a live allowance check or scan-only consumption. Empty readings
mean allowance was unavailable. There is no token-to-allowance conversion here.
-Cost displays use [standard API prices](https://developers.openai.com/api/docs/pricing)
-and show a short/long-context range. They estimate model usage, excluding other
-tiers, fees, surcharges, and account-specific pricing. They do not measure ChatGPT
-subscription allowance. Models without known prices have no estimate.
+Cost displays use Standard token rates for the model. OpenAI models use
+[OpenAI API pricing](https://developers.openai.com/api/docs/pricing). OpenAI
+models on Bedrock use [AWS pricing for the model, inference route, and region](https://github.com/openai/codex-security/blob/main/docs/bedrock.md#scan-cost-estimates).
+The short/long-context range estimates usage at published rates. Other service
+tiers, account discounts, and ChatGPT subscription allowance are outside that
+estimate. Models without known prices have no estimate.
JSON, history, and bulk receipts include `cost.estimatedUsdRange` (`min`, `max`,
and `context: "unknown"`). A null maximum means no upper estimate is available.
diff --git a/sdk/typescript/scripts/package-dist-files.mjs b/sdk/typescript/scripts/package-dist-files.mjs
index 30e3ffa2e..064f43def 100644
--- a/sdk/typescript/scripts/package-dist-files.mjs
+++ b/sdk/typescript/scripts/package-dist-files.mjs
@@ -77,6 +77,8 @@ const distModules = [
"contract",
"cost",
"cost-model",
+ "bedrock-pricing",
+ "bedrock-oss-pricing",
"coverage-presentation",
"custom-validation",
"custom-validation-prompt",
diff --git a/sdk/typescript/src/api.ts b/sdk/typescript/src/api.ts
index 3da0dbb24..6f4c62b2f 100644
--- a/sdk/typescript/src/api.ts
+++ b/sdk/typescript/src/api.ts
@@ -90,6 +90,7 @@ import {
MODEL_FILE_SETTINGS,
resolveNestedConfigPaths,
scanApprovalPolicy,
+ scanPricingContext,
scanCyberAccessConfig,
scanModelConfiguration,
scanModelProvider,
@@ -111,7 +112,7 @@ import {
type ScanUsageObservation,
type ScanWorkerEvent,
} from "./cost.js";
-import type { ScanTokenUsage } from "./cost-model.js";
+import type { ScanPricingContext, ScanTokenUsage } from "./cost-model.js";
import { measuredScanCost, combineMeasurements } from "./scan-measurement.js";
import { ScanProgressTracker, type DeepScanProgress } from "./scan-progress.js";
import { findScanSession } from "./scan-logs.js";
@@ -345,6 +346,7 @@ interface PreparedRuntime {
}
interface PreparedSession {
+ pricingContext?: ScanPricingContext;
safetyIdentifier?: string;
runtime: PreparedRuntime;
runtimeHome: string;
@@ -1368,6 +1370,7 @@ export class CodexSecurity {
tracker = new ScanCostTracker({
codexHome: runtime.codexHome,
model: scanModelConfiguration(session.effectiveConfig).model,
+ pricingContext: session.pricingContext,
repository: inputs.repository,
onMeasurement: (measurement) => {
if (this.#operationTelemetry !== null)
@@ -1538,6 +1541,7 @@ export class CodexSecurity {
? repository.replace(/[A-Z]/g, (letter) => letter.toLowerCase())
: repository;
const configured = scanModelConfiguration(session.effectiveConfig);
+ const { pricingContext } = session;
const model = options.model ?? configured.model;
const reasoningEffort =
options.reasoningEffort ?? configured.reasoningEffort;
@@ -1565,6 +1569,7 @@ export class CodexSecurity {
tracker = new ScanCostTracker({
codexHome: session.runtime.codexHome,
model,
+ pricingContext,
repository,
onMeasurement: (measurement) => {
if (this.#operationTelemetry !== null)
@@ -1661,7 +1666,10 @@ export class CodexSecurity {
tracker = null;
const snapshot = await activeTracker.stop(turn.usage).catch((error) => {
reportTrackingError(error);
- return { usage: turn.usage, cost: estimateScanCost(model, turn.usage) };
+ return {
+ usage: turn.usage,
+ cost: estimateScanCost(model, turn.usage, pricingContext),
+ };
});
return {
...outcome,
@@ -1715,7 +1723,11 @@ export class CodexSecurity {
);
const model = scanModelConfiguration(configuration);
const modelProvider = scanModelProvider(configuration);
- validateScanCostLimit(options.maxCostUsd, model.model);
+ validateScanCostLimit(
+ options.maxCostUsd,
+ model.model,
+ scanPricingContext(configuration, this.#dependencies.environment),
+ );
const archiveDir =
options.archiveExisting === true
? await planOutputArchive(inputs.outputDir)
@@ -1861,9 +1873,9 @@ export class CodexSecurity {
signal,
temporaryRoot,
);
- const { runtime, effectiveConfig } = session;
+ const { runtime, effectiveConfig, pricingContext } = session;
const model = scanModelConfiguration(effectiveConfig);
- validateScanCostLimit(options.maxCostUsd, model.model);
+ validateScanCostLimit(options.maxCostUsd, model.model, pricingContext);
for (const path of [
"references/threat-model.md",
"references/security-guidance.md",
@@ -2017,6 +2029,7 @@ export class CodexSecurity {
const tracker = new ScanCostTracker({
codexHome: runtime.codexHome,
model: model.model,
+ pricingContext,
repository: target.repository,
scanDirectory: outputDir,
maxCostUsd: options.maxCostUsd,
@@ -2026,6 +2039,7 @@ export class CodexSecurity {
this.#operationTelemetry.measurement = combineMeasurements(
[...stageMeasurements.values()],
model.model,
+ pricingContext,
);
},
onCost:
@@ -2072,7 +2086,7 @@ export class CodexSecurity {
warn(
`Could not track policy-generation cost: ${errorMessage(error)}`,
);
- const cost = estimateScanCost(model.model, usage);
+ const cost = estimateScanCost(model.model, usage, pricingContext);
if (cost !== null) reportCost(cost);
return { usage, cost };
});
@@ -2192,6 +2206,7 @@ export class CodexSecurity {
pythonPath: string;
protectedRoot: string;
model: string;
+ pricingContext?: ScanPricingContext;
threadId: string | null;
} | null = null;
let preparedTargetWarnings: string[] = [];
@@ -2328,6 +2343,7 @@ export class CodexSecurity {
effectiveConfig,
preflightConfig,
modelProvider,
+ pricingContext,
authentication,
approvalPolicy,
python,
@@ -2508,7 +2524,7 @@ export class CodexSecurity {
pluginVersion: runtime.plugin.version,
};
const { model } = scanModelConfiguration(effectiveConfig);
- validateScanCostLimit(options.maxCostUsd, model);
+ validateScanCostLimit(options.maxCostUsd, model, pricingContext);
if (mode === "deep" && options.maxCostUsd !== undefined) {
budgetRecovery = {
expectation,
@@ -2516,6 +2532,7 @@ export class CodexSecurity {
pythonPath: session.python,
protectedRoot,
model,
+ pricingContext,
threadId: null,
};
}
@@ -2929,6 +2946,7 @@ export class CodexSecurity {
const tracker = new ScanCostTracker({
codexHome: runtime.codexHome,
model,
+ pricingContext,
repository: repo,
scanDirectory: scanDir,
...(mode === "deep"
@@ -3482,6 +3500,7 @@ export class CodexSecurity {
modelProvider,
workbenchValidated: true,
model,
+ pricingContext,
onTurnStatus: (status, message) => {
scanTurnStatus = status;
// A later caller abort during iterator cleanup cannot undo a reported failure.
@@ -3606,7 +3625,7 @@ export class CodexSecurity {
reportTrackingError(error);
return {
usage,
- cost: estimateScanCost(model, usage),
+ cost: estimateScanCost(model, usage, pricingContext),
measurement: latestMeasurement,
};
});
@@ -3759,6 +3778,7 @@ export class CodexSecurity {
protectedRoot,
expectation,
model,
+ pricingContext,
onReconnect: options.onReconnect,
onWorkerStatus: options.onWorkerStatus,
onObserverError: options.onObserverError,
@@ -3987,6 +4007,7 @@ export class CodexSecurity {
{
status: "completed",
model: budgetRecovery.model,
+ ...turnPricingMetadata(budgetRecovery.pricingContext),
usage: snapshot.usage,
},
budgetRecovery.threadId,
@@ -4961,6 +4982,10 @@ export class CodexSecurity {
preflightConfig,
sessionConfig,
modelProvider,
+ pricingContext: scanPricingContext(
+ effectiveConfig,
+ runtime.environment,
+ ),
providerEnvKey,
nativeProviderEnvironment:
providerEnvKey !== undefined &&
@@ -5174,11 +5199,14 @@ export class CodexSecurity {
notifyObserver(options, "onOutputDirReady")(scanDir);
}
const revision = await repositoryRevision(local.repository, signal);
- const { model } = scanModelConfiguration(
- await mergedCodexConfig(
- this.config,
- scanCodexHome(this.#dependencies.environment),
- ),
+ const config = await mergedCodexConfig(
+ this.config,
+ scanCodexHome(this.#dependencies.environment),
+ );
+ const { model } = scanModelConfiguration(config);
+ const pricingContext = scanPricingContext(
+ config,
+ this.#dependencies.environment,
);
const workbenchOptions: WorkbenchCommandOptions = {
python,
@@ -5265,7 +5293,7 @@ export class CodexSecurity {
cached_input_tokens: 0,
output_tokens: 0,
};
- const cost = estimateScanCost(model, usage);
+ const cost = estimateScanCost(model, usage, pricingContext);
await workbench(
workbenchOptions,
workbenchArgs("prepare-scan-completion", { scanId }),
@@ -5297,6 +5325,7 @@ export class CodexSecurity {
model,
usage,
mock: true,
+ ...turnPricingMetadata(pricingContext),
finalResponse:
"Synthetic mock scan; no security analysis was performed.",
},
@@ -5690,6 +5719,7 @@ interface ScanEventRunOptions extends Pick<ScanOptions, "onReconnect"> {
modelProvider?: unknown;
workbenchValidated?: boolean;
model?: string;
+ pricingContext?: ScanPricingContext;
onFinalize?: (usage: unknown) => Promise<unknown>;
onTurnStatus?: (status: "completed" | "failed", message?: string) => void;
onThreadStarted?: (threadId: string) => Promise<void> | void;
@@ -5857,6 +5887,7 @@ export async function runScanEvents(
finalResponse,
usage,
...(options.model === undefined ? {} : { model: options.model }),
+ ...turnPricingMetadata(options.pricingContext),
},
threadId,
options.scanDir,
@@ -6250,14 +6281,31 @@ async function prepareScanOutputDir(
return output;
}
+function turnPricingMetadata(context?: ScanPricingContext) {
+ return context === undefined ? {} : { bedrockRegion: context.region };
+}
+
function validateScanCostLimit(
maxCostUsd: number | undefined,
model: string,
+ pricingContext?: ScanPricingContext,
): void {
if (maxCostUsd === undefined) return;
- if (estimateScanCost(model, { input_tokens: 0, output_tokens: 0 }) === null) {
+ if (
+ estimateScanCost(
+ model,
+ { input_tokens: 0, output_tokens: 0 },
+ pricingContext,
+ ) === null
+ ) {
+ const region = pricingContext?.region;
+ const location = region ? ` in ${region}` : "";
+ const detail =
+ region === null
+ ? " Set the Bedrock billing region to check its regional price."
+ : "";
throw new CodexSecurityError(
- `A scan cost limit is not available for the configured model: ${model}.`,
+ `A scan cost limit is not available for the configured model: ${model}${location}.${detail}`,
);
}
}
diff --git a/sdk/typescript/src/bedrock-oss-pricing.ts b/sdk/typescript/src/bedrock-oss-pricing.ts
new file mode 100644
index 000000000..80b890c52
--- /dev/null
+++ b/sdk/typescript/src/bedrock-oss-pricing.ts
@@ -0,0 +1,81 @@
+// Standard on-demand input and output in USD per million tokens (AWS quotes
+// 1K-token units in the bulk catalog). The four OSS models have a 128K context
+// window and one input/output rate for the whole request.
+type Prices = readonly [input: number, output: number];
+type ModelRates = Readonly<Record<string, Prices | undefined>>;
+
+function models(
+ base20b: Prices,
+ base120b: Prices,
+ safeguard20b?: Prices,
+ safeguard120b?: Prices,
+): ModelRates {
+ return {
+ "openai.gpt-oss-20b": base20b,
+ "openai.gpt-oss-120b": base120b,
+ "openai.gpt-oss-20b-1:0": base20b,
+ "openai.gpt-oss-120b-1:0": base120b,
+ "openai.gpt-oss-safeguard-20b": safeguard20b,
+ "openai.gpt-oss-safeguard-120b": safeguard120b,
+ };
+}
+
+const us = models([0.07, 0.3], [0.15, 0.6], [0.07, 0.2], [0.15, 0.6]);
+const gov = models([0.084, 0.36], [0.18, 0.72]);
+const tokyo = models([0.08, 0.36], [0.18, 0.73], [0.08, 0.24], [0.18, 0.73]);
+
+const regions: Readonly<Record<string, ModelRates>> = {
+ "us-east-1": us,
+ "us-east-2": us,
+ "us-west-2": us,
+ "us-gov-east-1": gov,
+ "us-gov-west-1": gov,
+ "ap-northeast-1": tokyo,
+ "ap-south-1": models([0.08, 0.35], [0.18, 0.71], [0.08, 0.24], [0.18, 0.71]),
+ // The Standard table on AWS's pricing page explicitly lists Sydney, where
+ // the bulk catalog only has Mantle SKUs.
+ "ap-southeast-2": models(
+ [0.0721, 0.309],
+ [0.1545, 0.618],
+ [0.0721, 0.206],
+ [0.1545, 0.618],
+ ),
+ "ap-southeast-3": models(
+ [0.07, 0.31],
+ [0.16, 0.62],
+ [0.084, 0.24],
+ [0.18, 0.72],
+ ),
+ "eu-central-1": models([0.09, 0.4], [0.2, 0.79], [0.084, 0.24], [0.18, 0.72]),
+ "eu-north-1": models([0.07, 0.3], [0.15, 0.6], [0.084, 0.24], [0.18, 0.72]),
+ "eu-south-1": models([0.09, 0.4], [0.2, 0.79], [0.08, 0.23], [0.18, 0.7]),
+ "eu-west-1": models([0.08, 0.35], [0.18, 0.7], [0.08, 0.23], [0.18, 0.7]),
+ "eu-west-2": models([0.11, 0.47], [0.23, 0.93], [0.11, 0.31], [0.23, 0.93]),
+ "sa-east-1": tokyo,
+};
+
+export function bedrockOssPricing(model: string, region?: string | null) {
+ const govProfile =
+ model === "us-gov.openai.gpt-oss-20b-1:0" ||
+ model === "us-gov.openai.gpt-oss-120b-1:0";
+ // Both possible source regions of these inference profiles have the same
+ // rates. A bare model ID requires the billing region to resolve its price.
+ const origin = region ?? (govProfile ? "us-gov-east-1" : undefined);
+ if (origin === undefined || !Object.hasOwn(regions, origin)) return undefined;
+ if (govProfile && origin !== "us-gov-east-1" && origin !== "us-gov-west-1")
+ return undefined;
+ const rates = regions[origin]!;
+ const id = govProfile ? model.slice("us-gov.".length) : model;
+ const price = Object.hasOwn(rates, id) ? rates[id] : undefined;
+ if (!price) return undefined;
+ const [input, output] = price;
+ const tokens = [input, input, input, output] as const;
+ return {
+ short: tokens,
+ long: tokens,
+ source:
+ origin === "ap-southeast-2" && model.endsWith("-1:0")
+ ? "https://aws.amazon.com/bedrock/pricing/#OpenAI"
+ : "https://pricing.us-east-1.amazonaws.com/offers/v1.0/aws/AmazonBedrock/20261102165015/index.json",
+ };
+}
diff --git a/sdk/typescript/src/bedrock-pricing.ts b/sdk/typescript/src/bedrock-pricing.ts
new file mode 100644
index 000000000..31ab2f239
--- /dev/null
+++ b/sdk/typescript/src/bedrock-pricing.ts
@@ -0,0 +1,256 @@
+import { bedrockOssPricing } from "./bedrock-oss-pricing.js";
+
+// AWS Bedrock Standard prices, in USD per million tokens. Each route lists
+// the source regions of the request (not cross-region routing destinations).
+type Prices = readonly [
+ input: number,
+ read: number,
+ write: number,
+ output: number,
+];
+type Route = "regional" | "us" | "us-gov" | "in" | "global";
+type Origins = Partial<Record<Route, readonly string[]>>;
+
+const usEast = ["us-east-1", "us-east-2"];
+const usMantle = [...usEast, "us-west-2"];
+const usGeo = [...usMantle, "us-west-1", "ca-central-1", "ca-west-1"];
+const gov = ["us-gov-east-1", "us-gov-west-1"];
+const india = ["ap-south-1", "ap-south-2"];
+const astraGlobal = [
+ ...usGeo,
+ "eu-central-1",
+ "eu-north-1",
+ "eu-west-1",
+ "eu-west-2",
+ "eu-west-3",
+ "ap-northeast-1",
+ "ap-northeast-2",
+ "ap-northeast-3",
+ "ap-south-1",
+ "ap-southeast-1",
+ "ap-southeast-2",
+ "sa-east-1",
+];
+const commonGlobal = [
+ ...astraGlobal,
+ "eu-central-2",
+ "eu-south-1",
+ "eu-south-2",
+ "ap-east-2",
+ "ap-south-2",
+ "ap-southeast-3",
+ "ap-southeast-4",
+ "ap-southeast-5",
+ "ap-southeast-6",
+ "ap-southeast-7",
+ "il-central-1",
+ "af-south-1",
+];
+const gpt56Global = [...commonGlobal, "me-central-1", "me-south-1"];
+// GPT-6.1 Sol links here for its global source regions:
+// https://docs.aws.amazon.com/bedrock/latest/userguide/models-region-compatibility.html#model-regions-openai
+const gpt6Global = [...commonGlobal, "mx-central-1"];
+
+function perToken([input, read, write, output]: Prices, scale: number): Prices {
+ return [
+ Math.round(input * scale),
+ Math.round(read * scale),
+ Math.round(write * scale),
+ Math.round(output * scale),
+ ];
+}
+
+function offer(
+ origins: Origins,
+ short: Prices,
+ long?: Prices,
+ source?: string,
+) {
+ return { origins, short, long, source };
+}
+
+function card(slug: string, ...offers: ReturnType<typeof offer>[]) {
+ return {
+ source: `https://docs.aws.amazon.com/bedrock/latest/userguide/model-card-openai-gpt-${slug}.html`,
+ offers,
+ };
+}
+
+// Prefer AWS's route-specific model card, including its GovCloud prices.
+// GPT-5.4 and GPT-5.5 list no separate write fee for their implicit cache.
+// Their cards publish regional rates. AWS prices geographic inference at the
+// source region's Standard rate, and does not specify the Global rate here:
+// https://docs.aws.amazon.com/bedrock/latest/userguide/models-region-compatibility.html
+const cards: Readonly<Record<string, ReturnType<typeof card>>> = {
+ "openai.gpt-5.4": card(
+ "54",
+ offer(
+ { regional: usMantle, us: usGeo },
+ [2.75, 0.275, 2.75, 16.5],
+ [5.5, 0.55, 5.5, 24.75],
+ ),
+ offer({ regional: ["us-gov-west-1"] }, [3.375, 0.3375, 3.375, 20.25]),
+ offer(
+ { regional: ["us-gov-east-1"] },
+ [3.3, 0.33, 3.3, 19.8],
+ undefined,
+ "https://pricing.us-east-1.amazonaws.com/offers/v1.0/aws/AmazonBedrock/20261102165015/us-gov-east-1/index.json",
+ ),
+ ),
+ "openai.gpt-5.5": card(
+ "55",
+ offer(
+ { regional: usEast, us: usGeo },
+ [5.5, 0.55, 5.5, 33],
+ [11, 1.1, 11, 49.5],
+ ),
+ ),
+ "openai.gpt-daybreak-blue-5.6-sol": card(
+ "daybreak-blue-56-sol",
+ offer(
+ { regional: ["us-east-2"] },
+ [4.4, 0.44, 5.5, 22],
+ [8.8, 0.88, 11, 33],
+ ),
+ ),
+ "openai.gpt-5.6-cyber": card(
+ "56-cyber",
+ offer({ regional: ["us-east-2"] }, [13.75, 1.375, 17.1875, 82.5]),
+ offer({ regional: ["us-gov-east-1"] }, [16.875, 1.6875, 21.09375, 101.25]),
+ ),
+ "openai.gpt-5.6-luna": card(
+ "56-luna",
+ offer(
+ { regional: usMantle, us: usGeo, in: india },
+ [0.22, 0.022, 0.275, 1.32],
+ [0.44, 0.044, 0.55, 1.98],
+ ),
+ offer(
+ { global: gpt56Global },
+ [0.2, 0.02, 0.25, 1.2],
+ [0.4, 0.04, 0.5, 1.8],
+ ),
+ offer(
+ { regional: gov, "us-gov": gov },
+ [0.27, 0.027, 0.3375, 1.62],
+ [0.54, 0.054, 0.675, 2.43],
+ ),
+ ),
+ "openai.gpt-5.6-terra": card(
+ "56-terra",
+ offer(
+ { regional: usMantle, us: usGeo, in: india },
+ [2.2, 0.22, 2.75, 13.2],
+ [4.4, 0.44, 5.5, 19.8],
+ ),
+ offer({ global: gpt56Global }, [2, 0.2, 2.5, 12], [4, 0.4, 5, 18]),
+ offer(
+ { regional: gov, "us-gov": gov },
+ [2.7, 0.27, 3.375, 16.2],
+ [5.4, 0.54, 6.75, 24.3],
+ ),
+ ),
+ "openai.gpt-5.6-sol": card(
+ "56-sol",
+ offer(
+ { regional: usEast, us: usGeo },
+ [4.4, 0.44, 5.5, 22],
+ [8.8, 0.88, 11, 33],
+ ),
+ offer({ global: gpt56Global }, [4, 0.4, 5, 20], [8, 0.8, 10, 30]),
+ ),
+ "openai.gpt-6-astra": card(
+ "6-astra",
+ offer(
+ { regional: ["us-east-1", "us-west-2"], us: usGeo },
+ [11, 1.1, 13.75, 55],
+ [22, 2.2, 27.5, 82.5],
+ ),
+ offer({ global: astraGlobal }, [10, 1, 12.5, 50], [20, 2, 25, 75]),
+ ),
+ "openai.gpt-6-luna": card(
+ "6-luna",
+ offer(
+ { regional: ["us-east-1"], us: usGeo },
+ [0.11, 0.011, 0.1375, 0.55],
+ [0.22, 0.022, 0.275, 0.825],
+ ),
+ offer(
+ { global: gpt6Global },
+ [0.1, 0.01, 0.125, 0.5],
+ [0.2, 0.02, 0.25, 0.75],
+ ),
+ ),
+ "openai.gpt-6-sol": card(
+ "6-sol",
+ offer(
+ { regional: ["us-east-1"], us: usGeo },
+ [2.2, 0.22, 2.75, 11],
+ [4.4, 0.44, 5.5, 16.5],
+ ),
+ offer({ global: gpt6Global }, [2, 0.2, 2.5, 10], [4, 0.4, 5, 15]),
+ ),
+ "openai.gpt-6.1-sol": card(
+ "6-1-sol",
+ offer(
+ { regional: ["us-east-1"], us: usGeo },
+ [2.2, 0.11, 2.75, 11],
+ [4.4, 0.22, 5.5, 16.5],
+ ),
+ offer({ global: gpt6Global }, [2, 0.1, 2.5, 10], [4, 0.2, 5, 15]),
+ ),
+ "openai.gpt-6.1-terra": card(
+ "6-1-terra",
+ offer(
+ { regional: ["us-east-1"], us: usGeo },
+ [1.1, 0.055, 1.375, 5.5],
+ [2.2, 0.11, 2.75, 8.25],
+ ),
+ offer({ global: gpt6Global }, [1, 0.05, 1.25, 5], [2, 0.1, 2.5, 7.5]),
+ ),
+};
+
+function frontierPricing(model: string, region?: string | null) {
+ const match = /^(?:(us|us-gov|in|global)\.)?(openai\.gpt-[\w.-]+)$/.exec(
+ model,
+ );
+ if (!match) return undefined;
+ const modelCard = cards[match[2]!];
+ if (!modelCard) return undefined;
+ const route = (match[1] ?? "regional") as Route;
+ // The first regional offer is the model-only SDK's commercial default.
+ // A live scan with an unresolved region (null) needs an unambiguous rate.
+ const offers = modelCard.offers.filter(
+ ({ origins }) =>
+ origins[route] && (region == null || origins[route].includes(region)),
+ );
+ if (region === null && offers.length > 1) return undefined;
+ const rate = offers[0];
+ if (!rate) return undefined;
+ return {
+ short: rate.short,
+ long: rate.long,
+ source: rate.source ?? modelCard.source,
+ };
+}
+
+export function bedrockModelPricing(model: string, region?: string | null) {
+ const rate =
+ frontierPricing(model, region) ?? bedrockOssPricing(model, region);
+ if (!rate) return undefined;
+
+ // Keep nanodollar arithmetic where possible. GovCloud Cyber cache writes
+ // require quarter-nanodollars, and Sydney OSS prices need tenths.
+ const prices = [...rate.short, ...(rate.long ?? [])];
+ const scale = [1_000, 2_000, 4_000, 10_000].find((scale) =>
+ prices.every((price) => Math.round(price * scale) / scale === price),
+ );
+ if (!scale) throw new Error("Bedrock price exceeds supported precision");
+ return {
+ short: perToken(rate.short, scale),
+ long: rate.long && perToken(rate.long, scale),
+ unitsPerUsd: scale * 1_000_000,
+ source: rate.source,
+ asOf: "2026-11-05",
+ };
+}
diff --git a/sdk/typescript/src/cli.ts b/sdk/typescript/src/cli.ts
index 9bba63c6d..4421bf78c 100644
--- a/sdk/typescript/src/cli.ts
+++ b/sdk/typescript/src/cli.ts
@@ -195,6 +195,7 @@ import {
resolveCodexProfile,
mergedCodexConfig,
readCodexFileProfile,
+ scanPricingContext,
scanModel,
scanModelConfiguration,
scanModelProvider,
@@ -5863,14 +5864,17 @@ export async function main(
dependencies.environment["CI"] === undefined &&
dependencies.environment["TERM"] !== "dumb"
) {
+ const dashboardConfig = await mergedCodexConfig(
+ config,
+ scanCodexHome(dependencies.environment),
+ );
const candidate = new ScanDashboard(errorOutput, {
repository,
presentation: "components",
- model: scanModelConfiguration(
- await mergedCodexConfig(
- config,
- scanCodexHome(dependencies.environment),
- ),
+ model: scanModelConfiguration(dashboardConfig),
+ pricingContext: scanPricingContext(
+ dashboardConfig,
+ dependencies.environment,
),
mode: settings.mode,
maxCostUsd: settings.maxCostUsd,
@@ -10391,6 +10395,10 @@ async function executeScan(
mode: arguments_.mode,
showCost,
model: scanModelConfiguration(effectiveConfiguration),
+ pricingContext: scanPricingContext(
+ effectiveConfiguration,
+ dependencies.environment,
+ ),
...(arguments_.maxCostUsd === undefined
? {}
: { maxCostUsd: arguments_.maxCostUsd }),
diff --git a/sdk/typescript/src/config.ts b/sdk/typescript/src/config.ts
index 1ffa9d0f6..0f679535d 100644
--- a/sdk/typescript/src/config.ts
+++ b/sdk/typescript/src/config.ts
@@ -1,3 +1,4 @@
+import type { ScanPricingContext } from "./cost-model.js";
import { randomUUID } from "node:crypto";
import { mkdir, open, readFile, rename, unlink } from "node:fs/promises";
import { dirname, join, resolve } from "node:path";
@@ -122,6 +123,50 @@ export function scanModelProvider(config: Readonly<JsonObject>): unknown {
return selectedProfile?.["model_provider"] ?? config["model_provider"];
}
+function bedrockPricingRegion(
+ provider: JsonValue | undefined,
+ environment: Readonly<Record<string, string | undefined>>,
+): string | null {
+ const baseUrl = isObject(provider) ? provider["base_url"] : undefined;
+ if (typeof baseUrl === "string") {
+ // Codex sends inference to an explicit endpoint even when aws.region differs.
+ const host = URL.parse(baseUrl)?.hostname ?? "";
+ return (
+ /^bedrock-(?:mantle|runtime)(?:-fips)?\.([a-z0-9-]+)\.(?:api\.aws|amazonaws\.com(?:\.cn)?)$/.exec(
+ host,
+ )?.[1] ?? null
+ );
+ }
+ const aws = isObject(provider) ? provider["aws"] : undefined;
+ const region = isObject(aws) ? aws["region"] : undefined;
+ return (
+ (typeof region === "string" && region.trim()) ||
+ environment["AWS_REGION"]?.trim() ||
+ environment["AWS_DEFAULT_REGION"]?.trim() ||
+ null
+ );
+}
+
+/** @internal Keep billing selection separate from model names shared by providers. */
+export function scanPricingContext(
+ config: Readonly<JsonObject>,
+ environment: Readonly<Record<string, string | undefined>>,
+): ScanPricingContext | undefined {
+ const selected = scanModelProvider(config);
+ if (selected !== "amazon-bedrock" && selected !== "amazon-bedrock-runtime") {
+ return undefined;
+ }
+ const resolved = selectedScanProfile(config)
+ ? resolveCodexProfile(config)
+ : config;
+ const providers = resolved["model_providers"];
+ const provider = isObject(providers) ? providers[selected] : undefined;
+ return {
+ provider: "amazon-bedrock",
+ region: bedrockPricingRegion(provider, environment),
+ };
+}
+
/** @internal Use the selected native provider's configured credential key. */
export function scanProviderEnvKey(
config: Readonly<JsonObject>,
diff --git a/sdk/typescript/src/cost-model.ts b/sdk/typescript/src/cost-model.ts
index 98200d173..752e51c29 100644
--- a/sdk/typescript/src/cost-model.ts
+++ b/sdk/typescript/src/cost-model.ts
@@ -1,6 +1,12 @@
+import { bedrockModelPricing } from "./bedrock-pricing.js";
import { isRecord } from "./record.js";
import { isSafeNonNegativeInteger } from "./value.js";
+export interface ScanPricingContext {
+ provider: "amazon-bedrock";
+ region: string | null;
+}
+
export interface ScanCost {
model: string;
inputTokens: number;
@@ -47,38 +53,6 @@ export interface ScanTokenUsage {
total_tokens: number;
}
-const BEDROCK_MODEL_PRICING: Readonly<
- Record<
- string,
- {
- short: ModelPricing;
- long?: ModelPricing;
- unitsPerUsd: number;
- source: string;
- asOf: string;
- }
- >
-> = {
- // AWS commercial in-region Standard prices already include the 10% fee.
- "openai.gpt-daybreak-blue-5.6-sol": {
- short: [4_400, 440, 5_500, 22_000],
- long: [8_800, 880, 11_000, 33_000],
- unitsPerUsd: 1_000_000_000,
- source:
- "https://docs.aws.amazon.com/en_en/bedrock/latest/userguide/model-card-openai-gpt-daybreak-blue-56-sol.html",
- asOf: "2026-10-01",
- },
- "openai.gpt-5.6-cyber": {
- // Half-nanodollar units preserve the $17.1875/M cache-write rate exactly,
- // including an odd number of cache-write tokens.
- short: [27_500, 2_750, 34_375, 165_000],
- unitsPerUsd: 2_000_000_000,
- source:
- "https://docs.aws.amazon.com/en_en/bedrock/latest/userguide/model-card-openai-gpt-56-cyber.html",
- asOf: "2026-10-01",
- },
-};
-
const MODEL_PRICING_NANODOLLARS: Readonly<Record<string, ModelPricing>> = {
// GPT-5.5 has no additional cache-write charge.
"gpt-5.5": [5_000, 500, 5_000, 30_000],
@@ -169,22 +143,25 @@ export function tokenUsage(value: unknown): ScanTokenUsage | null {
};
}
+/**
+ * Estimates Standard token costs using the selected provider's published rates.
+ * A string (or null for an unknown region) selects Bedrock for compatibility.
+ * Model-only Bedrock frontier estimates use the commercial in-region price.
+ */
export function estimateScanCost(
model: string | undefined,
usage: unknown,
+ context?: ScanPricingContext | string | null,
): ScanCost | null {
if (model === undefined) return null;
- const bedrockPricing = Object.hasOwn(BEDROCK_MODEL_PRICING, model)
- ? BEDROCK_MODEL_PRICING[model]
- : undefined;
- const unitsPerUsd = bedrockPricing?.unitsPerUsd ?? 1_000_000_000;
- const pricingModel = model.startsWith("openai.")
- ? model.slice("openai.".length)
- : model;
+ const region =
+ typeof context === "object" && context !== null ? context.region : context;
+ const providerPricing = bedrockModelPricing(model, region);
+ const unitsPerUsd = providerPricing?.unitsPerUsd ?? 1_000_000_000;
const pricing =
- bedrockPricing?.short ??
- (Object.hasOwn(MODEL_PRICING_NANODOLLARS, pricingModel)
- ? MODEL_PRICING_NANODOLLARS[pricingModel]
+ providerPricing?.short ??
+ (context === undefined && Object.hasOwn(MODEL_PRICING_NANODOLLARS, model)
+ ? MODEL_PRICING_NANODOLLARS[model]
: undefined);
const normalized = tokenUsage(usage);
if (pricing === undefined || normalized === null) return null;
@@ -203,9 +180,9 @@ export function estimateScanCost(
outputTokens * outputRate;
if (!Number.isSafeInteger(costUnits)) return null;
- const longPricing = bedrockPricing
- ? bedrockPricing.long
- : LONG_CONTEXT_PRICING_NANODOLLARS[pricingModel];
+ const longPricing = providerPricing
+ ? providerPricing.long
+ : LONG_CONTEXT_PRICING_NANODOLLARS[model];
let maximumUnits: number | null = null;
if (longPricing !== undefined) {
const [longInput, longRead, longWrite, longOutput] = longPricing;
@@ -239,22 +216,22 @@ export function estimateScanCost(
context: "unknown",
},
pricing: {
- source: bedrockPricing
- ? bedrockPricing.source
- : pricingModel.startsWith("gpt-5.5")
+ source: providerPricing
+ ? providerPricing.source
+ : model.startsWith("gpt-5.5")
? "https://developers.openai.com/api/docs/models/gpt-5.5"
- : pricingModel === "gpt-6-sol"
+ : model === "gpt-6-sol"
? "https://developers.openai.com/api/docs/models/gpt-6-sol"
: "https://developers.openai.com/api/docs/pricing",
- asOf: bedrockPricing
- ? bedrockPricing.asOf
- : pricingModel === "gpt-6-sol"
+ asOf: providerPricing
+ ? providerPricing.asOf
+ : model === "gpt-6-sol"
? "2026-10-10"
- : pricingModel === "gpt-5.6-cyber"
+ : model === "gpt-5.6-cyber"
? "2026-10-06"
- : pricingModel === "gpt-6-cyber"
+ : model === "gpt-6-cyber"
? "2026-11-05"
- : pricingModel === "gpt-6.1-sol" || pricingModel === "gpt-6-luna"
+ : model === "gpt-6.1-sol" || model === "gpt-6-luna"
? "2026-09-30"
: "2026-09-14",
serviceTier: "standard",
diff --git a/sdk/typescript/src/cost.ts b/sdk/typescript/src/cost.ts
index 580511537..dfa362db5 100644
--- a/sdk/typescript/src/cost.ts
+++ b/sdk/typescript/src/cost.ts
@@ -21,6 +21,7 @@ import {
estimateScanCost,
tokenUsage,
type ScanCost,
+ type ScanPricingContext,
type ScanTokenUsage,
} from "./cost-model.js";
import {
@@ -332,6 +333,7 @@ function sessionObservation<T>(
interface ScanCostTrackerOptions {
codexHome: string;
+ pricingContext?: ScanPricingContext;
model: string;
repository?: string;
scanDirectory?: string;
@@ -2445,7 +2447,11 @@ export class ScanCostTracker {
this.#snapshot.measurement.missingThreadCount > 0
)
return this.#snapshot;
- const cost = estimateScanCost(this.#options.model, fallbackUsage);
+ const cost = estimateScanCost(
+ this.#options.model,
+ fallbackUsage,
+ this.#options.pricingContext,
+ );
const usage = tokenUsage(fallbackUsage);
this.#setSnapshot(usage, cost, usage !== null, usage === null ? 0 : 1, 0);
this.#reportUsage(usage);
@@ -2835,7 +2841,11 @@ export class ScanCostTracker {
details?: Pick<ScanUsageMeasurement, "coverageScope" | "warnings">,
reportCost = true,
): void {
- const cost = estimateScanCost(this.#options.model, usage);
+ const cost = estimateScanCost(
+ this.#options.model,
+ usage,
+ this.#options.pricingContext,
+ );
this.#setSnapshot(
usage,
cost,
diff --git a/sdk/typescript/src/index.ts b/sdk/typescript/src/index.ts
index 29f6060ac..ec1114745 100644
--- a/sdk/typescript/src/index.ts
+++ b/sdk/typescript/src/index.ts
@@ -94,7 +94,7 @@ export type {
ScanUsageObservation,
ScanWorkerEvent,
} from "./cost.js";
-export type { ScanTokenUsage } from "./cost-model.js";
+export type { ScanPricingContext, ScanTokenUsage } from "./cost-model.js";
export type { DeepScanProgress } from "./scan-progress.js";
export type { CustomValidationResult } from "./custom-validation.js";
export type { ScanActivity, ScanActivityStatus } from "./scan-activity.js";
diff --git a/sdk/typescript/src/result.ts b/sdk/typescript/src/result.ts
index d413ea336..31fca3fe6 100644
--- a/sdk/typescript/src/result.ts
+++ b/sdk/typescript/src/result.ts
@@ -15,6 +15,8 @@ export interface TurnResultMetadata {
id?: string;
status?: string;
model?: string;
+ /** Source region recorded for Bedrock scans. */
+ bedrockRegion?: string | null;
durationMs?: number;
finalResponse?: string;
usage?: unknown;
@@ -88,7 +90,11 @@ export class ScanResult {
this.threatModelPath = options.threatModelPath ?? null;
this.cost =
options.recordedCost ??
- estimateScanCost(options.turnResult.model, options.turnResult.usage);
+ estimateScanCost(
+ options.turnResult.model,
+ options.turnResult.usage,
+ options.turnResult.bedrockRegion,
+ );
if (options.sarifPath !== undefined) {
this.sarifPath = options.sarifPath;
} else {
diff --git a/sdk/typescript/src/scan-dashboard.ts b/sdk/typescript/src/scan-dashboard.ts
index d99a11642..bc54a73fc 100644
--- a/sdk/typescript/src/scan-dashboard.ts
+++ b/sdk/typescript/src/scan-dashboard.ts
@@ -26,6 +26,7 @@ import {
formatScanCosts,
formatScanCostTokens,
formatTokenUsage,
+ type ScanPricingContext,
type ScanTokenUsage,
} from "./cost-model.js";
import { formatDeepScanReviews } from "./scan-progress.js";
@@ -92,6 +93,7 @@ interface ScanDashboardOptions {
componentName?: string;
mode?: ScanMode;
model?: ScanModelConfiguration;
+ pricingContext?: ScanPricingContext;
maxCostUsd?: number;
showCost?: boolean;
clock: DashboardClock;
@@ -1165,10 +1167,11 @@ export class ScanDashboard {
if (
this.#cost === null &&
this.#options.maxCostUsd === undefined &&
- estimateScanCost(this.#options.model?.model, {
- input_tokens: 0,
- output_tokens: 0,
- }) === null
+ estimateScanCost(
+ this.#options.model?.model,
+ { input_tokens: 0, output_tokens: 0 },
+ this.#options.pricingContext,
+ ) === null
) {
return [];
}
diff --git a/sdk/typescript/src/scan-measurement.ts b/sdk/typescript/src/scan-measurement.ts
index 178cda879..3b7cf1a39 100644
--- a/sdk/typescript/src/scan-measurement.ts
+++ b/sdk/typescript/src/scan-measurement.ts
@@ -2,17 +2,19 @@ import { addTokenUsage, type ScanUsageMeasurement } from "./cost.js";
import {
estimateScanCost,
tokenUsage,
+ type ScanPricingContext,
type ScanTokenUsage,
} from "./cost-model.js";
export function measurementFromUsage(
value: unknown,
model: string,
+ pricingContext?: ScanPricingContext,
): ScanUsageMeasurement {
const usage = tokenUsage(value);
return {
usage,
- cost: estimateScanCost(model, usage),
+ cost: estimateScanCost(model, usage, pricingContext),
coverage: usage === null ? "unavailable" : "complete",
observedAt: new Date().toISOString(),
threadCount: usage === null ? 0 : 1,
@@ -24,6 +26,7 @@ export function measurementFromUsage(
export function combineMeasurements(
values: readonly Readonly<ScanUsageMeasurement>[],
model: string,
+ pricingContext?: ScanPricingContext,
): ScanUsageMeasurement {
let usage: ScanTokenUsage | null = null;
for (const value of values) {
@@ -33,7 +36,7 @@ export function combineMeasurements(
...new Set(values.flatMap((value) => value.warnings ?? [])),
].sort();
return {
- ...measurementFromUsage(usage, model),
+ ...measurementFromUsage(usage, model, pricingContext),
...(warnings.length ? { warnings } : {}),
coverage:
usage === null
diff --git a/sdk/typescript/tests-ts/api.test.ts b/sdk/typescript/tests-ts/api.test.ts
index b48d3e705..e9d763f1a 100644
--- a/sdk/typescript/tests-ts/api.test.ts
+++ b/sdk/typescript/tests-ts/api.test.ts
@@ -1176,6 +1176,7 @@ describe("CodexSecurity orchestration", () => {
CODEX_API_KEY: "synthetic-codex-key",
OPENROUTER_API_KEY: "synthetic-openrouter-key",
FIREWORKS_API_KEY: "synthetic-fireworks-key",
+ AWS_DEFAULT_REGION: "us-east-2",
...credentials,
};
const client = new TestClient(
@@ -1246,11 +1247,13 @@ describe("CodexSecurity orchestration", () => {
const { root, repository, codexHome, scanDir } = await scanDirectories();
const createCodex = mock(completedCodex(root));
const onAuthentication = mock<(selected: ScanAuthentication) => void>();
+ const onCost = mock<(cost: ScanCost) => void>();
+ let savedCost: ScanCost | undefined;
let savedRecipe: Record<string, unknown> | undefined;
const environment = {
OPENAI_API_KEY: "synthetic-openai-key-must-not-be-used",
AWS_BEARER_TOKEN_BEDROCK: "synthetic-bedrock-bearer",
- AWS_REGION: "us-east-2",
+ AWS_REGION: "us-gov-west-1",
};
const client = new TestClient(
{
@@ -1259,13 +1262,13 @@ describe("CodexSecurity orchestration", () => {
model_provider: "openai",
profiles: {
bedrock: {
- model: "openai.gpt-5.6-luna",
+ model: "openai.gpt-5.6-cyber",
model_provider: "amazon-bedrock",
},
},
model_providers: {
"amazon-bedrock": {
- aws: { region: "us-east-2", profile: "security-prod" },
+ aws: { region: "us-gov-east-1", profile: "security-prod" },
},
},
},
@@ -1281,6 +1284,9 @@ describe("CodexSecurity orchestration", () => {
): Promise<JsonObject> => {
if (args[0] === "register-cli-scan") {
savedRecipe = JSON.parse(input!).recipe;
+ } else if (args[0] === "complete-scan") {
+ const costJson = args[args.indexOf("--cost-json") + 1]!;
+ savedCost = JSON.parse(costJson).cost;
}
return mockWorkbench(args, input);
},
@@ -1293,7 +1299,7 @@ describe("CodexSecurity orchestration", () => {
await expect(
client.preflight(repository, { maxCostUsd: 1 }),
).resolves.toMatchObject({
- model: "openai.gpt-5.6-luna",
+ model: "openai.gpt-5.6-cyber",
modelProvider: "amazon-bedrock",
maxCostUsd: 1,
authentication: {
@@ -1305,6 +1311,7 @@ describe("CodexSecurity orchestration", () => {
const result = await client.run(repository, {
maxCostUsd: 1,
onAuthentication,
+ onCost,
});
expect(onAuthentication.mock.lastCall?.[0]).toEqual({
@@ -1314,7 +1321,7 @@ describe("CodexSecurity orchestration", () => {
});
expect(createCodex.mock.lastCall?.[0]?.env).toMatchObject({
AWS_BEARER_TOKEN_BEDROCK: "synthetic-bedrock-bearer",
- AWS_REGION: "us-east-2",
+ AWS_REGION: "us-gov-west-1",
});
expect(createCodex.mock.lastCall?.[0]?.config).toMatchObject({
model_reasoning_summary: "none",
@@ -1323,20 +1330,38 @@ describe("CodexSecurity orchestration", () => {
expect(createCodex.mock.lastCall?.[0]?.env).not.toHaveProperty(
"OPENAI_API_KEY",
);
- expect(result.cost).toMatchObject({ model: "openai.gpt-5.6-luna" });
+ expect(result.turnResult.bedrockRegion).toBe("us-gov-east-1");
+ expect(result.cost).toMatchObject({
+ model: "openai.gpt-5.6-cyber",
+ inputTokens: 10,
+ cachedInputTokens: 2,
+ outputTokens: 3,
+ estimatedUsd: 0.000442125,
+ estimatedUsdRange: { min: 0.000442125, max: null },
+ pricing: {
+ usdPerMillionTokens: {
+ input: 16.875,
+ cacheRead: 1.6875,
+ cacheWrite: 21.09375,
+ output: 101.25,
+ },
+ },
+ });
+ expect(onCost.mock.lastCall?.[0]).toEqual(result.cost!);
+ expect(savedCost).toEqual(result.cost!);
expect(savedRecipe).toMatchObject({
config: {
model_provider: "openai",
profile: "bedrock",
profiles: {
bedrock: {
- model: "openai.gpt-5.6-luna",
+ model: "openai.gpt-5.6-cyber",
model_provider: "amazon-bedrock",
},
},
model_providers: {
"amazon-bedrock": {
- aws: { region: "us-east-2", profile: "security-prod" },
+ aws: { region: "us-gov-east-1", profile: "security-prod" },
},
},
},
@@ -1344,6 +1369,36 @@ describe("CodexSecurity orchestration", () => {
await client.close();
});
+ test.each([
+ ["missing", {}],
+ [
+ "unsupported",
+ { AWS_REGION: "us-gov-west-1", AWS_DEFAULT_REGION: "us-gov-east-1" },
+ ],
+ ] as const)(
+ "rejects Bedrock budgets when the billing region is %s",
+ async (_reason, environment) => {
+ const { repository } = await repositoryDirectories();
+ const { client, prepareRuntime } = localClient(
+ {
+ codexOverrides: {
+ model: "openai.gpt-5.6-cyber",
+ model_provider: "amazon-bedrock",
+ },
+ },
+ environment,
+ );
+ await expect(
+ client.preflight(repository, { maxCostUsd: 1 }),
+ ).rejects.toThrow("cost limit is not available for the configured model");
+ await expect(client.preflight(repository)).resolves.toMatchObject({
+ model: "openai.gpt-5.6-cyber",
+ });
+ expect(prepareRuntime).not.toHaveBeenCalled();
+ await client.close();
+ },
+ );
+
test.each(["standard", "deep"] as const)(
"identifies Bedrock account advisory applicability in the %s parent prompt",
async (mode) => {
diff --git a/sdk/typescript/tests-ts/bedrock-oss-pricing.test.ts b/sdk/typescript/tests-ts/bedrock-oss-pricing.test.ts
new file mode 100644
index 000000000..57e46b217
--- /dev/null
+++ b/sdk/typescript/tests-ts/bedrock-oss-pricing.test.ts
@@ -0,0 +1,53 @@
+import { expect, test } from "bun:test";
+import { estimateScanCost } from "../src/cost-model.js";
+
+test.each([
+ ["openai.gpt-oss-20b-1:0", "ap-southeast-2", 0.0721, 0.309],
+ ["openai.gpt-oss-20b", "us-east-1", 0.07, 0.3],
+ ["openai.gpt-oss-20b-1:0", "eu-west-2", 0.11, 0.47],
+ ["openai.gpt-oss-safeguard-20b", "ap-southeast-2", 0.0721, 0.206],
+ ["openai.gpt-oss-safeguard-120b", "ap-southeast-2", 0.1545, 0.618],
+ ["openai.gpt-oss-120b", "us-gov-west-1", 0.18, 0.72],
+ ["us-gov.openai.gpt-oss-120b-1:0", "us-gov-east-1", 0.18, 0.72],
+ ["us-gov.openai.gpt-oss-20b-1:0", null, 0.084, 0.36],
+] as const)(
+ "prices Standard tokens for %s in %s",
+ (model, region, input, output) => {
+ const cost = estimateScanCost(
+ model,
+ {
+ input_tokens: 3,
+ cached_input_tokens: 1,
+ cache_write_input_tokens: 1,
+ output_tokens: 1,
+ },
+ region,
+ );
+ expect(cost?.pricing?.usdPerMillionTokens).toEqual({
+ input,
+ cacheRead: input,
+ cacheWrite: input,
+ output,
+ });
+ expect(cost?.estimatedUsd).toBeCloseTo(
+ (3 * input + output) / 1_000_000,
+ 16,
+ );
+ expect(cost?.estimatedUsdRange?.max).toBe(cost?.estimatedUsd);
+ },
+);
+
+test.each([
+ ["openai.gpt-oss-20b", undefined],
+ ["openai.gpt-oss-20b", null],
+ ["openai.gpt-oss-20b-1:0", "ap-southeast-4"],
+ ["openai.gpt-oss-safeguard-20b", "us-gov-east-1"],
+ ["us-gov.openai.gpt-oss-20b-1:0", "us-west-2"],
+ ["us-gov.openai.gpt-oss-20b", "us-gov-east-1"],
+ ["global.openai.gpt-oss-120b-1:0", "us-east-1"],
+ ["openai.gpt-oss-safeguard-120b-1:0", "us-east-1"],
+] as const)("keeps %s in %s unpriced", (model, region) => {
+ expect(
+ estimateScanCost(model, { input_tokens: 1, output_tokens: 1 }, region),
+ ).toBeNull();
+});
diff --git a/sdk/typescript/tests-ts/config.test.ts b/sdk/typescript/tests-ts/config.test.ts
index 7e710fa84..a6558936c 100644
--- a/sdk/typescript/tests-ts/config.test.ts
+++ b/sdk/typescript/tests-ts/config.test.ts
@@ -11,6 +11,7 @@ import {
modelProviderConfigOverride,
resolveCodexProfile,
resolveCommandAuthConfig,
+ scanPricingContext,
scanModelConfiguration,
scanModelProvider,
} from "../src/config.js";
@@ -637,6 +638,74 @@ describe("Codex configuration", () => {
}
});
+ test.each(["amazon-bedrock", "amazon-bedrock-runtime"])(
+ "uses the selected %s provider's AWS region before the environment",
+ (provider) => {
+ const config = {
+ profile: "gov",
+ model_provider: "openai",
+ model_providers: { [provider]: { aws: { region: "us-east-2" } } },
+ profiles: {
+ gov: {
+ model_provider: provider,
+ model_providers: {
+ [provider]: { aws: { region: " us-gov-east-1 " } },
+ },
+ },
+ },
+ };
+ expect(scanPricingContext(config, { AWS_REGION: "us-west-1" })).toEqual({
+ provider: "amazon-bedrock",
+ region: "us-gov-east-1",
+ });
+ },
+ );
+
+ test("resolves Bedrock environment regions and preserves an unknown region", () => {
+ const config = { model_provider: "amazon-bedrock" };
+ const environment = {
+ AWS_REGION: "us-gov-west-1",
+ AWS_DEFAULT_REGION: "us-east-1",
+ };
+ expect(scanPricingContext(config, environment)).toEqual({
+ provider: "amazon-bedrock",
+ region: "us-gov-west-1",
+ });
+ expect(
+ scanPricingContext(config, { ...environment, AWS_REGION: " " }),
+ ).toEqual({ provider: "amazon-bedrock", region: "us-east-1" });
+ expect(scanPricingContext(config, { AWS_PROFILE: "synthetic" })).toEqual({
+ provider: "amazon-bedrock",
+ region: null,
+ });
+ expect(
+ scanPricingContext({ model_provider: "openai" }, environment),
+ ).toBeUndefined();
+ });
+
+ test.each([
+ ["https://bedrock-mantle.us-gov-east-1.api.aws/openai/v1", "us-gov-east-1"],
+ [
+ "https://bedrock-runtime-fips.us-gov-west-1.amazonaws.com",
+ "us-gov-west-1",
+ ],
+ ["https://gateway.example.test", null],
+ ])(
+ "uses the billing region from a Bedrock endpoint override %s",
+ (baseUrl, region) => {
+ const config = {
+ model_provider: "amazon-bedrock",
+ model_providers: {
+ "amazon-bedrock": { base_url: baseUrl, aws: { region: "us-east-1" } },
+ },
+ };
+ expect(scanPricingContext(config, { AWS_REGION: "us-west-2" })).toEqual({
+ provider: "amazon-bedrock",
+ region,
+ });
+ },
+ );
+
test("rejects invalid model settings from the selected Codex profile", async () => {
for (const [profile, message] of [
[{ model: " " }, "model must be a nonempty string"],
diff --git a/sdk/typescript/tests-ts/cost-context.test.ts b/sdk/typescript/tests-ts/cost-context.test.ts
index 2e10b1895..70ce0045e 100644
--- a/sdk/typescript/tests-ts/cost-context.test.ts
+++ b/sdk/typescript/tests-ts/cost-context.test.ts
@@ -20,20 +20,18 @@ test.each([
"%s bounds each token category with verified rates",
(model, short, long) => {
for (const category of [0, 1, 2, 3] as const) {
- for (const selected of [model, `openai.${model}`]) {
- const cost = estimateScanCost(selected, {
- input_tokens: category === 3 ? 0 : 1_000_000,
- cached_input_tokens: category === 1 ? 1_000_000 : 0,
- cache_write_input_tokens: category === 2 ? 1_000_000 : 0,
- output_tokens: category === 3 ? 1_000_000 : 0,
- })!;
- expect(cost.estimatedUsd).toBe(short[category]);
- expect(cost.estimatedUsdRange).toEqual({
- min: short[category],
- max: long[category],
- context: "unknown",
- });
- }
+ const cost = estimateScanCost(model, {
+ input_tokens: category === 3 ? 0 : 1_000_000,
+ cached_input_tokens: category === 1 ? 1_000_000 : 0,
+ cache_write_input_tokens: category === 2 ? 1_000_000 : 0,
+ output_tokens: category === 3 ? 1_000_000 : 0,
+ })!;
+ expect(cost.estimatedUsd).toBe(short[category]);
+ expect(cost.estimatedUsdRange).toEqual({
+ min: short[category],
+ max: long[category],
+ context: "unknown",
+ });
}
},
);
@@ -62,8 +60,8 @@ test("prices the exact Bedrock Daybreak Blue ID with AWS short and long context
estimatedUsdRange: { min: 6.138, max: 11.176, context: "unknown" },
pricing: {
source:
- "https://docs.aws.amazon.com/en_en/bedrock/latest/userguide/model-card-openai-gpt-daybreak-blue-56-sol.html",
- asOf: "2026-10-01",
+ "https://docs.aws.amazon.com/bedrock/latest/userguide/model-card-openai-gpt-daybreak-blue-56-sol.html",
+ asOf: "2026-11-05",
serviceTier: "standard",
context: "short",
usdPerMillionTokens: {
@@ -135,8 +133,8 @@ test("prices the exact Bedrock Daybreak Red ID without inventing long-context ra
estimatedUsdRange: { min: 20.55625, max: null, context: "unknown" },
pricing: {
source:
- "https://docs.aws.amazon.com/en_en/bedrock/latest/userguide/model-card-openai-gpt-56-cyber.html",
- asOf: "2026-10-01",
+ "https://docs.aws.amazon.com/bedrock/latest/userguide/model-card-openai-gpt-56-cyber.html",
+ asOf: "2026-11-05",
serviceTier: "standard",
context: "short",
usdPerMillionTokens: {
@@ -162,6 +160,69 @@ test.each([1, 3])(
},
);
+// USD per million tokens in the AWS Standard short-context tables.
+test.each([
+ ["us.openai.gpt-5.4", "us-east-2", 2.75, 0.275, 2.75, 16.5],
+ ["us.openai.gpt-5.5", "us-east-1", 5.5, 0.55, 5.5, 33],
+ ["openai.gpt-5.4", "us-gov-west-1", 3.375, 0.3375, 3.375, 20.25],
+ ["openai.gpt-5.6-cyber", "us-gov-east-1", 16.875, 1.6875, 21.09375, 101.25],
+ ["openai.gpt-5.6-terra", "us-gov-west-1", 2.7, 0.27, 3.375, 16.2],
+ ["us-gov.openai.gpt-5.6-luna", "us-gov-east-1", 0.27, 0.027, 0.3375, 1.62],
+ ["openai.gpt-6.1-sol", "us-east-1", 2.2, 0.11, 2.75, 11],
+ ["us.openai.gpt-6.1-sol", "us-east-1", 2.2, 0.11, 2.75, 11],
+ ["global.openai.gpt-6.1-sol", "us-east-1", 2, 0.1, 2.5, 10],
+] as const)(
+ "uses the Standard rates for %s in %s",
+ (model, region, input, cacheRead, cacheWrite, output) => {
+ const cost = estimateScanCost(
+ model,
+ {
+ input_tokens: 3,
+ cached_input_tokens: 1,
+ cache_write_input_tokens: 1,
+ output_tokens: 1,
+ },
+ region,
+ );
+ expect(cost?.pricing?.usdPerMillionTokens).toEqual({
+ input,
+ cacheRead,
+ cacheWrite,
+ output,
+ });
+ expect(cost?.estimatedUsd).toBeCloseTo(
+ (input + cacheRead + cacheWrite + output) / 1_000_000,
+ 15,
+ );
+ },
+);
+
+test("keeps unverified Bedrock IDs, regions, and long-context rates unpriced", () => {
+ const usage = { input_tokens: 1_000_000, output_tokens: 0 };
+ for (const [model, region] of [
+ ["global.openai.gpt-5.5", "us-east-1"],
+ ["openai.gpt-6-cyber", "us-east-2"],
+ ["gpt-5.6-cyber", "us-gov-east-1"],
+ ["openai.gpt-5.5-2026-04-23", "us-east-1"],
+ ["openai.gpt-5.6-cyber", "us-gov-west-1"],
+ ["openai.gpt-daybreak-blue-5.6-sol", "us-gov-east-1"],
+ ["global.openai.gpt-5.6-terra", "us-gov-west-1"],
+ ] as const) {
+ expect(estimateScanCost(model, usage, region)).toBeNull();
+ }
+ expect(
+ estimateScanCost("openai.gpt-5.4", usage, "us-gov-west-1")
+ ?.estimatedUsdRange,
+ ).toEqual({ min: 3.375, max: null, context: "unknown" });
+ expect(
+ estimateScanCost(
+ "openai.gpt-5.6-luna",
+ { ...usage, cache_write_input_tokens: 0 },
+ "us-gov-west-1",
+ )?.estimatedUsdRange,
+ ).toEqual({ min: 0.27, max: 0.54, context: "unknown" });
+});
+
test("reports a range for cache-heavy scans without changing the budget baseline", () => {
const cost = estimateScanCost("gpt-5.6-sol", {
input_tokens: 150_000_000,
diff --git a/sdk/typescript/tests-ts/cost-model.property.test.ts b/sdk/typescript/tests-ts/cost-model.property.test.ts
index aa9297265..7ed0a495e 100644
--- a/sdk/typescript/tests-ts/cost-model.property.test.ts
+++ b/sdk/typescript/tests-ts/cost-model.property.test.ts
@@ -13,6 +13,9 @@ const rates = [
["gpt-5.6-sol", 4000n, 400n, 5000n, 20000n],
["gpt-5.6-terra", 2000n, 200n, 2500n, 12000n],
["gpt-5.6-luna", 200n, 20n, 250n, 1200n],
+ ["openai.gpt-6-sol", 2200n, 220n, 2750n, 11000n],
+ ["openai.gpt-6.1-sol", 2200n, 110n, 2750n, 11000n],
+ ["global.openai.gpt-6-sol", 2000n, 200n, 2500n, 10000n],
["openai.gpt-daybreak-blue-5.6-sol", 4400n, 440n, 5500n, 22000n],
] as const;
const count = fc.integer({ min: 0, max: 1_000_000_000 });
@@ -49,7 +52,7 @@ describe("cost-model invariants", () => {
expect(formatUsd(1234.5)).toBe("$1,234.50");
});
- test("matches an integer nanodollar oracle for every priced model", () => {
+ test("matches an integer nanodollar oracle for direct and Bedrock models", () => {
fc.assert(
fc.property(usageParts, (parts) => {
const tokens = usage(parts);
@@ -69,19 +72,14 @@ describe("cost-model invariants", () => {
BigInt(parts.cached) * cachedRate +
BigInt(parts.written) * writeRate +
BigInt(parts.output) * outputRate;
- const identifiers = model.startsWith("openai.")
- ? [model]
- : [model, `openai.${model}`];
- for (const selected of identifiers) {
- expect(estimateScanCost(selected, tokens)).toMatchObject({
- model: selected,
- inputTokens: tokens.input_tokens,
- cachedInputTokens: parts.cached,
- cacheWriteInputTokens: parts.written,
- outputTokens: parts.output,
- estimatedUsd: Number(nanos) / 1_000_000_000,
- });
- }
+ expect(estimateScanCost(model, tokens)).toMatchObject({
+ model,
+ inputTokens: tokens.input_tokens,
+ cachedInputTokens: parts.cached,
+ cacheWriteInputTokens: parts.written,
+ outputTokens: parts.output,
+ estimatedUsd: Number(nanos) / 1_000_000_000,
+ });
}
}),
propertyOptions,
diff --git a/sdk/typescript/tests-ts/cost.test.ts b/sdk/typescript/tests-ts/cost.test.ts
index 238db1766..649c48608 100644
--- a/sdk/typescript/tests-ts/cost.test.ts
+++ b/sdk/typescript/tests-ts/cost.test.ts
@@ -2292,45 +2292,43 @@ describe("scan cost", () => {
},
});
});
- test.each(["gpt-6-sol", "openai.gpt-6-sol"])(
- "prices every token category and both context ranges for %s",
- (model) => {
- expect(
- estimateScanCost(model, {
- input_tokens: 1_000_000,
- cached_input_tokens: 200_000,
- cache_write_input_tokens: 300_000,
- output_tokens: 100_000,
- }),
- ).toEqual({
- model,
- inputTokens: 1_000_000,
- cachedInputTokens: 200_000,
- cacheWriteInputTokens: 300_000,
- outputTokens: 100_000,
- estimatedUsd: 2.79,
- estimatedUsdRange: { min: 2.79, max: 5.08, context: "unknown" },
- pricing: {
- source: "https://developers.openai.com/api/docs/models/gpt-6-sol",
- asOf: "2026-10-10",
- serviceTier: "standard",
- context: "short",
- usdPerMillionTokens: {
- input: 2,
- cacheRead: 0.2,
- cacheWrite: 2.5,
- output: 10,
- },
- longContextUsdPerMillionTokens: {
- input: 4,
- cacheRead: 0.4,
- cacheWrite: 5,
- output: 15,
- },
+ test("prices every token category and both context ranges for gpt-6-sol", () => {
+ const model = "gpt-6-sol";
+ expect(
+ estimateScanCost(model, {
+ input_tokens: 1_000_000,
+ cached_input_tokens: 200_000,
+ cache_write_input_tokens: 300_000,
+ output_tokens: 100_000,
+ }),
+ ).toEqual({
+ model,
+ inputTokens: 1_000_000,
+ cachedInputTokens: 200_000,
+ cacheWriteInputTokens: 300_000,
+ outputTokens: 100_000,
+ estimatedUsd: 2.79,
+ estimatedUsdRange: { min: 2.79, max: 5.08, context: "unknown" },
+ pricing: {
+ source: "https://developers.openai.com/api/docs/models/gpt-6-sol",
+ asOf: "2026-10-10",
+ serviceTier: "standard",
+ context: "short",
+ usdPerMillionTokens: {
+ input: 2,
+ cacheRead: 0.2,
+ cacheWrite: 2.5,
+ output: 10,
},
- });
- },
- );
+ longContextUsdPerMillionTokens: {
+ input: 4,
+ cacheRead: 0.4,
+ cacheWrite: 5,
+ output: 15,
+ },
+ },
+ });
+ });
test.each([
[{ cache_write_tokens: 15 }, 15],
[{ cache_write_input_tokens: 0, cache_write_tokens: 15 }, 15],
diff --git a/sdk/typescript/tests-ts/scan-dashboard.test.ts b/sdk/typescript/tests-ts/scan-dashboard.test.ts
index 83f16b329..b549b8f8e 100644
--- a/sdk/typescript/tests-ts/scan-dashboard.test.ts
+++ b/sdk/typescript/tests-ts/scan-dashboard.test.ts
@@ -732,21 +732,35 @@ describe("live scan dashboard", () => {
});
test.each([
- ["unknown-model", false],
- ["gpt-5.6-cyber", true],
- ] as const)("shows a cost row only when %s has pricing", (model, priced) => {
- const stderr = capture(true);
- const dashboard = createDashboard(stderr.stream, {
- model: { model, reasoningEffort: "xhigh" },
- showCost: true,
- });
- dashboard.start();
- const frame = lastFrame(stderr);
- expect(/^\s*COST\b/m.test(frame)).toBe(priced);
- if (priced) expect(frame).toMatch(/COST\s+waiting for usage/);
- expect(frame).not.toContain("model pricing missing");
- dashboard.stop();
- });
+ ["unknown-model", undefined, false],
+ ["gpt-5.6-cyber", undefined, true],
+ [
+ "openai.gpt-5.6-cyber",
+ { provider: "amazon-bedrock", region: "us-gov-east-1" },
+ true,
+ ],
+ [
+ "openai.gpt-5.6-cyber",
+ { provider: "amazon-bedrock", region: "us-gov-west-1" },
+ false,
+ ],
+ ] as const)(
+ "shows a cost row only when %s has pricing in %j",
+ (model, pricingContext, priced) => {
+ const stderr = capture(true);
+ const dashboard = createDashboard(stderr.stream, {
+ model: { model, reasoningEffort: "xhigh" },
+ pricingContext,
+ showCost: true,
+ });
+ dashboard.start();
+ const frame = lastFrame(stderr);
+ expect(/^\s*COST\b/m.test(frame)).toBe(priced);
+ if (priced) expect(frame).toMatch(/COST\s+waiting for usage/);
+ expect(frame).not.toContain("model pricing missing");
+ dashboard.stop();
+ },
+ );
test("keeps cost bounds and assumptions readable on a narrow terminal", () => {
const stderr = capture(true);