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(pricing): remove Daybreak latest alias pricing (#2165)

Michael D'Angelo25b98eed85df
diff --git a/plugins/codex-security/mcp-app/tests/test_deep_scan_executor.ts b/plugins/codex-security/mcp-app/tests/test_deep_scan_executor.ts
index b81bf6351..35022da9d 100644
--- a/plugins/codex-security/mcp-app/tests/test_deep_scan_executor.ts
+++ b/plugins/codex-security/mcp-app/tests/test_deep_scan_executor.ts
@@ -1227,7 +1227,7 @@ env_key = "CODEX_API_KEY"
`CAPTURE_SYNTHETIC_OPENAI_AUTH\n${virtualenv === undefined ? "" : "CAPTURE_SYNTHETIC_PYTHON\n"}`,
);
const executor = new CodexSdkWorkerExecutor({
- model: "gpt-daybreak-blue-latest",
+ model: "gpt-6.1-sol",
parentSandbox: trustedParentSandbox,
});
for (const kind of ["discovery", "dedup"] as const) {
@@ -1250,11 +1250,7 @@ env_key = "CODEX_API_KEY"
runtimeEnvironment.LD_LIBRARY_PATH,
);
assert.equal(invocation.codexHome, fixture.root);
- assertFlagPair(
- invocation.argv,
- "--model",
- "gpt-daybreak-blue-latest",
- );
+ assertFlagPair(invocation.argv, "--model", "gpt-6.1-sol");
assert.deepEqual(invocation.runtimeEnvironment, runtimeEnvironment);
if (virtualenv !== undefined) {
assert.equal(invocation.pythonPrefix, virtualenv);
diff --git a/plugins/codex-security/scripts/workbench_finding_workflows.py b/plugins/codex-security/scripts/workbench_finding_workflows.py
index 010b87d71..28ca98751 100644
--- a/plugins/codex-security/scripts/workbench_finding_workflows.py
+++ b/plugins/codex-security/scripts/workbench_finding_workflows.py
@@ -12,6 +12,7 @@ from pathlib import Path
from typing import Any
sys.path.insert(0, str(Path(__file__).resolve().parent))
+from workbench_scan_usage import stored_scan_cost_fields
from workbench_target import (
UnreadableLocalFile,
UnsupportedLocalFileType,
@@ -19,6 +20,7 @@ from workbench_target import (
git_bytes,
git_revision,
)
+from workbench_validation import _valid_legacy_scan_cost
WORKFLOW_BINDINGS = {
"repositoryPath": "repository_path",
@@ -31,13 +33,80 @@ WORKFLOW_BINDINGS = {
WORKFLOW_STAGES = ("scan", "publish", "dedupe")
+def _usable_cost_range(value: Any) -> bool:
+ if not isinstance(value, dict) or value.get("context") != "unknown" or "max" not in value:
+ return False
+ minimum, maximum = value.get("min"), value["max"]
+ return (
+ type(minimum) in (int, float)
+ and 0 <= minimum <= sys.float_info.max
+ and (
+ maximum is None
+ or type(maximum) in (int, float)
+ and minimum <= maximum <= sys.float_info.max
+ )
+ )
+
+
+def _recorded_cost(value: str | None) -> dict[str, Any] | None:
+ """Recover the SDK cost fields that are usable from this historical receipt."""
+ try:
+ cost = stored_scan_cost_fields(value).get("cost")
+ except (TypeError, ValueError, RecursionError):
+ return None
+ if not _valid_legacy_scan_cost(cost):
+ return None
+
+ fields = (
+ "model",
+ "inputTokens",
+ "cachedInputTokens",
+ "cacheWriteInputTokens",
+ "outputTokens",
+ "estimatedUsd",
+ )
+ result = {field: cost[field] for field in fields}
+ if isinstance(cost.get("cacheWriteInputTokensReported"), bool):
+ result["cacheWriteInputTokensReported"] = cost["cacheWriteInputTokensReported"]
+ bounds = cost.get("estimatedUsdRange")
+ if _usable_cost_range(bounds):
+ result["estimatedUsdRange"] = {field: bounds[field] for field in ("min", "max", "context")}
+
+ pricing = cost.get("pricing")
+ if (
+ isinstance(pricing, dict)
+ and pricing.get("serviceTier") == "standard"
+ and pricing.get("context") == "short"
+ and isinstance(pricing.get("source"), str)
+ and isinstance(pricing.get("asOf"), str)
+ ):
+ details = {field: pricing[field] for field in ("source", "asOf", "serviceTier", "context")}
+ rate_keys = ("input", "cacheRead", "cacheWrite", "output")
+ for field in ("usdPerMillionTokens", "longContextUsdPerMillionTokens"):
+ rates = pricing.get(field)
+ if isinstance(rates, dict) and all(
+ type(rates.get(key)) in (int, float) and 0 <= rates[key] <= sys.float_info.max
+ for key in rate_keys
+ ):
+ details[field] = {key: rates[key] for key in rate_keys}
+ if "usdPerMillionTokens" in details:
+ result["pricing"] = details
+ return result
+
+
def read_workflow(connection: sqlite3.Connection, workflow_id: str) -> dict[str, Any] | None:
row = connection.execute(
- "SELECT * FROM finding_workflows WHERE id = ?", (workflow_id,)
+ """SELECT w.*, s.cost_json FROM finding_workflows AS w
+ LEFT JOIN scans AS s ON s.id = w.scan_id AND s.scan_dir = w.scan_dir
+ AND s.status = 'complete'
+ WHERE w.id = ?""",
+ (workflow_id,),
).fetchone()
if row is None:
return None
state = {"id": row["id"], "stages": {}}
+ if (cost := _recorded_cost(row["cost_json"])) is not None:
+ state["recordedCost"] = cost
for field, column in WORKFLOW_BINDINGS.items():
if row[column] is not None:
state[field] = row[column]
diff --git a/sdk/typescript/docs/cli.md b/sdk/typescript/docs/cli.md
index 002dec82e..8227cf17c 100644
--- a/sdk/typescript/docs/cli.md
+++ b/sdk/typescript/docs/cli.md
@@ -170,7 +170,8 @@ and project have access. An omitted `access_programs.cyber` field does not by
itself mean that Daybreak treatment is disabled.
Choose a compatible model available to your project and select Daybreak Blue
-explicitly:
+explicitly. If you use `--max-cost` or the SDK's `maxCostUsd`, choose a model
+with built-in pricing so the scan can enforce the limit.
```bash
codex-security scan /path/to/repository \
@@ -183,11 +184,8 @@ Use `--mode deep` in the CLI or `mode: "deep"` in SDK scan options for Deep Scan
The selection reaches discovery and reducer workers, including resumed workers. See [Select a Cyber access program](#select-a-cyber-access-program)
for configuration precedence and the API-key feature setting.
-The previously documented `gpt-daybreak-blue-latest` alias is
-[deprecated](https://developers.openai.com/api/docs/models/gpt-daybreak-blue-latest).
-Use a current model available to your project. `access_programs.cyber` is a
-Responses API field; use the supported selector instead of passing it as a
-`--codex` configuration key.
+`access_programs.cyber` is a Responses API field. Use `--cyber-access-program`
+instead of passing it as a `--codex` configuration key.
### Select a Cyber access program
diff --git a/sdk/typescript/src/api.ts b/sdk/typescript/src/api.ts
index c4015be5a..077c268db 100644
--- a/sdk/typescript/src/api.ts
+++ b/sdk/typescript/src/api.ts
@@ -823,13 +823,16 @@ export class CodexSecurity {
expectedScanId: state.scanId,
signal,
});
- await workflow.bind({ artifactDigest: scanArtifactDigest(contract) });
+ const verified = await workflow.bind({
+ artifactDigest: scanArtifactDigest(contract),
+ });
metadata ??= { threadId: "", turnResult: { status: "completed" } };
await workflow.complete("scan", metadata);
return new ScanResult({
...contract,
scanDir: state.scanDir,
...metadata,
+ recordedCost: verified.recordedCost,
threatModelPath: await readThreatModelPath(state.scanDir, {
pythonPath: this.config.pythonPath,
protectedRoot: local.protectedRoot,
diff --git a/sdk/typescript/src/cost-model.ts b/sdk/typescript/src/cost-model.ts
index fa2fa63ea..34853147e 100644
--- a/sdk/typescript/src/cost-model.ts
+++ b/sdk/typescript/src/cost-model.ts
@@ -93,15 +93,13 @@ const MODEL_PRICING_NANODOLLARS: Readonly<Record<string, ModelPricing>> = {
"gpt-5.6-luna": [200, 20, 250, 1_200],
// https://developers.openai.com/api/docs/pricing#cyber-models
"gpt-5.6-cyber": [12_500, 1_250, 15_625, 75_000],
- "gpt-daybreak-blue-latest": [4_000, 400, 5_000, 20_000],
- "gpt-daybreak-red-latest": [12_500, 1_250, 15_625, 75_000],
};
// Verified Standard rates: https://developers.openai.com/api/docs/pricing
// GPT-5.5: https://developers.openai.com/api/docs/models/gpt-5.5
// GPT-6 Sol: https://developers.openai.com/api/docs/models/gpt-6-sol
// Do not infer tiers from aggregate scan tokens: the runtime does not report
-// which usage received long-context pricing. Cyber/Daybreak Red has no verified
+// which usage received long-context pricing. GPT-5.6-Cyber has no verified
// long-context rate in the pricing table, so its upper estimate stays unavailable.
const LONG_CONTEXT_PRICING_NANODOLLARS: Readonly<Record<string, ModelPricing>> =
{
@@ -115,7 +113,6 @@ const LONG_CONTEXT_PRICING_NANODOLLARS: Readonly<Record<string, ModelPricing>> =
"gpt-5.6-sol": [8_000, 800, 10_000, 30_000],
"gpt-5.6-terra": [4_000, 400, 5_000, 18_000],
"gpt-5.6-luna": [400, 40, 500, 1_800],
- "gpt-daybreak-blue-latest": [8_000, 800, 10_000, 30_000],
};
function usdPerMillionTokens(pricing: ModelPricing, unitsPerUsd: number) {
diff --git a/sdk/typescript/src/finding-workflow.ts b/sdk/typescript/src/finding-workflow.ts
index 1efeaba35..06d0d19a5 100644
--- a/sdk/typescript/src/finding-workflow.ts
+++ b/sdk/typescript/src/finding-workflow.ts
@@ -2,6 +2,7 @@ import { createHash } from "node:crypto";
import { isAbsolute, relative, sep } from "node:path";
import type { JsonObject } from "./config.js";
import type { LoadedContract } from "./contract.js";
+import type { ScanCost } from "./cost-model.js";
import { CodexSecurityError, errorMessage } from "./errors.js";
import { findingsBaseUrl } from "./findings-client.js";
import type { FindingSearchScope } from "./finding-retrieval.js";
@@ -27,6 +28,7 @@ export interface WorkflowBinding {
}
export interface WorkflowState extends WorkflowBinding {
id: string;
+ recordedCost?: ScanCost;
stages: Record<
WorkflowStage,
{
diff --git a/sdk/typescript/src/result.ts b/sdk/typescript/src/result.ts
index e2738328f..d413ea336 100644
--- a/sdk/typescript/src/result.ts
+++ b/sdk/typescript/src/result.ts
@@ -44,6 +44,8 @@ export interface ScanResultOptions {
scanDir: string;
threadId: string;
turnResult: TurnResultMetadata;
+ /** Cost saved when this scan completed. */
+ recordedCost?: Readonly<ScanCost>;
sarifPath?: string | null;
threatModelPath?: string | null;
repositoryFindings?: readonly RepositoryFinding[];
@@ -84,10 +86,9 @@ export class ScanResult {
this.ownerSuggestionsPath = options.ownerSuggestionsPath;
this.ownerSuggestionsError = options.ownerSuggestionsError;
this.threatModelPath = options.threatModelPath ?? null;
- this.cost = estimateScanCost(
- options.turnResult.model,
- options.turnResult.usage,
- );
+ this.cost =
+ options.recordedCost ??
+ estimateScanCost(options.turnResult.model, options.turnResult.usage);
if (options.sarifPath !== undefined) {
this.sarifPath = options.sarifPath;
} else {
diff --git a/sdk/typescript/tests-ts/api-attribution-concurrency.test.ts b/sdk/typescript/tests-ts/api-attribution-concurrency.test.ts
index d32313749..d530487e2 100644
--- a/sdk/typescript/tests-ts/api-attribution-concurrency.test.ts
+++ b/sdk/typescript/tests-ts/api-attribution-concurrency.test.ts
@@ -51,7 +51,7 @@ describe("delegated scan attribution", () => {
(["cli", "sdk"] as const).map(async (surface) => {
const model = selectModel
? surface === "cli"
- ? "gpt-daybreak-blue-latest"
+ ? "gpt-6.1-sol"
: "gpt-5.6-sol"
: undefined;
const program = surface === "cli" ? programs[0] : programs[1];
@@ -371,7 +371,7 @@ describe("delegated scan attribution", () => {
expect(
await client.preflight(repository, { mode }),
).toMatchObject({
- model: index === 0 ? "gpt-daybreak-blue-latest" : "gpt-5.6-sol",
+ model: index === 0 ? "gpt-6.1-sol" : "gpt-5.6-sol",
authentication: { method: "api_key", verified: false },
});
}),
diff --git a/sdk/typescript/tests-ts/api-workflow.test.ts b/sdk/typescript/tests-ts/api-workflow.test.ts
index 2ea1a50ff..8ebffedf2 100644
--- a/sdk/typescript/tests-ts/api-workflow.test.ts
+++ b/sdk/typescript/tests-ts/api-workflow.test.ts
@@ -1,5 +1,6 @@
import { mkdir, readFile, rm, writeFile } from "node:fs/promises";
import { dirname, join } from "node:path";
+import { Database } from "bun:sqlite";
import { afterEach, expect, spyOn, test } from "bun:test";
import { runWorkbench } from "../src/runtime.js";
import * as runtime from "../src/runtime.js";
@@ -16,9 +17,15 @@ import {
scanRuntimeDependencies,
} from "./support/api-events.js";
import { createApiTestFixtures } from "./support/temporary-directories.js";
+import { writePublicationFindings } from "./support/sealed-scan.js";
+import {
+ scanRegistrationArguments,
+ workbenchCommand,
+} from "./support/workbench-command.js";
import { FindingWorkflow } from "../src/finding-workflow.js";
import { readKnowledgeBaseSnapshot } from "../src/knowledge-base.js";
import { DEFAULT_DEEP_SCAN_SETTINGS } from "../src/deep-scan-defaults.js";
+import { estimateScanCost } from "../src/cost-model.js";
const { cleanup, temporaryDirectory } = createApiTestFixtures();
afterEach(cleanup);
@@ -106,13 +113,16 @@ test.each([
await rm(promptFile);
return state;
}
- if (args[0] === "get-scan")
+ if (args[0] === "get-scan") {
+ if (scenario === "completed" && attempt === 2)
+ throw new Error("Synthetic unavailable scan history");
return {
scan: {
progress: { status: completed ? "complete" : "failed" },
continuationThreadId: "thread-1",
},
};
+ }
if (args[0] === "register-cli-scan") {
expect(JSON.parse(input!).workflowId).toBe(workflowId);
expect(JSON.parse(input!).recipe.ownerSuggestions).toBe(
@@ -221,6 +231,150 @@ test.each([
},
);
+const historicalRange = { min: 0.001, max: 0.002, context: "unknown" } as const;
+const historicalCost = {
+ model: "retired-test-model",
+ inputTokens: 10,
+ cachedInputTokens: 2,
+ cacheWriteInputTokens: 0,
+ outputTokens: 3,
+ estimatedUsd: 0.001,
+};
+const historicalPricing = {
+ source: "https://example.test/pricing",
+ asOf: "2026-01-01",
+ serviceTier: "standard",
+ context: "short",
+ usdPerMillionTokens: { input: 4, cacheRead: 0.4, cacheWrite: 5, output: 20 },
+} as const;
+const fullCost = {
+ ...historicalCost,
+ cacheWriteInputTokensReported: false,
+ estimatedUsdRange: historicalRange,
+ pricing: {
+ ...historicalPricing,
+ longContextUsdPerMillionTokens: {
+ input: 8,
+ cacheRead: 0.8,
+ cacheWrite: 10,
+ output: 30,
+ },
+ },
+};
+const openRangeCost = {
+ ...historicalCost,
+ estimatedUsdRange: { ...historicalRange, max: null },
+ pricing: historicalPricing,
+};
+const fallbackModel = "gpt-5.6-sol";
+const turnUsage = {
+ input_tokens: 10,
+ cached_input_tokens: 2,
+ output_tokens: 3,
+};
+test.each([
+ ["completed receipt", fullCost, fullCost],
+ [
+ "unknown upper bound",
+ {
+ ...openRangeCost,
+ pricing: { ...historicalPricing, longContextUsdPerMillionTokens: {} },
+ },
+ openRangeCost,
+ ],
+ [
+ "lost receipt",
+ { ...historicalCost, estimatedUsdRange: null },
+ historicalCost,
+ ],
+ [
+ "malformed optional fields",
+ {
+ ...historicalCost,
+ estimatedUsdRange: { min: 0.001, context: "unknown" },
+ pricing: {},
+ cacheWriteInputTokensReported: "false",
+ },
+ historicalCost,
+ ],
+ ["different scan directory", fullCost, null],
+ ["unreadable cost", "{", estimateScanCost(fallbackModel, turnUsage)],
+] as const)(
+ "replays historical cost only for the bound scan: %s",
+ async (scenario, saved, expected) => {
+ const root = await temporaryDirectory();
+ const repository = join(root, "repository");
+ await mkdir(repository);
+ const scanDir = join(root, "scan");
+ const localDir =
+ scenario === "different scan directory" ? join(root, "local") : scanDir;
+ await mkdir(localDir, { mode: 0o700 });
+ const environment = {
+ PATH: process.env["PATH"],
+ SystemRoot: process.env["SystemRoot"],
+ CODEX_SECURITY_STATE_DIR: join(root, "state"),
+ };
+ const workbench = workbenchCommand(
+ await runtime.resolvePluginPython({ environment }),
+ () => environment,
+ );
+ const registration = await workbench(
+ scanRegistrationArguments(repository, localDir),
+ );
+ const scanId = registration["scanId"] as string;
+ await copyCompletedScan(root);
+ await writePublicationFindings(scanDir, scanId, 1);
+ const storedCost =
+ scenario === "lost receipt"
+ ? {
+ usage: {
+ coverage: "unavailable",
+ source: "codex_rollout",
+ threadCount: 0,
+ },
+ cost: saved,
+ }
+ : saved;
+ using database = new Database(
+ join(environment.CODEX_SECURITY_STATE_DIR, "workbench.sqlite3"),
+ );
+ database.run("UPDATE scans SET status='complete', cost_json=? WHERE id=?", [
+ typeof storedCost === "string" ? storedCost : JSON.stringify(storedCost),
+ scanId,
+ ]);
+ const workflow = new FindingWorkflow("historical-cost", environment);
+ await workflow.bind({ repositoryPath: repository, scanId, scanDir });
+ await workflow.begin("scan");
+ if (scenario !== "lost receipt") {
+ await workflow.complete("scan", {
+ threadId: "thread-1",
+ turnResult: {
+ status: "completed",
+ model:
+ scenario === "unreadable cost"
+ ? fallbackModel
+ : historicalCost.model,
+ usage: turnUsage,
+ },
+ });
+ }
+ const client = TestClient.withDependencies({
+ environment,
+ runWorkbench: async (options, args, input) => {
+ if (args[0] === "get-scan") expect(scenario).toBe("lost receipt");
+ return runWorkbench(options, args, input);
+ },
+ });
+ try {
+ const result = await client.run(repository, { workflowId: workflow.id });
+ expect(result.cost).toEqual(expected);
+ expect(result.toJSON()["cost"]).toEqual(expected);
+ } finally {
+ await client.close();
+ }
+ },
+);
+
test.each(["ambient settings", "shipped defaults"])(
"a deep workflow can resume after changing %s but rejects a different request",
async (change) => {
diff --git a/sdk/typescript/tests-ts/cost-context.test.ts b/sdk/typescript/tests-ts/cost-context.test.ts
index 6d0346367..f859c8479 100644
--- a/sdk/typescript/tests-ts/cost-context.test.ts
+++ b/sdk/typescript/tests-ts/cost-context.test.ts
@@ -16,7 +16,6 @@ test.each([
["gpt-5.6-sol", [4, 0.4, 5, 20], [8, 0.8, 10, 30]],
["gpt-5.6-terra", [2, 0.2, 2.5, 12], [4, 0.4, 5, 18]],
["gpt-5.6-luna", [0.2, 0.02, 0.25, 1.2], [0.4, 0.04, 0.5, 1.8]],
- ["gpt-daybreak-blue-latest", [4, 0.4, 5, 20], [8, 0.8, 10, 30]],
] as const)(
"%s bounds each token category with verified rates",
(model, short, long) => {
@@ -39,6 +38,16 @@ test.each([
},
);
+test.each(["blue", "red"])(
+ "leaves the Daybreak %s latest alias unpriced",
+ (program) => {
+ const alias = `gpt-daybreak-${program}-latest`;
+ const usage = { input_tokens: 1_000_000, output_tokens: 0 };
+ expect(estimateScanCost(alias, usage)).toBeNull();
+ expect(estimateScanCost(`openai.${alias}`, usage)).toBeNull();
+ },
+);
+
test("prices the exact Bedrock Daybreak Blue ID with AWS short and long context rates", () => {
const usage = {
input_tokens: 1_000_000,
@@ -185,7 +194,7 @@ test("widens the maximum for unreported cache writes without changing token coun
});
test("keeps unverified long-context prices unavailable", () => {
- const cost = estimateScanCost("gpt-daybreak-red-latest", {
+ const cost = estimateScanCost("gpt-5.6-cyber", {
input_tokens: 1_000_000,
cache_write_input_tokens: 0,
output_tokens: 0,
diff --git a/sdk/typescript/tests-ts/cost-model.property.test.ts b/sdk/typescript/tests-ts/cost-model.property.test.ts
index 2067dc039..aa9297265 100644
--- a/sdk/typescript/tests-ts/cost-model.property.test.ts
+++ b/sdk/typescript/tests-ts/cost-model.property.test.ts
@@ -13,8 +13,6 @@ 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],
- ["gpt-daybreak-blue-latest", 4000n, 400n, 5000n, 20000n],
- ["gpt-daybreak-red-latest", 12500n, 1250n, 15625n, 75000n],
["openai.gpt-daybreak-blue-5.6-sol", 4400n, 440n, 5500n, 22000n],
] as const;
const count = fc.integer({ min: 0, max: 1_000_000_000 });