Plans you can read at a glance. • Plan names match everywhere: the subscription screen now says Scholar and Master, the same names shown on your plan badge and in limit messages. (The cards used to say Pro and Team.) • Every plan limit is a number. Home shows today's runs used and runs left, and the saved-circuit limit, as figures. • When you reach today's simulation limit, the app tells you how many runs your plan has, that the counter resets at midnight on your device, and which plan raises the limit. Saved circuits and results stay available. • Subscription Limits (Settings) no longer lists Journal, SQL Optimizer or Oracle rows: SwiftQuantum has none of these features. Saved circuits now sit under “Storage & Export”. • Numbers follow the language you chose in the app, and plan screens that still showed English text in other languages are translated. • The sign-in prompt states the real number: sign in for 10 simulations a day. • Clearer answers for large circuits: a circuit with non-Clifford gates above 25 qubits now stops on your device with a note on what to change, and nothing is sent. Server errors appear in your language. • Changing your password keeps you signed in on this device. • The business and privacy information screens now show our registered company details and where your data is stored. Thank you for building with us.

Numbers you can check. • Error correction: the logical error rate chart now says where each curve comes from. With our engine, it is a measured Monte Carlo. Offline, it shows measured distance-3 points from an on-device decoder (20,000 shots, 95% confidence interval) and dashed reference curves labelled as a formula. Nothing is made up to fill the chart. • Transpiler: the accuracy card shows the measured state overlap between your circuit and the converted code, or “Not verified” when it cannot be checked. The old estimated score is gone. • Benchmark (Home > Quick Actions): each engine reports its median run time, a 95% confidence interval and the number of runs, measured on your device, with a note on the device, OS and version for reproduction. The “N× faster” card is gone; ratios appear only between engines that were both measured. • AI to editor: gates that could not be applied are listed in a banner, and a sealed generation shows whether the editor still matches its record. • Honest wording: leftover mentions of quantum hardware and “127+ qubit computers” are removed. SwiftQuantum runs classical simulations — on your device, or on our engine service for the large circuits you choose to send. The in-app post-quantum cryptography is described as an educational reference, not a certified implementation. • Privacy: a circuit that fits on your device never leaves it, even after you agree to send a larger one in the same session. • Plans: prices on the plan screens now come from the App Store in your currency, and they no longer mention hardware access the app does not have. Choosing a locked engine in the editor opens the subscription screen, and Scholar and Master subscribers get every engine their plan includes. • Subscription Limits (Settings): every row now shows only its name in your language. Internal code names that appeared in parentheses next to the labels are gone. • The “Academy level” row is removed: it described courses SwiftQuantum does not offer. Journal and export rows now sit under “Storage & Export”, and the Pre-Flight Estimator row is listed with the compute limits. • Privacy requests: when you are signed in, a data deletion request and the California “Do Not Sell” choice are now sent to your account. EU content reports open an email to legal@swiftquantum.tech. • The usage-counter switch in the Coherence panel now explains that it is on by default and counts on your device. • Fixed a crash on the first launch after updating. Thank you for building with us.

Results you can trust, and privacy you can predict. • Every engine now computes exactly the circuit you drew. S, T, CZ, SWAP, Toffoli and Fredkin gates are never skipped — in shot mode too. • On the free plan, the Standard engine now runs your circuits on your device instead of our server. • A circuit is sent to our server only when it needs it — more than 25 qubits with non-Clifford gates — and only after you tap Send. Smaller circuits no longer show the prompt: they always run on your device. • On Team and Enterprise, the Turbo engine now runs large Clifford circuits (up to 127 qubits) on your device, just like HyperCore, instead of sending them to the server. • OpenQASM import now works: paste a program and get the circuit, with a clear message for anything unsupported. • Error correction (Team and Enterprise): surface-code experiments now run as a real Monte Carlo on our engine and show the shots, the decoder and a 95% confidence interval. • The Bloch sphere shows each qubit's actual state, including mixed states. The QFT example is now a real QFT, and circuit depth counts layers. • Subscriptions show App Store prices in your currency, and a purchase unlocks right away. • Sign-up now checks your email address on the spot and explains problems clearly (address already in use, invalid address, too many attempts). Thank you for building with us. Questions and ideas are always welcome.

New: on-device AI circuit generation — your words become circuits, entirely on your device. • Download a compact AI model once, then generate quantum circuits from plain language with no network at all. Your prompts never leave your device. • Every accepted result is sealed into a local, tamper-evident record you can inspect and verify in Settings — and records you sealed before keep verifying. • Honest by design: results carry an advisory confidence rating (low ratings included), and a seal proves a generation happened on your device — it is not a correctness score. • Off by default. Turn it on in Settings → On-Device Model; delete the download whenever you like. Also new: Q-Footprints — a record of what you actually ran. • Opt-in and off by default; your circuit itself is never sent — only the shape of the run. • See your activity, the gates you have exercised, and your two highest qubit counts, kept honestly separate. • A seal attests that a run happened. It is not a skill score and not a certification. Plus: deleting your account now removes your saved work everywhere it was stored, and subscribers get their full simulation limits on the web workspace. Thank you for building with us. Questions and ideas are always welcome.

Clearer and more honest. • Qubit capabilities are now stated precisely throughout the app: Clifford (stabilizer) circuits up to 127 qubits on-device, and general circuits up to 25 qubits — the numbers we actually measured on real hardware. • Terms and Privacy links on the subscription screen now open reliably. • Reliability and polish across the app. Thank you for building with us. Questions and ideas are always welcome.

No new features this time — this update is about the store page itself. SwiftQuantum is now listed in Korean, Japanese, Chinese, German, French and Spanish, not just English, so you can read about it in your own language before you download it. We also corrected our App Privacy details to describe exactly what the app does and doesn't collect, and cleaned up wording that was written for engineers rather than for you. Everything from v10.2.0 — on-device Clifford circuits up to 127 qubits, ask-before-server for larger general circuits — is unchanged and still there.

v10.2.0. Clifford circuits now run entirely on your device — up to 127 qubits, with no server and no network. Build a GHZ or stabilizer circuit and simulate it on-device in about two seconds. When a general (non-Clifford) circuit is too large to run locally, the app asks before sending anything to our servers — your circuit stays on your device unless you allow it. General circuits run on-device up to 25 qubits. Also in this update (since the last App Store release): the premium engines work end-to-end — quantum error-correction (QEC) decoding on Master and up, native Turbo and HyperCore engines, and Enterprise API keys and custom QEC settings that save and run reliably. Subscribe as a guest — signing in is optional and only adds cross-device sync. Plus a smarter AI Apply that drops gates straight onto your circuit, and clearer in-app upgrade prompts. Refined translations across all seven languages.

v10.1.5. The premium engines now work end-to-end. Run quantum error-correction (QEC) decoding on Master and up, run large circuits on the on-device native Turbo and HyperCore engines, and on Enterprise manage your API keys and custom QEC settings — everything now saves and runs reliably. You can subscribe as a guest: signing in is optional and only adds cross-device sync for your saved circuits. Plus dependable sign-in across every Apple device, a smarter AI Apply that drops gates straight onto your circuit, fresh AI suggestions on each refresh, and a one-minute first-run guide to Coherence (Live mode · Three candidates · Conductor · Time axis). This update also makes in-app upgrade prompts clearer, with UI polish throughout. Polished translations across all seven languages.

v10.1.4. The premium engines now work end-to-end. Run quantum error-correction (QEC) decoding on Master and up, push large circuits through the HyperCore engine, and on Enterprise manage your API keys and custom QEC settings — everything now saves and runs reliably. You can subscribe as a guest: signing in is optional and only adds cross-device sync for your saved circuits. Plus dependable sign-in across every Apple device, a smarter AI Apply that drops gates straight onto your circuit, fresh AI suggestions on each refresh, and a one-minute first-run guide to Coherence (Live mode · Three candidates · Conductor · Time axis). This update also makes in-app upgrade prompts clearer, with UI polish throughout. Polished translations across all seven languages.

v10.1.3. The premium engines now work end-to-end. Run quantum error-correction (QEC) decoding on Master and up, push large circuits through the HyperCore engine, and on Enterprise manage your API keys and custom QEC settings — everything now saves and runs reliably. You can subscribe as a guest: signing in is optional and only adds cross-device sync for your saved circuits. Plus dependable sign-in across every Apple device, a smarter AI Apply that drops gates straight onto your circuit, fresh AI suggestions on each refresh, and a one-minute first-run guide to Coherence (Live mode · Three candidates · Conductor · Time axis). Polished translations across all seven languages.

Reliable sign-in across every Apple device, a smarter AI Apply that actually drops gates onto your circuit, fresher AI suggestions on every refresh, and a first-run welcome guide that teaches Coherence (Live mode · Three candidates · Conductor · Time axis) in under a minute. Plus polished translations across all seven languages.

v9.4.5: - AI Circuit Generation — describe circuits in natural language, built on-device. Your data never leaves your device. - Siri Shortcuts — voice-control your quantum workflow in 7 languages - Liquid Glass UI — refined translucent design across canvas, panels, and toolbar - Post-quantum cryptography security tools - Enhanced stability and performance

v9.4.4: - AI Circuit Generation — describe circuits in natural language, built on-device. Your data never leaves your device. - Siri Shortcuts — voice-control your quantum workflow in 7 languages - Liquid Glass UI — refined translucent design across canvas, panels, and toolbar - Post-quantum cryptography security tools - Enhanced stability and performance KO

v9.4.2 — 5-Perspective Security Audit + L10n Hardening - Security: CORS domain whitelist, IDOR prevention (SHA-256), SSO Enterprise tier enforcement, clipboard 30s auto-clear - Auth: Subscription expiry enforcement, admin privilege escalation prevention - UX: Destructive action confirmation dialogs, timer leak fix - L10n: EnterpriseContactButton 6 hardcoded → L10n (7 keys × 7 languages = 49 translations) - Code: Force-unwrap elimination, input validation strengthening, Pydantic validators - Previous: macOS Guideline 4 + App Name Unification + PQC Migration (v9.3.7)

v9.4.1 — 5-Perspective Security Audit + L10n Hardening - Security: CORS domain whitelist, IDOR prevention (SHA-256), SSO Enterprise tier enforcement, clipboard 30s auto-clear - Auth: Subscription expiry enforcement, admin privilege escalation prevention - UX: Destructive action confirmation dialogs, timer leak fix - L10n: EnterpriseContactButton 6 hardcoded → L10n (7 keys × 7 languages = 49 translations) - Code: Force-unwrap elimination, input validation strengthening, Pydantic validators - Previous: macOS Guideline 4 + App Name Unification + PQC Migration (v9.3.7)

v9.2.4 - B2B Team Invite, Enterprise Fixes & QPU Dashboard UI - B2B Team Invite: Enterprise team member invitation card with 7-language localization - Team API 404 Fix: CodingKeys double-conversion removed - QPU Dashboard: Full-width card layout + 2-column grid + scroll bottom padding - Enterprise org_id fix: API Key, SSO, White-label views corrected - Session persistence: Login session maintained across app rebuilds - AI assistant hidden from UI - 5 Legacy Backend APIs added (enterprise_dashboard, qec, subscription, transpiler, vqe) - 127-qubit stabilizer tableau, QEC cloud history, endpoint resilience (v9.2.0-9.2.1)