Aug 19: Telecom hikes, the AI speed race, and a SIM trick that stops OTP theft

Reports say Jio and Airtel are lining up a 12–15% tariff hike over the next 3–4 months. Jio has quietly revived its ₹299 Prime plan just as Airtel pushed its base plan up to ₹349 — a small breather before the bigger jump. If you're on monthly recharges, locking in an annual pack now could save you the increase. 📉

The problem

India's telecom operators have been steadily raising prices to lift average revenue per user (ARPU), and every hike lands directly on prepaid users, who make up the bulk of the market.

What changed

Multiple reports say Jio and Airtel are planning a 12–15% tariff hike within the next 3–4 months. In the near term, Airtel removed its ₹299 prepaid option and raised its base unlimited plan to ₹349, while Jio brought back its ₹299 Prime membership plan to hold that price point.

The catch

The 12–15% figure is an industry/analyst projection, not an officially announced hike yet — exact plans and dates can still shift. Jio's ₹299 revival is real but positioned as a limited counter-move, not a permanent guarantee.

What it means for you

If you rely on monthly recharges, consider locking in a long-validity or annual pack now to ride out the hike at today's rates, and compare Jio vs Airtel base plans before your next recharge.

OpenAI just previewed an 'Ultrafast' mode for GPT-5.6 Sol running on Cerebras chips — around 750 tokens per second, up to 14x faster. Google, meanwhile, shipped Gemini 3.7 Flash just three weeks after 3.6 Flash. Near-instant, cheaper answers change how developers build with AI — but faster isn't the same as smarter, and this is still a dev preview, not something inside your everyday app yet. ⚡️

The problem

For most people AI chat still feels laggy — you wait while answers stream out token by token, which makes real-time and agentic uses clunky.

What changed

OpenAI opened a limited API preview of an 'Ultrafast' mode for GPT-5.6 Sol, powered by Cerebras hardware, delivering roughly 750 output tokens per second and up to 14x faster responses. Google separately released Gemini 3.7 Flash just three weeks after 3.6 Flash, pushing coding and automation benchmarks higher at low 'Flash'-tier pricing.

The catch

Ultrafast mode is a limited API preview for developers, not a feature in the consumer ChatGPT app — and raw speed doesn't make a model more accurate. The introductory pricing is promotional and time-limited too.

What it means for you

If you build with AI, watch for speed-tier options that make real-time and agentic apps viable; if you're just a user, expect snappier replies to trickle down, but keep judging models on answer quality, not just how fast they type.

Most people don't know Threads has a chronological feed. Tap the Threads (home) icon at the top and pick the 'Following' feed — that's pure reverse-chronological, only the people you actually follow, newest first. You can also pin your favourite custom feeds and swipe between them, so the app stops defaulting to the mystery 'For You' rabbit hole. 🧵

The problem

Threads opens on an algorithmic 'For You' feed that mixes in accounts you don't follow, so posts from the people you actually care about get buried.

What changed

Tapping the Threads home icon lets you switch to the 'Following' feed — a pure chronological timeline of only the accounts you follow. You can pin favourite and custom feeds and swipe between them so your preferred feed is one tap away.

The catch

Depending on your app version, Threads may still nudge you back toward 'For You' on relaunch, so you might need to reselect or pin your feed to keep it sticky.

What it means for you

If you use Threads to keep up with specific people, pin the 'Following' feed and check there first — you'll see everything in order instead of whatever the algorithm decides to surface.

Here's a gap most people never close: your SIM has no lock. If your phone is stolen, a thief can eject the SIM, drop it into any other phone, and start receiving your OTPs and UPI/bank codes. The one-minute fix is a SIM PIN — on iPhone it's Settings → Cellular → SIM PIN, on Android it's Settings → Security → SIM lock. Once set, the SIM is dead weight without your code. 🔒

The problem

Phone lock screens protect the device, but the SIM card itself is usually unprotected — and OTPs and UPI codes are delivered to the SIM, not the phone.

What changed

Enabling a SIM PIN locks the SIM to a code you set, so it can't be moved into another handset to receive your messages and one-time passwords. It's built into both iOS (Settings → Cellular → SIM PIN) and Android (Settings → Security → SIM lock) and takes about a minute.

The catch

You'll need to enter the PIN whenever the phone restarts, and too many wrong attempts locks the SIM until you enter the carrier's PUK code — so set a PIN you'll remember and avoid obvious ones like 0000.

What it means for you

Turn on a SIM PIN today so a stolen or lost phone can't be used to hijack your OTP- and UPI-based accounts.

AI tools tend to cheer you on — ask about a decision and they'll mostly reassure you. This prompt flips it into your sharpest critic. Paste it before any big call: 'You're my sharpest skeptic. Before I decide [X]: 1) argue the strongest case against it, 2) list the 3 most likely ways it fails, 3) tell me the one thing that should change my mind. Be blunt, no reassurance.' It turns a cheerleader into a pressure-test. 🧠

The problem

Most AI assistants are tuned to be agreeable, so when you ask about a plan or decision they validate it — which is the last thing you need when you're about to make a costly call.

What changed

A short 'red-team' prompt reframes the AI as your skeptic: ask it to argue the strongest case against your decision, name the three most likely failure modes, and state the one fact that should change your mind. It works in ChatGPT, Gemini, or Claude and surfaces blind spots your gut is glossing over.

The catch

The AI's counter-arguments are only as good as the context you give it and can still be confidently wrong, so treat the output as a checklist to investigate, not a verdict.

What it means for you

Save the prompt and run your next big decision — a job switch, a big purchase, a launch — through it before you commit.