Monthly Tech News Roundup – April 2026

AI wars. Smartphone shifts. Indian startup energy. Your complete digest of the biggest stories this month.

Editor’s Note

Welcome to the April 2026 edition of the Tech Roundup. If you had to distil this month into a single theme, it would be: acceleration. The AI model race moved faster in April than in any previous month in history — benchmark leaderboards changed hands multiple times in a single week. India’s smartphone market hit a structural pivot, with nine out of ten buyers now saying AI features influence their purchase decision. And Indian startups continue to attract capital in a global funding environment that remains cautious everywhere else.

This edition is organized by category. Jump to the section most relevant to you, or read through for the full picture. A ‘What to Watch in May’ section at the end flags upcoming events and launches worth tracking.

AI & Machine Learning

Major Model Releases — April Was Historic

April 2026 is already being called the most consequential single month in the history of AI model development. The benchmark leaderboards changed hands multiple times in under 30 days.

• GPT-5.5 (OpenAI, April 23): OpenAI shipped its fully retrained GPT-5.5 — the first retrained base architecture since GPT-4.5. It claimed the top spot on the Artificial Analysis Intelligence Index at 60 points. Pricing: $2.50 per million input tokens, $15 per million output.
• Claude Opus 4.7 (Anthropic, April 16): Anthropic’s most capable general-availability model, designed for agentic workflows. Wins on 12 of 14 benchmarks compared to Opus 4.6, with particular gains in advanced software engineering and scaled tool use. GitHub Copilot’s coding agent runs on Claude Sonnet 4.6.
• DeepSeek V4 Preview (DeepSeek, April 24): The open-source bombshell. A 1.6 trillion parameter model available under MIT license, priced at just $0.14 per million input tokens for the Flash variant. Places within 7–8 benchmark points of GPT-5.5 at a fraction of the cost. Built on Huawei Ascend chips — a significant geopolitical signal.
• Grok 4.3 (xAI, April 24): xAI’s surprise full release, dropped hours after DeepSeek V4. Strong coding benchmarks and real-time X/Twitter data integration remain its differentiators.
• Meta Muse Spark (Meta, April 8): Meta’s first proprietary closed-weights model — a reversal from their open-source Llama strategy. Uses ‘thought compression’ during RL training. Scores 86.4% on figure-understanding vision tasks, outperforming GPT-5.4.
• Google Gemma 4 (April 2, open source): Apache 2.0 licensed, frontier-level performance, practical for local deployment. Widely considered the best open-source option for most use cases.

Key Stat In Q1 2026, LLM Stats logged 255 model releases from major organizations — roughly three significant AI model releases per day. The model-of-the-month era is over; we are now in the model-of-the-week era.

 

Claude Mythos — The Model That Wasn’t Launched

The most dramatic story of April was a model that did not launch publicly. After an internal data leak on March 26 confirmed Anthropic’s ‘Project Glasswing’ model (internally dubbed Mythos), Anthropic announced on April 7 that Claude Mythos will NOT receive a public release. The reason: the model’s capabilities in cybersecurity and code generation were deemed too dangerous for open deployment. It is being trialled with fewer than 50 vetted partner organizations under strict agreements. This marks the first time a major AI lab has publicly withheld a completed frontier model from the market on safety grounds.

AI Policy News

• India’s Digital India Act: The Ministry of Electronics and Information Technology (MeitY) released the updated draft framework for AI regulation in India this month. The framework proposes a tiered risk classification for AI systems — high-risk systems (used in credit, hiring, law enforcement) face mandatory audit requirements. Industry stakeholders have been given until June 15 to submit comments. This is India’s most concrete AI regulatory step to date.
• EU AI Act Enforcement Begins: The EU AI Act’s high-risk AI provisions came into enforcement for new deployments in April 2026. Companies operating in Europe — including several Indian IT service providers with EU clients — are now required to document AI system conformity. The compliance burden is significant for small and mid-sized vendors.
• US Executive Order on AI in National Security: The US government issued guidance limiting the use of AI models from Chinese origin (including DeepSeek) in federal systems. This mirrors existing restrictions on Huawei hardware and marks a significant escalation in the AI dimension of the tech cold war.

Interesting AI Tools Launched This Month

• Gemini 3.1 Flash Live: Google’s real-time multimodal AI is now powering Search Live in 200+ countries. You can speak a question, point your camera at an object, and get an instant voice + text response. Early reviews are calling it the most practical consumer AI tool of the year.
• Claude Code for VS Code and JetBrains: Anthropic expanded Claude Code availability to two of the most popular developer IDEs. Claude Code now competes directly with GitHub Copilot and Cursor for the developer workflow.
• Perplexity Assistant on Android: Perplexity’s AI assistant was pre-installed on select Android phones in India (Motorola, OnePlus) as an alternative default search assistant. A significant distribution win for the startup.
• Sarvam AI — Bulbul v3: Indian AI startup Sarvam AI launched Bulbul v3, a multilingual voice AI supporting 11 Indian languages with significantly improved naturalness scores. It is being deployed in government service applications across several states.

 

Notable Research Simplified

ARC-AGI-2: The AI field’s hardest benchmark had a major breakthrough month. Gemini 3.1 Pro scored 77.1% — more than double its predecessor’s 31.1%. For context, human performance on ARC-AGI-2 is around 85%. We are watching AI approach human-level performance on this benchmark in real time. The practical implication: AI systems are getting much better at reasoning about novel problems they have never seen before.

Smartphone & Gadget Launches

New Phone Launches in India — April 2026

• OnePlus Nord CE 6 & CE 6 Lite: Two new 5G budget devices from OnePlus, expected to launch at ₹20,000 and ₹15,000 respectively. The CE 6 features a Dimensity 8300 chipset with 12 GB RAM — strong mid-range specs. The CE 6 Lite targets first-time 5G buyers.
• Vivo Y500s: Vivo’s latest Y-series entry with a large battery (6,500 mAh), 50MP AI camera, and 5G connectivity under ₹18,000. Aimed at battery-conscious users in Tier-2 and Tier-3 Indian cities.
• Redmi Note 15 Pro+: Xiaomi’s Redmi Note 15 Pro+ launched with a 200MP periscope camera, Dimensity 9300, and 120W HyperCharge under ₹35,000 — a spec sheet that would have cost ₹60,000 three years ago.
• Samsung Galaxy A57: Samsung’s A-series continues to dominate the ₹25,000–₹35,000 segment. The A57 adds an AMOLED 144Hz display and Galaxy AI features (translation, call assist) at an accessible price.

The AI Smartphone Shift – Flipkart x Counterpoint Report

The bigger story is the shift from specification-led to experience-led buying. Indian consumers are no longer chasing RAM and chipset numbers — they are asking: does this phone help me with my work, content, and communication in a seamless, consistent way? Performance, camera quality, and battery life have become table stakes. AI tools are the new differentiator.

Price still matters enormously: 60% of buyers cite value for money as the primary driver. But 43% are using EMI options to access higher-spec devices — a meaningful shift in how Indians finance technology purchases.

Notable Global Launches

• Apple: No new iPhone launch this month, as expected. However, Apple confirmed Gemini-powered Siri features are rolling out to iPhones in more markets, including India, as part of iOS 19.3.
• Samsung Galaxy Buds 4 Pro: Samsung’s new TWS flagship received strong early reviews for its IP57 water resistance, Samsung Seamless Codec for Galaxy devices, and improved spatial audio. India pricing to be confirmed.
• Google Pixel 9a: Google’s mid-range Pixel 9a was officially confirmed for India availability at ₹52,000. Features Tensor G4, 6.1″ OLED display, and 7 years of software updates.

Gadget Accessories Worth Noting

• Lenovo L24D-4C Monitor: A 24″ FHD USB-C monitor priced under ₹15,000, receiving strong reviews from home-office buyers. USB-C power delivery and built-in speakers at this price are genuinely rare.
• boAt Airdopes 413 ANC: boAt’s newest ANC flagship at ₹3,999, targeting the OnePlus Nord Buds 4 Pro market directly. Reviews are mixed — solid hardware, but software falls short of OnePlus.
• Cleartrip + IRCTC: Not a gadget, but tech-relevant: Cleartrip (a Flipkart company) launched train ticket booking via IRCTC integration, adding a meaningful travel use case to the Flipkart ecosystem.

 

Big Tech & Startups

Indian Startup Ecosystem Highlights

• Sarvam AI funding: Sarvam AI, the Bengaluru-based multilingual AI startup, reportedly closed a significant Series B round this month. The company’s Bulbul voice AI and Sarvam-2B language models are gaining traction in government and enterprise deployments.
• Archimedis Digital expansion: The Tiruchirappalli-based life sciences IT company is doubling its workforce of 250–300 employees and expanding its AI Centre of Excellence — an indicator of AI specialization opportunities in Tier-2 Indian cities.
• Paisabazaar going phygital: India’s leading financial marketplace is opening physical stores in South India (Chennai, Bengaluru) — a bet that high-value financial product decisions still benefit from in-person interaction, even in a digital-first market.
• NatWest + IIT Delhi: Global bank NatWest Group partnered with IIT Delhi’s FITT to drive industry-academia collaboration in AI, fintech, and cybersecurity. The first international bank to partner with FITT, signalling growing global appetite for Indian AI research talent.

Big Tech Company News

• xAI acquires SpaceX (structurally): A landmark restructuring merged xAI with SpaceX in an entity valued at over $1 trillion combined. xAI employees received SpaceX stock conversion — widely reported as a major morale event. Strategic implications for AI compute, satellite infrastructure, and Grok’s future are still unfolding.
• Meta’s pivot to proprietary AI: Meta’s release of Muse Spark as a closed-weights model marks a significant strategic reversal from their years-long open-source Llama strategy. Meta Superintelligence Labs — built around Alexandr Wang’s team from Scale AI — is now producing Meta’s flagship closed models.
• TCS x Siemens Energy: TCS signed agreements with Siemens Energy India to support AI-ready data centre energy infrastructure across India. As AI compute demand grows, data centre energy management is becoming a major enterprise opportunity in India.
• BITS Pilani Smart Manufacturing Centre: BITS Pilani’s Work Integrated Learning Programmes opened its second Smart Manufacturing Competency Centre in Bengaluru, focused on Industrial AI, IIoT, Digital Twins, and Advanced Robotics. Targeted at helping Indian MSMEs adopt Industry 4.0 technologies.

 

Sonata Software Leadership Change

Sonata Software appointed Rajsekhar Datta Roy as CEO, effective May 9, 2026. Roy brings over 30 years of experience in scaling global technology practices. The transition comes as Sonata navigates its international expansion strategy amid growing competition in the mid-market IT services segment.

Cybersecurity & Privacy

Notable Vulnerabilities & Security Events

• The Anthropic Mythos Leak: The most significant security story in AI this month. Internal Anthropic documents were leaked on March 26, revealing the existence and capabilities of Claude Mythos. The incident — still under investigation — has prompted every major AI lab to review their internal data handling and access controls. The lesson: frontier AI model details are high-value intelligence targets.
• DeepSeek Security Scrutiny: Following the US Executive Order restricting DeepSeek in federal systems, cybersecurity researchers published findings about data handling practices in DeepSeek’s consumer apps. Users should be cautious about sharing sensitive information with any AI app from any jurisdiction without understanding the privacy policy.
• Indian Telecom Data Concerns: TRAI (Telecom Regulatory Authority of India) is investigating reports of unauthorized data sharing by telecom operators with third-party advertising networks. Action is expected in Q2 2026.

Privacy Policy Updates from Major Platforms

• Meta WhatsApp: WhatsApp updated its privacy policy in April to clarify AI data usage — specifically, that messages processed by Meta AI features (like AI chat assistants in groups) may be used to improve Meta’s AI systems. End-to-end encryption of standard messages is unaffected.
• LinkedIn: LinkedIn updated its data policy to explicitly allow user data to be used for training LinkedIn’s own AI models, with an opt-out option buried in Settings. Worth checking your account settings if you have concerns.
• Google Maps Timeline: Google completed the rollout of on-device Timeline storage (rather than cloud storage) for Maps location history. This is a meaningful privacy improvement — your location data no longer lives on Google’s servers by default.

Security Tips Based on This Month

• Review your AI app permissions: Check which apps on your phone have microphone and camera access. The rapid growth of AI assistants means more apps than ever are requesting these permissions.
• Update your LinkedIn privacy settings: Go to Settings > Data Privacy > Data for Generative AI Improvement and toggle off if you prefer not to contribute your data.
• If you use DeepSeek or similar AI services: Avoid pasting sensitive personal, professional, or financial information into any AI chatbot. Treat AI chat inputs as potentially visible to the service provider.
• Enable 2FA everywhere: With the pace of AI-powered phishing attacks increasing, SMS-based 2FA is no longer sufficient for important accounts. Use an authenticator app (Google Authenticator, Authy) for critical services.

 

Quick Hits

1. OneAssist HawkEye Launch

Insurance-tech startup OneAssist launched HawkEye — an AI-powered protection plan for existing (not new) smartphones. Addresses a real gap: of 15 crore smartphones sold annually in India, fewer than 1.5 crore are covered by any protection plan.

3. Xiaomi JeevikaSaathi+ Women Initiative

Xiaomi India launched JeevikaSaathi+, a women entrepreneurship programme supporting female micro-entrepreneurs in Tier-2 and Tier-3 Indian cities. Combines digital literacy training with device access.

4. Google’s AI Cost-Cutting Algorithm

Google quietly announced a compression algorithm that reduces KV-cache memory requirements by 6x. This is not a model release — it is an infrastructure improvement. But its implications are enormous: existing models become much cheaper to run at scale. The economics of AI deployment are about to get much more accessible.

5. GLM-5.1 Open Source Milestone

Chinese AI lab Zhipu AI’s GLM-5.1 briefly held the #1 spot on SWE-bench Pro (code generation benchmark) — the first time an open-weight model ever topped that benchmark. It was displaced by Claude Opus 4.7 nine days later, but the milestone signals that open-source AI is closing the gap faster than anyone predicted.

6. Grokipedia

xAI launched Grokipedia — a Wikipedia-like knowledge base powered by Grok, pulling from real-time X/Twitter data and structured knowledge. Still early, but the combination of live data and structured encyclopaedic format is genuinely novel.

 

What to Watch in May 2026

Upcoming Launches & Events

• Google I/O 2026 (expected mid-May): Google’s annual developer conference. Watch for Android 16 announcements, Gemini API updates, and Pixel hardware previews. Google has been on a roll in 2026 — this should be a major event.
• OnePlus Nord CE 6 & CE 6 Lite India launch: Confirmed for May. Likely to make a significant impact in the ₹15,000–₹22,000 5G segment, currently dominated by Realme and Xiaomi.
• Anthropic’s next move: With Claude Mythos withheld, the community expects either a new Claude public model announcement or significant Claude Code / agentic product launches in May.
• Samsung Galaxy Buds 4 Pro India pricing: Samsung confirmed the Buds 4 Pro for India but has not announced INR pricing yet. Expected in the ₹18,000–₹22,000 range.
• EU AI Act compliance deadline for specific high-risk categories: Several more AI Act provisions come into force in May 2026. Indian IT companies with EU contracts should be monitoring this closely.
• WWDC 2026 (Apple, likely early June): Though it is a June event, Apple’s developer announcements typically start leaking in May. Expect rumours about iOS 20, Apple Intelligence updates, and possible new Macs with next-gen chips.

Editor’s Pick for May Watch Google I/O 2026 closely. Google has led benchmark rankings for most of early 2026, and their ability to combine that model quality with Android’s 3 billion active devices gives them a distribution moat that no other AI lab can match. What they announce at I/O will shape the second half of 2026.