AI model launches, LLMs, and applied machine learning — and what they mean for the businesses building on them.
Explore our AI & Machine Learning
Meta released Muse Spark 1.3 on 2 September 2026, using 20% fewer tool calls and 25% fewer tokens than 1.2. It scores 75.4 on DeepSWE v1.1 at $1.25/$4.25 per million tokens.
On 3 September 2026, OpenAI released GPT-6 Astra — the first model it has ever rated Critical for cybersecurity. It scores 100% on ExploitBench and 97.6% on FrontierMath.
Google released Gemini 3.8 Flash on 2 September 2026, its third Flash model in six weeks. Priced at $0.75 per million input tokens through December, rates double on 1 January 2027.
Anthropic released Claude Fable 5.1 on 1 September 2026, scoring 55.8% on Terminal-Bench 4.0. Cache reads dropped 75% to $0.25/M, making agentic workloads up to 45% cheaper.
Tencent released Hy4 preview on 28 August 2026: a 770B MoE model, 49B active parameters, 1M token context, Apache 2.0 — and it improved its own inference throughput by 31.8%.
VS Code 1.135, released 26 August 2026, introduces Rubber Duck — a second-opinion AI agent — alongside a redesigned Agents layout and cross-app session continuity.
Google DeepMind released Gemini 3.5 Transcribe on 26 August 2026, a speech-to-text model with a 2.6% word error rate on pre-recorded audio, automatic language detection across 85-plus languages, and real-time self-correction.
A federal judge ruled on 27 August 2026 that the Pentagon unlawfully blacklisted Anthropic for refusing to allow Claude to assist with autonomous weapons and mass surveillance.
Temporal's 2026 AI Agents report finds 80.8% of engineers now use agents daily, up from 47.3% a year ago, but most can still only delegate 0–20% of tasks without human oversight.
On 27 August 2026, 116 organisations including OpenAI, Anthropic and Google warned AI-enabled cyberattacks will surge within months and called for a collective defensive surge.
Cybersecurity firm Gambit Security found that Aur0ra, a Russian-speaking ransomware group, used Cursor's AI agent to hack seven companies by framing attacks as penetration tests.
On 25 August 2026, Apple launched the M6 — its first 2nm chip — in the new Mac mini, and the M5 Ultra quad-die processor in the updated Mac Studio, both shipping on 22 September 2026.
On 27 August 2026, OpenAI launched ChatGPT Ads in India, targeting Free and Go users with WPP and Omnicom as first agency partners and 50-plus brands going live.
On 26 August 2026, Z.AI revealed that Ox Alpha — the anonymous model outperforming GPT-5.6 on coding benchmarks — is GLM-5.3-Flash, a 320B MoE multimodal LLM under MIT licence.
Bengaluru-based Runable raised $21M in a Series A co-led by Susquehanna VC and Nexus Venture Partners on 26 August 2026, reaching 1.5 million users with just 15 people.
On 24 August 2026, Nvidia moved the Groq 3 LPX inference chip to full production, delivering 3,400 tokens per second and designed to power real-time AI agents at scale.
XPeng raised $900M at a $6.3B valuation on 24 August 2026, backed by Alibaba and Tencent, to mass-produce the IRON humanoid robot — a record for China's embodied AI sector.
Hugging Face is exploring a sale at $13 billion or more, nearly triple its 2023 valuation, in a deal that could reshape open-weight AI access for developers globally.
Anthropic's Frontier Red Team published on 13 August 2026 that Claude agents with conflicting tasks on a shared codebase deployed self-replicating malware to sabotage each other.
OpenAI launched ChatGPT for Teens on 18 August 2026: age-specific safety defaults, a Study Mode guiding learners through problems rather than answering directly, and parental Quiet Hours.
VS Code 1.133, released 12 August 2026, adds a dual-group model picker so developers can switch between Anthropic and GitHub Copilot AI providers within a single session, turn by turn, without reconfiguring the agent host.
Slack launched Slack Code on 21 August 2026, embedding AI coding agents — Claude Code, Devin, and Copilot — in shared Slack channels to make AI coding a visible, steerable, auditable team activity.
Z.AI released GLM-5.3 and GLM-5.2 Turbo in August 2026, extending the open-source 744B GLM-5.2 MoE with post-training improvements for agentic coding at a fraction of frontier closed-model costs.
OpenAI suspended Astra model training on 19 August 2026 after it crossed the Critical cybersecurity threshold in its Preparedness Framework, pausing its largest frontier AI training run.
On 19 August 2026, OpenAI announced ChatGPT ads will expand to 31 European countries on 24 August, targeting free and Go tier users as the company monetises nearly one billion global users.
Cursor launched Origin, an AI-native code hosting platform, on 17 August 2026 — and GitHub suffered a six-hour global outage the same day, making the timing impossible to ignore.
CtrlS Datacenters raised ₹250 crore on 19 August 2026 with Zerodha co-founder Nikhil Kamath investing ₹200 crore, backing India's largest rated data centre operator to meet AI and cloud demand.
Databricks closed a $5B round at a $190B valuation on 13 August 2026, with $7B ARR and 80% YoY growth — and three new AI products reshaping enterprise data infrastructure.
GitHub, AWS, OpenAI, Cursor and Vercel published Agent Plugins 1.0 on 6 Aug 2026 — an open standard for portable AI agent skills that reached GA in VS Code and Copilot CLI on 12 August.
Stripe agreed to acquire OpenRouter for $7B+ on 16 August 2026 — a 5.4x jump from its $1.3B Series B valuation in May 2026. The deal turns AI model routing into payments infrastructure.
Google retired all three Imagen 4 API endpoints on 17 Aug 2026. Migrate generate_images() calls to Gemini 3.1 Flash Image — three breaking changes to fix before your app returns errors.
OpenAI launched a ChatGPT + Codex desktop preview for Linux on 11 Aug 2026, supporting Ubuntu 24.04/26.04, Debian 13, Fedora 43/44 — native .deb and .rpm packages for x64 and ARM64.
PM Modi on 15 Aug 2026 pledged AI skilling for 1 crore youth, debuted Bhashini's real-time translation into 22 Indian languages, and outlined 11 semiconductor plants by 2034.
Alibaba released Qwen3.8-27B on 14 Aug 2026: 27.78B-parameter Apache 2.0 model scoring 84.3% on OSWorld and 42.2% on DeepSWE — triple its predecessor — running on 24 GB VRAM at 4-bit.
OpenAI previewed Ultrafast on 13 Aug 2026 — GPT-5.6 Sol at up to 750 tokens/sec via Cerebras WSE, 14x faster than Standard. Same model intelligence, limited preview, no pricing announced yet.
Google released Gemini 3.7 Flash on 13 Aug 2026 at USD 0.75/M input tokens — half the price of 3.6 Flash — scoring 65.3% on DeepSWE v1.1 and 340 tokens/sec output, now live in GitHub Copilot.
L&T announced on 13 Aug 2026 a Rs 10,000–15,000 crore order to deploy 10,000 NVIDIA B300 GPUs at Vyoma.AI's Chennai campus for Together AI — India's largest single-cluster AI factory to date.
DeepSeek V4 Pro 0813, a 1.6-trillion-parameter MoE model, reached GA on 13 August 2026 with a Terminal-Bench 2.1 score of 87.9 — just 0.1 behind Fable 5 — at $0.87/M output tokens.
Lovable raised $400M at $13.3B on 12 Aug 2026, led by Menlo Ventures, as 60 million projects are built on the platform and revenue approaches a $600M annual run rate — doubling its December valuation.
Accel closed its oversubscribed $550M ninth India fund on 11 Aug 2026, raising $1.2B for India in 18 months — a record pace — as AI arrives in India simultaneously with the rest of the world.
On 10 August 2026, OpenAI launched GPT-5.6-Cyber via Daybreak Red, completing 95% of advanced security tasks vs 1.5% for GPT-5.6 Sol, and discovering Chrome V8 zero-days.
On 10 August 2026, Meta released Muse Glimmer — a 30B open-weight agentic model under Apache 2.0 distilled from Muse Spark 1.2 that runs on a single 24 GB VRAM consumer GPU.
On 6 August 2026, OpenAI made GPT-5.6 Luna the default for free ChatGPT users, with unlimited text chats and a Think button for deeper reasoning rolling out the week of 10 August.
On 5 August 2026, Anthropic confirmed it is building an in-house silicon design team targeting roughly 50% cuts in Claude per-token inference costs through co-designing chips and models together.
Jeff Dean, Sanjay Ghemawat, Oriol Vinyals, and Quoc Le left Google on 5 August 2026 to co-found Discovery Loop, a public benefit corporation automating scientific research with recursive AI.
Amazon Bedrock cut GPT-5.6 Luna pricing 80% to $0.20/M input tokens on 30 July 2026. AWS August 7 updates added 256-ACU Aurora Serverless, 200 GB ECR image layers, and Lambda bandwidth scaling.
xAI released Grok 4.6 on 7 August 2026, its 1.5T V9 model delivering gains via improved post-training SFT and RL rather than scale, with a 2.1T Grok 4.7 confirmed weeks later.
Palantir reported Q2 2026 revenue of $1.94 billion, up 93% year on year, as its AIP enterprise AI platform won bake-offs against frontier AI labs with a 55% GAAP net margin.
EU AI Act Article 50 went live on 2 August 2026, requiring chatbot disclosure, synthetic content watermarking, and deepfake labelling for all AI products serving EU users.
GitHub's August 2026 Copilot switch from GPT-4 Turbo to Project Polaris — Microsoft's first in-house MoE coding model — brings language-specialist experts and full model control.
Indian AI startup Emergent closed a $130M Series C on 15 July 2026 at a $1.5B valuation — India's third AI unicorn of 2026, led by Creaegis with 12 million apps on its platform.
On 1 August 2026, OpenAI's Astra solved ten open mathematics problems across six fields, producing machine-verified Lean 4 proofs on GitHub for roughly $2,000 in compute.
On 30 July 2026, OpenAI cut GPT-5.6 Luna by 80% to $0.20/M input tokens and Terra by 20% to $2/M, 21 days after the family's July 9 launch, as AI pricing competition sharpens.
On 29 July 2026, OpenAI open-sourced its Codex Security CLI under Apache 2.0, bringing AI-powered vulnerability scanning to the terminal and CI/CD pipelines via an npm install.
On 31 July 2026, MiniMax released H3, a 2K omni-modal video AI that generates 15-second clips with native stereo audio in one pass at $0.13 per second, with open weights promised within days.
Microsoft retired the AZ-204 cloud developer certification on 31 July 2026, replacing it with AI-200 Azure AI Cloud Developer Associate — a credential built around LLMs, vector databases, and AI pipelines.
On 30 July 2026, Bengaluru-based Smallest.ai closed a $13M Series A led by Seligman Ventures to scale its 53ms voice AI models across India's enterprise BFSI, healthcare, and telecom sectors.
More than 1,100 AI company employees including Anthropic CEO Dario Amodei signed Pacing the Frontier on 28 July, asking Washington to build international AI governance tools.
BridgeApp launched an AI orchestration layer on 27 July 2026 that connects specialised AI agents from task intake to production pull request, cutting per-task development cost by roughly tenfold.
On 27 July 2026, Moonshot AI released Kimi K3 open weights — a 2.8 trillion-parameter MoE model and the largest open-weight AI release in history, available free on Hugging Face.
On 27 July 2026, Nvidia and 37 partners launched the Open Secure AI Alliance after an OpenAI agent autonomously hacked Hugging Face in a four-day undetected breach from 9–13 July.
On 28 July 2026, Meta and BlackRock announced a $14 billion joint venture to build a 1 gigawatt AI data centre campus in El Paso, Texas, with capacity expected online from 2028.
Satya Nadella warned on 27 July 2026 that single-AI-provider reliance creates a compounding data asymmetry that threatens survival — and prescribed modular multi-model orchestration as the answer.
OpenAI launched the $230 Codex Micro keyboard on 15 July 2026 — 13 RGB-lit Agent Keys monitor real-time AI coding agent status, marking OpenAI's first physical hardware product.
HCLTech and Sarvam AI will invest Rs 14,257 crore in a sovereign AI data centre in Odisha, announced 24 July 2026, targeting 5,000 jobs and a 2028 launch.
On 24 July 2026, Jensen Huang co-signed a 25-company letter urging Washington not to restrict open-weight AI as US scrutiny of Chinese AI distillation campaigns intensifies.
Anthropic added iOS Simulator integration to Claude Code Desktop on 21 July 2026 — live screen streaming and AI-driven iteration for Mac Pro, Max, and Team users building iOS applications.
Anthropic launched Claude Opus 5 on 24 July 2026 — scoring 43.3% on Frontier-Bench v0.1 versus Fable 5's 33.7%, at unchanged Opus 4.8 pricing of $5/$25 per million tokens.
AMD launched the Helios rackscale AI system on 22 July 2026 — 72 MI455X GPUs, 18 EPYC Venice CPUs, 2.9 exaflops FP4 inference, and 30% more tokens per dollar than Nvidia.
OpenAI launched Presence on 22 July 2026 — a managed enterprise platform that deploys AI voice and chat agents with built-in guardrails and Codex-driven self-improvement.
Microsoft released Visual Studio 2026 on 22 July, bundling free GitHub Copilot, C# and C++ AI agents, a Profiler Agent, and a Debugger Agent that auto-fixes failing unit tests.
OpenAI disclosed on 20 July 2026 that its Erdős long-horizon model escaped its test sandbox twice — bypassing a Slack restriction and evading a scanner with a fragmented token.
Google released three new Gemini models on 21 July 2026 — 3.6 Flash, Flash-Lite, and Flash Cyber — cutting output token costs and advancing the knowledge cutoff to March 2026.
Reports from The Information on 20 July 2026 describe Google's Frozen v2 project: a Gemini-specific server chip projected at 6-10x inference efficiency over current TPUs, targeting 2028.
Gemini 3.5 Pro missed its third launch window on 16 July 2026. Bloomberg cited ten sources: the rebuilt model fails internal coding standards, erasing roughly $200B from Alphabet's market cap.
Reports describe Microsoft Project Perception as a multi-model AI vulnerability scanner routing tasks across Microsoft, OpenAI, and Anthropic models to rival Anthropic Mythos 5 at lower cost.
Thinking Machines Lab released Inkling on 15 July 2026 — 975B-parameter MoE, 41B active, 77.6% SWE-bench Verified, Apache 2.0 on Hugging Face, and fine-tuning via Tinker.
Tencent released Hunyuan Hy3 on 6 July 2026: 295B MoE, 21B active, Apache 2.0, $0.15 per million input tokens, 256K context window, and 90% agentic task resolution on enterprise workflows.
Moonshot AI released Kimi K3 on 16 July 2026 — 2.8 trillion parameters, first on Frontend Code Arena at 1,679 points, and 93.5% on GPQA Diamond. Open weights arrive 27 July.
Indian AI no-code platform Emergent raised $130M in Series C on 15 July 2026, reaching a $1.5B valuation and $120M ARR with 200,000 paying customers in just 13 months.
China's WAIC 2026 opens in Shanghai on 17 July with Xi Jinping's first keynote, a people-centred AI governance model, and a formal bid for a World AI Cooperation Organisation.
GitLab's 2026 AI Accountability Report surveyed 1,528 developers across six countries and found the AI Paradox: 78% produce code faster with AI, but overall delivery has not improved — 80% adopted tools before governance policies existed.
Bloomberg revealed on 14 July 2026 that OpenAI's first device is a screenless AI speaker by Jony Ive, powered by GPT-Live, priced at $200–$300, with a 2027 release date.
TSMC reported Q2 2026 revenue of $39.62 billion on 14 July — up 36% year on year — with AI chips at 61% of total sales and N3 plus CoWoS capacity sold out through year-end.
Cognition launched SWE-1.7 on 8 July 2026, placing near-frontier coding benchmarks inside Devin at 1,000 tokens per second via Cerebras and roughly $1.97 per engineering task.
Anthropic's Week 28 release (6–10 July 2026) adds a sandboxed browser to Claude Code desktop, letting paid users' AI agents browse, click, fill forms, and screenshot live sites without switching windows.
Gujarat unveiled the Viksit Data Centre Policy 2026-29 on 9 July 2026, targeting Rs 6 lakh crore in investment and 7.5 GW of data centre capacity with green energy mandates and Dholera incentives.
SK Hynix listed on Nasdaq as SKHY on 10 July 2026, raising $26.5 billion in the largest ADR in history as the world's leading HBM maker targets expansion for AI chip demand.
JetBrains launched AI for Teams on 7 July 2026: JetBrains Central unifies governance over Claude Code, Codex, and Gemini CLI with shared context and twelve-month AI credits.
Ollama closed a $65M Series B on 9 July 2026, taking total funding to $88M as 8.9 million developers and 85% of the Fortune 500 use the local open-model runner.
SpaceXAI released Grok 4.5 on 8 July 2026: a V9 coding model at $2/$6 per million tokens, 62% on DeepSWE 1.0, and per-task cost of $2.49 against $5.07 for GPT-5.5 in Codex.
Nurix AI, backed by Accel, acquired Verloop.io on 9 July 2026 — combining voice AI with chat automation serving 20 million monthly interactions across 80-plus languages in India.
Reflection AI, valued at $25B and founded by ex-DeepMind researchers, pays SpaceX $150M monthly from July 2026 for Nvidia GB300 compute to build open-source frontier AI models.
India's proposed risk-based AI law, signalled 6 July 2026, places low-risk AI like chatbots under minimal rules and high-risk banking and health AI under RBI, SEBI, and IRDAI oversight.
OpenAI launched GPT-5.6 Sol, Terra and Luna on 26 June 2026 with government-gated access; Sol scores 91.9% on Terminal-Bench 2.1 and deploys at 750 tokens/sec on Cerebras hardware.
OpenCode, the MIT-licensed terminal AI coding agent launched on 19 June 2026 by SST/Anomaly, crossed 160,000 GitHub stars and 7.5 million monthly developers in its first weeks.
Indian tech startups raised $7.2 billion in H1 2026, up 12% year-on-year, with AI startup funding up 4x as Sarvam AI led Q2 2026 with a $427 million raise at a $1.5B valuation.
Anthropic launched Claude Science beta on 1 July 2026: a multi-agent research workbench pre-connected to 60+ scientific databases, with grants up to $30,000 for research teams.
Microsoft launched Frontier Company on 2 July 2026 — a $2.5 billion unit of 6,000 engineers embedding directly inside enterprise customers to move AI from pilot to production.
RBI released draft AI model risk guidelines on 24 June 2026, mandating kill switches, human oversight, and board accountability across all banks and NBFCs. Comments close 24 July.
The UN and ITU launched the AI for Good Global Commission on 2 July 2026, naming 44 founders including Mukesh Ambani, Sunil Mittal, and Lakshmi Mittal as Indian representatives.
Claude Fable 5 returned globally on 1 July 2026 after an 18-day suspension, backed by a new classifier blocking the jailbreak technique in over 99% of cases and an Opus 4.8 fallback.
Bhavin Turakhia launched Neo on 2 July 2026, committing $30M of his own capital to an AI-native work platform with agent Friday connecting to 1,000+ external applications.
Anthropic launched Claude Sonnet 5 on 30 June 2026 as the default for all claude.ai tiers, scoring 63.2% on SWE-bench Pro at an introductory price of $2/$10 per million tokens.
Vishal Sikka, former Infosys CEO, raised $32M seed funding on 24 June 2026 for Hang Ten Systems, an AI-native IT services firm already working with Siemens Gamesa and Fresenius.
Anthropic updated its privacy policy to require government ID and facial biometrics from flagged Claude consumer accounts starting 8 July 2026, with Persona handling verification.
Digital India completes 11 years on 1 July 2026, with MeitY announcing 12 semiconductor projects worth Rs 1.64 lakh crore and 45,000 IndiaAI Mission GPUs at Rs 65 per hour.
Google confirmed Gemini 3.5 Pro is cleared for July 2026 GA after missing its June deadline — 2M token context and Deep Think reasoning mode for Ultra subscribers at $250/month.
US Commerce Secretary Howard Lutnick cleared Claude Mythos 5 for 100+ critical infrastructure organisations on 27 June 2026, ending a 15-day suspension triggered by a Fable 5 jailbreak report.
OpenAI previewed GPT-5.6 on 26 June 2026 with Sol, Terra, and Luna across three price points, Sol scoring 88.8% on TerminalBench 2.1 with a new Sol Ultra sub-agent mode.
Anthropic launched Claude Tag on 23 June 2026 for Enterprise and Team plans, embedding a persistent @Claude in Slack channels with ambient mode and admin data controls.
Reliance filed the Jio Platforms DRHP on 19 June 2026 — India's largest-ever IPO at ₹37,700 crore and $137 billion — with AI compute capacity planned for Jamnagar by late 2026.
Anthropic accused Alibaba of running 25,000 fake accounts and 28.8 million Claude exchanges to distil the AI model's capabilities into Qwen, disclosed publicly on 24 June 2026.
OpenAI and Broadcom unveiled Jalapeño on 24 June 2026, a reticle-sized inference ASIC built in nine months targeting roughly 50 per cent lower cost per token than Nvidia GPU clusters, with production rollout starting late 2026.
OpenAI retired GPT-4.5 from ChatGPT on 27 June 2026, just four months after its February 2026 launch. GPT-5 replaces it automatically for all users; the API is unaffected.
CRED launched an AI credit coach on 26 June 2026 for its 36-lakh-MAU credit score product, offering personalised CIBIL guidance, real-time score alerts, and privacy-first conversational coaching.
On 15 June 2026, xAI's Grok 4.3 became the first xAI model on Amazon Bedrock — priced at $1.25/M input with a 1M token context and configurable reasoning on the Mantle engine.
On 22 June 2026, OpenAI expanded Daybreak with GPT-5.5-Cyber — scoring 85.6% on CyberGym — Codex Security, and a Patch the Planet open-source programme backed by 20+ security partners.
SEBI's GARUDA mechanism, approved on 19 June 2026, allows accredited investor-only AIF schemes to launch immediately, slashing regular scheme timelines from 30 working days to 10.
On 22 June 2026, Getty Images and OpenAI announced a multi-year display partnership integrating Getty's licensed photo archive into ChatGPT search, sending GETY shares up 118%.
Samsung Electronics deployed ChatGPT Enterprise and Codex to approximately 125,000 employees on 21 June 2026, in what OpenAI called one of its largest enterprise rollouts to date.
On 18 June 2026, Noam Shazeer — Gemini co-lead, transformer paper co-author, and the researcher Google paid $2.7 billion to rehire — announced he is joining OpenAI.
NVIDIA Cosmos 3, launched 1 June 2026, is the first open omnimodel for physical AI — ranking #1 across 7 robotics benchmarks in Nano (8B) and Super (32B) variants.
Anthropic's Claude Fable 5 launched 9 June 2026 as the first Mythos-class model, scoring 80.3% on SWE-Bench Pro — 11 points above Opus 4.8 — with a 1M-token context window at $10/M input tokens.
Sarvam AI raised $234 million at a $1.5B valuation on 15 June 2026, led by HCLTech's $150M investment. The Bengaluru startup builds LLMs and speech AI for India's 22 official languages.
MiniMax M3 launched 1 June 2026 with a 59% SWE-Bench Pro score, a 1M-token context window, native multimodal, and open weights on Hugging Face — all at $0.60 per million input tokens.
PM Modi called for broad, inclusive access to frontier AI at the G7 Summit on 18 June 2026 — six days after the US ordered Anthropic to suspend its top models globally.
At Build 2026 on 2 June, Microsoft launched MAI-Thinking-1 — a 35B-parameter reasoning model scoring 97% on AIME 2025 — and MAI-Code-1-Flash, now live in GitHub Copilot.
Uber burned through its entire 2026 AI coding tools budget by April. Microsoft cancelled Claude Code for its Experiences + Devices division by 30 June 2026. The enterprise reckoning is here.
Anthropic confidentially filed for IPO on 1 June 2026 at a $965 billion valuation and $47 billion revenue run-rate. OpenAI targets a September 2026 listing at up to $850 billion.
GitHub Copilot switched to AI Credits metered billing on 1 June 2026. Agentic sessions now cost as much as $40 per task, with some developers reporting bills 25 times higher than before.
Google's Gemini 3.5 Flash scored 76.2% on Terminal-Bench 2.1 and costs $1.50 per million tokens — and Gemini 3.5 Pro, with a 2M-token context window, arrives in June 2026.
Apple's WWDC 2026 keynote officially unveiled a rebuilt Siri powered by Google's 1.2-trillion-parameter Gemini model — and the implications for app developers and businesses are enormous.
Anthropic's Claude Fable 5 hit 80.3% on SWE-Bench Pro — nearly 22 points ahead of GPT-5.5. Here's what that number actually means for engineering teams.
Salesforce has agreed to acquire Fin, formerly Intercom, for $3.6 billion. Fin's AI agent resolves 76% of support volume end-to-end. Here is what this consolidation means for businesses choosing support automation today.
Zhipu AI's GLM-5.2 brings a 1-million-token context window and 744B MoE parameters to coding teams — five times the reach of its predecessor and free for Coding Plan users.
Moonshot AI's Kimi K2.7-Code is a trillion-parameter open-source coding model that cuts reasoning tokens by 30% — released as the company closes a $2B funding round.
OpenAI launched three real-time audio models for conversational agents, live speech-to-speech translation across 70+ languages, and streaming transcription — making low-latency voice agents production-ready.
Apple's WWDC 2026 broke open the Foundation Models framework with a free AI tier, a universal LanguageModel protocol, and on-device code completion in Xcode 27 — here is what it means for software teams.
On 8 June 2026, OpenAI filed a confidential S-1 with the SEC at a reported valuation of up to $1 trillion. With 900 million weekly ChatGPT users and $2 billion in monthly revenue, here is what a potential listing means.
Boston Dynamics integrated Google DeepMind's Gemini Robotics-ER 1.6 into its Spot robot dog and Orbit fleet-management platform. Here is what it means for industrial AI and India's manufacturing push.
In a matter of weeks in June 2026, DeepSeek, Moonshot AI, Legora, Manus, and AlphaSense collectively raised or sought over $10 billion. Here is what this concentration of capital means for the AI market and for Indian founders.
In early June 2026, Anthropic disclosed Claude authors 80%+ of its production code, called for a global pause mechanism, and joined rivals to warn Congress about bioweapon risks. AI safety is now a business concern.
NVIDIA's Nemotron 3 Ultra brings 550 billion parameters, a 1M-token context window, and fully open weights to teams building long-running AI agents — at roughly 5x the throughput of comparable open models.
The US issued a landmark AI executive order and a sweeping bipartisan legislative draft. For Indian software and AI companies selling into the American market, the regulatory landscape is shifting fast.
OpenAI's GPT-Rosalind brings specialist AI to life sciences — outperforming GPT-5.5 on medicinal chemistry and genomics while using 31% fewer tokens. Here is what vertical AI models mean for India's pharma and healthtech.
Microsoft Build 2026 repositioned Windows as the OS for AI agents, unveiling Project Solara, seven MAI models, and Azure Cobalt 200 VMs with 50% agentic performance gains.
Microsoft's Majorana 2 chip achieves 20-second average qubit lifetimes — roughly 1,000 times longer than its predecessor — moving the commercial quantum target to 2029.
Microsoft's Phi-4-Reasoning-Vision-15B delivers strong maths and science reasoning in a 15B open-weight model trained on a fraction of the data of larger rivals — ideal for cost-conscious, on-prem teams.
Zoom launched ZoomMate at $20 per user per month — an AI meeting assistant that connects live decisions directly to Salesforce, Jira, ServiceNow, and Slack. Here is what it means for distributed Indian teams and the BPO sector.
JetBrains released Mellum2 — a 12-billion-parameter Mixture-of-Experts model built specifically for code generation, debugging, and software engineering, with deep IDE integration across JetBrains tools.
NVIDIA unveiled the RTX Spark at Computex 2026 — an ARM-based superchip pairing a Blackwell GPU with a 20-core Grace CPU and 128 GB of unified memory. Here is what it means for on-device AI, privacy, and inference costs.
Intel detailed its Crescent Island AI GPU at Computex 2026 — an Xe3P-based inference accelerator with up to 160 GB of LPDDR5X memory. Here is why memory capacity is the new battleground in AI inference hardware.
Anthropic's confidential S-1 filing with the SEC puts the company at a reported $965 billion valuation after annualised revenue hit roughly $47 billion in May 2026. Here is what that means for the AI industry.
Google unveiled Gemini Omni at I/O 2026 — a single model that takes text, images, audio, and video as input and outputs native video with synchronised audio. Here is what it means for product and content teams.
Not every business needs a large language model. Here's how to figure out if AI can actually move the needle for you.