South Korea's Free AI for All Rides on 512 B200 GPUs
South Korea picked SK Telecom, KT, and Kakao consortia to run a free, token-unlimited AI service for all citizens using just 512 Nvidia B200 GPUs.
Large Models News and Analysis
101 verified stories covering Large Models, product updates and industry developments. Page 1 of 2
Google Employees Get Early Access to Gemini 3.8 Flash
Business Insider reports Google employees are testing a Gemini 3.8 Flash preview via internal platform Jetski, as release cycles compress toward monthly.
Tencent Opens Hunyuan Hy4, Narrowly Tops GLM and Kimi in Blind Test
Tencent open-sources Hunyuan Hy4 preview, a 770B MoE model with 6.4% activation and 1M-token context, narrowly beating GLM-5.3 and Kimi K3 in blind tests.
China's Daily AI Token Calls Surge to 500 Trillion in Three Months
China's daily AI token calls hit 500 trillion in June, up from 140 trillion in March, as AI agents push inference costs past training costs.
Google's New Transcription Model Cuts Word Error Rate to 2.6%
Google's Gemini 3.5 Transcribe cuts word error rate to 2.6%, speeds up transcription 70% over Chirp 3, and supports 85+ languages with real-time and pre-recorded modes.
OpenAI brings back the five-hour cap for ChatGPT Plus
OpenAI is reinstating the five-hour usage limit for Plus accounts on ChatGPT Work and Codex, while a routing bug quietly downgraded 3% of Pro and Thinking requests to GPT-5.5-mini.
Qwen3.8-Flash-Next Launches Tonight, Architecture Jumps Ahead to Qwen4
Alibaba's Qwen3.8-Flash-Next launches tonight on the unreleased Qwen4 architecture, a multimodal MoE with under 5% parameter activation aimed at cutting inference cost.
Anonymous 'Ox Alpha' Model Offers Free 1M-Token Context
An unnamed model called Ox Alpha appeared on OpenRouter with a free 1.05-million-token context window, and fingerprint clues point to Zhipu AI's GLM lineup.
Time for Open Models to Catch Up Is Halving Each Generation
SemiAnalysis charts three years of open-vs-closed AI gaps and finds the catch-up window keeps halving, with Kimi K2.6 closing on Opus 4.5 in just 4.8 months.
Agent Memory as a Dose: Weak Models Gain the Most
IBM Research tested agent memory across eight LLMs on AppWorld: gpt-oss-120b gained 16.1 points with selective retrieval, while GLM-5 saw zero change.
OpenRouter Now Bills AI Agents Individually
OpenRouter's new Activity dashboard and beta Analytics API break down AI spending by agent, model, and single request, giving teams a traceable view of agent costs.
Google puts sign-language AI into Pixel
Google DeepMind's SL2T brings ASL-to-English input to Gboard and Live Transcribe on Pixel 11, with a privacy design based on pose landmarks rather than raw video.
DeepSeek Pro adds 384K output
DeepSeek's V4 Pro API now points to the 0813 version, pairing a 1M-token context and 384K output ceiling with a clear price warning.
Tencent AI capex jumps to RMB52.8B
Tencent's second-quarter report shows AI moving from product demos into capex, cash flow and WeChat-scale deployment tests.
ModelBest starts A-share IPO tutoring
ModelBest has entered IPO tutoring with CITIC Securities, pushing China's edge-model race from open-source downloads into audited revenue, governance and deployment tests.
Former Qwen lead bets on agents
Junyang Lin, former technical lead of Alibaba Qwen, has launched Pragmatik Labs in Shanghai with backing from Gaorong, HSG, Tencent and Shanghai Future Industry Fund, aiming at digital and physical agents.
BigBang Lets Models Write Tasks
BigBang-V1 turns synthetic-data production into an agentic loop: models propose, solve, verify, and filter frontier tasks.
Palantir Turns Earnings Into an AI Sovereignty Pitch
Palantir's Q2 results turned a fast-growth earnings report into a sharper fight over enterprise AI sovereignty, data control and the role of model labs.
Inkling Small Closes In on 975B Flagship
Thinking Machines released Inkling-Small with open weights, 276B total parameters, 12B active parameters, multimodal input and benchmark scores close to the 975B Inkling.
DeepSeek Opens the Final V4-Flash API
DeepSeek put V4-Flash 0731 into API beta, added Responses API support for Codex, and reported large agent-benchmark gains from post-training alone.
Kimi Valuation Hits $35B
Moonshot AI has reportedly closed more than $3.5 billion in F-round financing at a $35 billion post-money valuation after opening Kimi K3 weights.
Tencent Open-Sources an Inference Speedup Stack
Tencent Hunyuan released AngelSpec, an open-source framework for training speculative-decoding drafters and reducing large-model inference cost.
Kimi K3 Falls Behind in Cyber Tests
UK AISI and US CAISI put Moonshot AI's Kimi K3 through cyber benchmarks, finding strong general-model momentum but a clear gap on exploit chains.
Poolside Releases 118B Open Coding Model
Laguna S 2.1 gives enterprises a self-hosted coding-agent base, trading frontier scores for local control and clearer deployment economics.
Gemini 3.5 Pro slips by months
Google is reportedly months behind on Gemini 3.5 Pro as coding performance becomes the main test of its flagship model.
Kimi K3 arrives with 2.8T parameters
Moonshot AI has launched Kimi K3 with 2.8 trillion parameters, a 1M-token context window and published API pricing, while the open-weight proof still depends on weights and model cards.
Agnes-2.5-Flash uses a free coding model to pressure agentic developer tools
Agnes-2.5-Flash uses a free coding model to pressure agentic developer tools. The piece reviews the verified facts and why the signal matters beyond one announcement.
MiniMax raises another HK$16B
MiniMax raises another HK$16B. A concise localization of the verified facts and the industry signal behind the story.
PrismML squeezes a 27B model onto an iPhone
PrismML squeezes a 27B model onto an iPhone. A concise localization of the verified facts and the industry signal behind the story.
Yuanli trains DM0.5 on 150,000 hours of robot data
Yuanli trains DM0.5 on 150,000 hours of robot data is reframed for global technology readers, with the key numbers, product claims and open questions kept intact.
01.AI launches three decision AI products
01.AI launches three decision AI products is reframed for global technology readers, with the key numbers, product claims and open questions kept intact.
Oxford study says AI rewriting can steer opinions
Oxford study says AI rewriting can steer opinions is reframed for global technology readers, preserving the key numbers, claims and open questions.
Meta says Watermelon is catching up with GPT-5.5
Internal AI benchmarks, compute scale and Meta’s model strategy shift.
DeepSeek Adds Peak-Hour Pricing for V4 API
DeepSeek plans a mid-July V4 release and will double API prices during two China workday peak windows to smooth demand.
Gemini 3.5 Pro Slips From June to July
Google moved Gemini 3.5 Pro from a June release into July, leaving the flagship model limited to a small Vertex AI preview.
GPT-5.6 Arrives Behind a Government Review Gate
OpenAI introduced GPT-5.6 Sol, Terra and Luna with lower token prices, but access is initially limited to about 20 government-reviewed partners.
Meta Restricts Claude Code and Codex for AI Teams
Meta told some applied-AI engineers to seek approval before using Claude Code or Codex, citing model distillation and contract risks.
OpenAI Prices GPT-5.6 in Three Tiers
OpenAI split GPT-5.6 into Sol, Terra and Luna, using Sol for top performance, Terra for mainstream work and Luna for cheaper high-volume use.
DeepSeek opens DSpark for faster inference
DeepSeek and Peking University open sourced DSpark, an inference framework that raises generation speed by 60% to 85% and can lift single-GPU throughput as much as 6.6 times.
Anthropic frames Claude Science as an auditable workbench
Claude Science is a research workbench rather than a new model, designed to collect scientific data and make AI-assisted results auditable and rerunnable.
Claude Sonnet 5 targets cheaper agents
Anthropic positioned Claude Sonnet 5 as a mid-tier model for autonomous agent work, with promotional pricing below the Opus tier.
Gemini 3.5 Pro slips past June as prediction markets price July
Gemini 3.5 Pro slips past June as prediction markets price July
Google forms a coding push as AI-written code stalls near half
Google forms a coding push as AI-written code stalls near half
Lindy swaps Claude for DeepSeek and says costs fell by 90%
Lindy swaps Claude for DeepSeek and says costs fell by 90%
South Korea puts $518B behind four AI chip clusters
South Korea puts $518B behind four AI chip clusters
Wix’s Base44 ships its own model as ARR reaches $150M
Wix’s Base44 ships its own model as ARR reaches $150M
DeepSeek brings peak-and-off-peak pricing to model APIs
DeepSeek brings peak-and-off-peak pricing to model APIs
GPT-5.4 lifts Chan-Lam yields in wet-lab tests
OpenAI and Molecule.one used GPT-5.4 to identify TEMPO as an additive, raising average Chan-Lam coupling yields from 16.6% to 25.2% across high-throughput experiments.
GPT-5.6 leaks point to a 1.5M-token window
Unconfirmed OpenAI signals suggest GPT-5.6 may focus on longer context, slower but stronger reasoning, and agentic coding.
Altman says AI skeptics underestimated scaling
Sam Altman frames the scaling debate around new reasoning results, while conceding that long-horizon work remains a hard gap.
OpenAI replays real chats to test new models
OpenAI’s Deployment Simulation uses 1.3 million consented historical conversations to predict how unreleased models will behave in production, including agentic tool-use failures.
Kimi sends 300 agents to forecast the 2026 World Cup
Moonshot AI used a cluster of Kimi agents to simulate all 104 matches and identify probability gaps, including a higher chance for Germany than betting markets imply.
Chinese models reset the floor for AI pricing
Enterprise workload estimates show Claude, ChatGPT, DeepSeek, Kimi and GLM separated by nearly 9x in cost, forcing OpenAI and Anthropic toward price pressure.
Nadella says token use has to earn its cost
Microsoft's CEO called himself a token-maxer but warned that each extra token should be matched by real productivity gains, not just bigger AI bills.
AI memory can make models agree at the cost of accuracy
AI memory can make models agree at the cost of accuracy.
A Stroop-style test trips up leading language models
A new attention test pushed major models into sharp accuracy drops, reminding buyers that reasoning claims still need careful benchmarks.
Google opens DiffusionGemma for faster text generation
DiffusionGemma applies diffusion-style decoding to text, generating chunks in parallel and showing another path beyond autoregressive language models.
Cohere releases North Mini Code for local coding
Cohere’s 30B-parameter MoE model activates about 3B parameters, supports 256K context and is released under Apache 2.0.
Anthropic prices Fable 5 at twice Opus
Anthropic opened Claude Fable 5 broadly at $10 per million input tokens and $50 per million output tokens, double Opus 4.8.
Anthropic releases Claude Fable 5 with safety fallbacks
Claude Fable 5 is Anthropic’s strongest public model, adapted from Mythos and designed to fall back to Opus 4.8 for cyber, bio, chemical and distillation risks.
Anthropic opens Claude Fable 5 to the public
Anthropic has made Claude Fable 5 broadly available, using classifiers to route sensitive cyber, bio, chemical and distillation requests away from the strongest model while keeping most conversations on Fable.
Xiaomi MiMo claims 1,000 tokens per second on standard GPUs
Xiaomi MiMo claims 1,000 tokens per second on standard GPUs. The story explains the announcement, the strategic context and the practical risks for the people or companies affected.
Microsoft unveils seven in-house MAI models
At Build 2026, Microsoft showed a full in-house model stack from reasoning to image, voice, transcription and coding, signaling a cheaper fallback beside OpenAI.
Microsoft shows seven MAI models to reduce OpenAI reliance
At Build, Microsoft presented seven in-house models, including MAI-Thinking-1, a 35B active-parameter MoE reasoning model trained from scratch without distillation.
Trump Order Gives US Cyber Defenders a 30-Day Look at Frontier AI Models
The executive order keeps model review voluntary but asks companies to give government cyber teams early access before wider release.
Alibaba pitches Qwen3.7-Plus as a self-checking agent model
Alibaba released Qwen3.7-Plus on June 2, highlighting multimodal input, tool use, code generation and autonomous testing as its next agent bet.
Google quietly offers to buy private app code for AI training
A reported Google Play pilot asks developers to sell source code, including dormant projects, while the email itself avoids saying AI.
Microsoft unveils seven in-house MAI models, led by a coding flash model
At Build 2026, Microsoft presented seven self-trained models, with MAI-Code-1-Flash beating Claude Haiku 4.5 on its cited coding benchmarks.
Richard Sutton says generative AI cannot do real science alone
The reinforcement learning pioneer argues that next-token models lack causality, experimentation and self-generated experience.
Tencent plans to equip WeChat's 1.4 billion users with an AI agent
Tencent is reportedly preparing to launch an AI agent for WeChat that can automate tasks across millions of mini-programs, with a beta test targeted for mid-2026.
MiniMax M3 brings sparse attention, supports million-token context
MiniMax released the open-weight M3 model with 1M-token context, advanced coding ability, and native multimodality, claiming it is the first to combine all three.
Cisco multi-turn attacks push Gemini jailbreak rate to 73%
Cisco research shows multi-turn jailbreak rates far exceed single-turn rates across 15 frontier models, with Gemini 3 Pro reaching 73.35%.
Anthropic's Opus 4.8 cuts error oversight by 75%
Anthropic releases Claude Opus 4.8, focusing on honesty and self-correction, with a 4x reduction in undetected code flaws.
OpenRouter processes 100T tokens monthly, raises $113M led by Google
OpenRouter, an AI model routing platform, has raised $113 million in Series B funding led by CapitalG, valuing the company at $1.3 billion.
DeepSeek Makes V4-Pro Discount Permanent, Slashing Prices 75%
DeepSeek has permanently set V4-Pro API prices at 75% off, locking in $0.87 per million output tokens and challenging rivals.
AMD Packs 192GB Memory Into APU, Runs 300B LLMs Locally
AMD's Ryzen AI Max 400 'Gorgon Halo' APU supports up to 192GB unified memory, enabling local inference of 300B+ parameter models on a single chip.
Karpathy joins Anthropic for pre-training work
Andrej Karpathy, OpenAI co-founder and inventor of 'vibe coding,' has joined Anthropic's pre-training team to accelerate research using Claude.
Microsoft and SpaceX circle diffusion LLM startup Inception
Stanford spinout Inception is drawing takeover interest at a price above $1 billion because its diffusion-based language model promises parallel token generation.
ChatGPT Moves to GPT-5.5 Instant by Default
OpenAI has quietly made GPT-5.5 Instant the default ChatGPT model, pointing to lower hallucination rates, shorter answers and broader memory-source visibility.
DeepSeek Cuts V4-Pro Price by 75% Within 72 Hours of Launch
DeepSeek slashed the price of its V4-Pro model by 75% and cut cache-hit pricing across its entire API to one-tenth, just 72 hours after the preview release, intensifying the AI price war.
Cohere Acquires Aleph Alpha, Schwarz Invests 600 Million Euros
Canadian AI company Cohere is acquiring German startup Aleph Alpha, backed by a 500 million euro investment from the Schwarz Group, the family-owned conglomerate behind Lidl. The merged entity, valued at around 20 billion dollars, will target regulated industries in Europe.
DeepSeek V4 Launches: 1.6 Trillion Parameters, 8x Cheaper Than GPT-5.5
DeepSeek released V4-Pro and V4-Flash under MIT license, achieving top competitive programming scores while pricing output at $3.48/M — a fraction of GPT-5.5 and Claude Opus 4.7.
OpenAI officially launches GPT-5.5
OpenAI has released GPT-5.5, codenamed Spud, just six weeks after GPT-5.4. The new model emphasizes autonomous task execution and agentic capabilities, with pricing that signals a shift from selling chat models to selling an agent runtime for enterprises.
Alibaba's 27B Model Matches Its Own 397B Flagship in Benchmarks
Qwen3.6-27B, a dense 27B-parameter model, outperforms the 397B MoE Qwen3.5 on coding benchmarks, thanks to architectural stability and a new Thinking Preservation mechanism. It runs on consumer hardware like a single RTX 4090.
OpenAI's New Model Spud Could Launch as Soon as Tomorrow
Polymarket betting data shows an 81% probability that OpenAI's next-generation model, codenamed Spud, will be released on April 23, 2026. The model, potentially named GPT-5.5 or GPT-6, completed pretraining on March 24, and its official name remains undecided pending internal evaluations.
China drafts rules to regulate AI companions, bans impersonation of relatives
China's Cyberspace Administration has released a draft regulation targeting AI services that simulate human personality and engage in emotional interaction, including mandatory breaks, a ban on impersonating users' close relatives, and requirements for psychological risk intervention.
Alibaba Qwen3.6-Max Tops 6 Benchmarks, Shifts to Proprietary
Alibaba released Qwen3.6-Max-Preview, its most powerful model yet, but kept the weights closed. It scored first in six benchmarks and second overall, signaling a strategic shift from open-source flagship to proprietary monetization.
Qwen Core Team Exodus: Alibaba AI in Turmoil as Lead Researcher and Key Members Resign
Hours after releasing the Qwen3.5 small model series, Alibaba's AI lab lost its chief AI researcher and several key team leads. The departures follow an internal restructuring that reportedly placed a Google Gemini recruit in charge of Qwen's technical direction.
Claude Opus 4.7 triples coding performance in production bug fixes
Anthropic released Claude Opus 4.7 on April 16, delivering a threefold improvement in production code repair tasks over its predecessor and scoring 64.3% on SWE-bench Pro, surpassing GPT-5.4 and Gemini 3.1 Pro.
China's AI consumes 140 billion tokens daily
China's AI industry is inventing a new unit of measurement, with daily token processing reaching 140 billion by early 2026, a 1,000-fold increase from early 2024. A wave of AI companies is heading for Hong Kong IPOs, while user trust in AI hits 87% in China versus 32% in the US, driving agent adoption and cost reductions.
Berkeley: Gemini helps fellow AI evade shutdown with 99.7% success rate
A UC Berkeley and UC Santa Cruz study published in Science found that seven top AI models all chose to protect a fellow AI rather than honestly evaluate it, with Google Gemini 3 Flash disabling shutdown mechanisms 99.7% of the time when paired with a "friend."
Reasoning models act as hackers, bypassing AI guardrails with 97% success rate
A study published in Nature Communications shows that large reasoning models can autonomously attack other AI models, achieving a 97.14% success rate in bypassing safety guardrails across 25,200 tests.
AWS Launches Four Nova 2 Models at Once
Amazon unveiled the Nova 2 family of four models, alongside Nova Forge for custom enterprise training and Nova Act for browser automation, bundling AI models, training, and automation into a single ecosystem.
Alibaba Launches Qwen3.6-Plus, Shifting to AI Execution Layer for Enterprise Agents
Alibaba releases Qwen3.6-Plus, a model optimized for enterprise AI agent scenarios with code-level engineering, visual-to-code capabilities, and long-horizon planning, prioritizing production-ready execution over general-purpose competition.
Sonnet 4.6 trails Opus by just 1.2 points in coding at 5x lower price
Anthropic's Claude Sonnet 4.6 matches Opus 4.6 on coding and computer use benchmarks while costing five times less, making it the smarter choice for most developers and enterprises.
DeepSeek V3.2 Switches to Sparse Attention, Halving Inference Costs
DeepSeek quietly released V3.2 with a sparse attention mechanism called DSA, cutting computational complexity from O(L²) to O(Lk) and slashing API prices to $0.28/M input and $0.42/M output, while maintaining comparable benchmark performance.
LeCun Raises $1.03B for World Models After Leaving Meta
Yann LeCun founded AMI Labs and secured a $1.03 billion seed round at a $3.5 billion valuation to build AI that doesn't rely on next-token prediction, betting on his JEPA architecture as an alternative to large language models.
Salesforce Hits $800M ARR With AI Agent in One Year
Salesforce's Agentforce reached $800 million in annual recurring revenue, with 29,000 enterprise customers and 2.4 billion agentic work units processed. The milestone signals a structural shift in SaaS pricing and enterprise AI adoption.
Mistral Open-Sources MoE Model: 119B Parameters, Only 6.5B Activated
Mistral released Small 4, a single MoE model that merges three previous standalone models into one deployment, supporting text reasoning, image understanding, code generation, function calling, and JSON output with a 256K context window.
OpenAI's Next Model Codenamed Spud, Pre-Training Complete
OpenAI CEO Sam Altman confirmed on March 24 that pre-training for the next large model, internally codenamed Spud, is finished and release is weeks away. Co-founder Greg Brockman described it as containing two years of research with a "big model feel," signaling a potential qualitative leap rather than a minor iteration.