Foundation Models & Reasoning
Core model architectures, training methods, chain-of-thought reasoning, and test-time compute scaling. The backbone of modern AI capabilities.
Key Benchmarks
Recent Papers
NeuReasoner: Towards Explainable, Controllable, and Unified Reasoning via Mixture-of-Neurons
Haonan Dong, Kehan Jiang, Haoran Ye +3 more
R2-Write: Reflection and Revision for Open-Ended Writing with Deep Reasoning
Wanlong Liu, Bo Zhang, Chenliang Li +4 more
Verbalizing LLMs' assumptions to explain and control sycophancy
Myra Cheng, Isabel Sieh, Humishka Zope +7 more
InCoder-32B-Thinking: Industrial Code World Model for Thinking
Jian Yang, Wei Zhang, Jiajun Wu +19 more
CoME-VL: Scaling Complementary Multi-Encoder Vision-Language Learning
Ankan Deria, Komal Kumar, Xilin He +4 more
BidirLM: From Text to Omnimodal Bidirectional Encoders by Adapting and Composing Causal LLMs
Nicolas Boizard, Théo Deschamps-Berger, Hippolyte Gisserot-Boukhlef +2 more
Goose: Anisotropic Speculation Trees for Training-Free Speculative Decoding
Tao Jin, Phuong Minh Nguyen, Naoya Inoue
Reliable Control-Point Selection for Steering Reasoning in Large Language Models
Haomin Zhuang, Hojun Yoo, Xiaonan Luo +2 more
The Expert Strikes Back: Interpreting Mixture-of-Experts Language Models at Expert Level
Jeremy Herbst, Jae Hee Lee, Stefan Wermter
Neuro-RIT: Neuron-Guided Instruction Tuning for Robust Retrieval-Augmented Language Model
Jaemin Kim, Jae O Lee, Sumyeong Ahn +1 more
Recent Milestones
Gemma 4 vs Llama 4 vs Qwen 3.5: Open Titans
US-based consultancy Lushbinary published an in-depth comparison on April 5, 2026 of three flagship open-weight model families: Google DeepMind’s Gemma 4, Meta’s Llama 4 and Alibaba’s Qwen 3.5. The piece benchmarks licensing, performance, context length, multimodality and deployment trade-offs for production use.([lushbinary.com](https://www.lushbinary.com/blog/gemma-4-vs-llama-4-vs-qwen-3-5-open-weight-model-comparison/))
400B Open Reasoning Model Undercuts Claude
On April 3, 2026, Arcee AI released Trinity-Large-Thinking, an Apache 2.0–licensed 400B-parameter sparse Mixture-of-Experts reasoning model that activates 13B parameters per token. The model scores 91.9 on PinchBench, within two points of Anthropic’s Claude Opus 4.6, while Arcee prices output at $0.90 per million tokens, roughly 96% cheaper than Opus. Trinity-Large-Thinking is available via OpenRouter, DigitalOcean’s Agentic Inference Cloud and downloadable weights on Hugging Face.
China Labs Turn Token Sales Into Real Revenue
On April 3, 2026 at 14:49 local time in Shanghai, Xinhua reported that Chinese labs MiniMax, Zhipu AI and Moonshot AI are driving a global ‘token economy’ with rapid growth in API usage and overseas adoption. Zhipu AI’s 2025 revenue jumped 131.9% year‑on‑year with token sales up 292.6%, while MiniMax’s 2025 revenue rose 158.9% with about 70% from international markets, and Moonshot’s Kimi K2.5 model was recently adopted as the base engine for U.S. coding platform Cursor.
Microsoft Pours $10B into Japan AI Stack
Microsoft announced on April 3, 2026 it will invest $10 billion in Japan between 2026 and 2029 to expand AI data centers, strengthen cybersecurity and train one million engineers. The package includes partnerships with SoftBank and Sakura Internet to provide sovereign GPU infrastructure and in-country AI compute for Japanese customers.
AI Soaks Up 81% of Record $300B Q1 VC
Multiple analyses published on April 2, 2026 report that global venture funding hit roughly $297–300 billion in Q1 2026, the highest quarter on record. Around $239–242 billion, or about 81%, went to AI companies, led by mega-rounds for OpenAI, Anthropic, xAI and Waymo.
Record $297B Q1 as AI megadeals dominate
On April 1, 2026, TechCrunch reported that global startup funding hit $297 billion in Q1 2026, the largest quarter on record. The spike was driven by four outsized rounds, including massive financings for OpenAI, Anthropic, xAI and Waymo that together accounted for roughly two‑thirds of the total. Seed‑stage AI startups are also raising at historically rich valuations.
OpenAI’s Record $122B Round Supercharges AGI Race
OpenAI said on March 31, 2026 it closed a $122 billion funding round at an $852 billion post‑money valuation, the largest private tech raise on record. The round, anchored by Amazon, Nvidia, SoftBank and Microsoft, brings OpenAI’s revenue run‑rate to $2 billion per month and funds massive chip and data center expansion. Follow‑on coverage on April 1 from outlets across India, Europe, Latin America and the Middle East detailed the investor mix, retail participation and plans for an AI “superapp.”
Hyperscalers Lock In Massive 2026 AI Capex
A March 8, 2026 sector report from Chinese brokerage Guosen Securities finds that Microsoft, Meta, Amazon and Google all sharply increased 2025 Q4 capital expenditure, with aggressive 2026 guidance largely driven by AI infrastructure. Microsoft’s FY26 Q2 capex hit $37.5 billion, Google guided $175–185 billion for 2026 capex, and Amazon plans about $200 billion, with much of the spend earmarked for GPUs, custom AI chips and cloud AI services.
Claude user surge challenges ChatGPT dominance
On March 7, 2026 AI Insider reported that Anthropic’s Claude app has overtaken ChatGPT in U.S. daily mobile downloads and reached 11.3 million daily active users, with more than 1 million sign‑ups per day since late February. The growth follows the Pentagon’s decision to label Anthropic a supply‑chain risk, even as Microsoft, Google and AWS reaffirmed they will keep offering Claude for non‑defense workloads and a separate Mozilla partnership saw Claude Opus 4.6 uncover 22 Firefox security vulnerabilities in two weeks. ([theaiinsider.tech](https://theaiinsider.tech/2026/03/07/claude-surges-in-user-growth-and-enterprise-adoption-as-anthropic-challenges-pentagon-restrictions/))
Japan backs seven homegrown LLMs for government
On March 6, 2026 Japan’s Digital Agency announced it has selected seven domestically developed large language models, including NTT Data, Customer Cloud, KDDI/ELYZA, SoftBank, NEC, Fujitsu and Preferred Networks, for trial use in its “Government AI” platform GENNAI. A related press release from Customer Cloud confirmed at 13:44 JST that its CC Gov‑LLM is among the models to be evaluated for administrative workflows.([digital.go.jp](https://www.digital.go.jp/news/10d55c63-b3e1-42b9-9cc5-93a06943ae0e))
OpenAI & Anthropic Hit Massive AI Revenues
A March 5, 2026 analysis based on The Information’s reporting says OpenAI has reached a $25 billion annualized revenue run rate, while rival Anthropic has climbed to about $19 billion. The gap between the two has narrowed sharply over the last year as Anthropic’s Claude products gained enterprise traction.
Armenia AI hub scales to 50,000 Nvidia GPUs
US‑based AI cloud firm Firebird announced on February 10 that it secured US export approvals to deliver an additional 41,000 Nvidia GB300 GPUs to Armenia as part of Phase 2 of its AI supercomputing project. The expansion brings the total cluster to 50,000 GPUs and roughly US$4 billion in planned investment, positioning Armenia among the world’s largest AI GPU hubs.
OpenAI revives ChatGPT growth, preps new model
BusinessToday reports that an internal Slack memo shows ChatGPT returned to more than 10% month‑on‑month growth after OpenAI declared a “code red” in December 2025. The same memo, cited by CNBC and summarized on February 10, says OpenAI plans to ship an updated chat model between February 9 and 15, alongside 50% growth in its Codex coding product.
Claude Opus 4.6 Targets Serious Knowledge Work
Italian data outlet InfoData describes Anthropic’s new flagship model Claude Opus 4.6 as optimized for deep, continuous "knowledge work" rather than casual conversation. The article highlights long‑context reasoning, enterprise‑oriented use cases and Anthropic’s strategy to position Claude as a reliable professional tool.([infodata.ilsole24ore.com](https://www.infodata.ilsole24ore.com/2026/02/09/tutto-quello-che-ce-da-sapere-di-claude-opus-4-6-meno-chatbot-da-conversazione-piu-intelligenza-artificiale-pensata-per-lavorare/))
Alphabet taps $20B+ bonds, century debt for AI
Alphabet raised US$20 billion on February 9, 2026 in its largest ever US dollar bond sale, upsizing the deal from an initial US$15 billion on orders exceeding US$100 billion. Follow‑on coverage on February 10 details plans for additional sterling and Swiss franc bonds, including a rare 100‑year “century bond”, to help finance up to US$185 billion in 2026 capex heavily focused on AI data centers.
Alibaba’s Qwen Hits 1B Downloads, Wins Top Prize
China’s Zhejiang provincial government awarded its 2024 Science and Technology Progress First Prize to the project “Key technologies and large-scale applications of the Qwen open-source large model” on February 9, 2026. Coverage notes that Alibaba’s Qwen family has released more than 400 open-source models, with over one billion cumulative downloads and over 1 million enterprise users worldwide.
India backs Sarvam as flagship sovereign LLM
On February 8, 2026, India’s IT minister Ashwini Vaishnaw said the country’s sovereign AI model strategy is “delivering results,” highlighting an advanced foundational model from startup Sarvam AI. The minister noted that Sarvam was selected from 67 proposals to build India’s first sovereign LLM under the Rs 10,300 crore India AI Mission and praised its Indic text, speech and OCR models.
Cerebras Raises $1B to Challenge Nvidia
AI chipmaker Cerebras Systems closed a $1 billion Series H round at a $23 billion valuation on February 4, 2026, led by Tiger Global with AMD, Benchmark, Fidelity and others participating. On February 7, MLQ.ai reported that Benchmark has layered at least $225 million into the round via two “Benchmark Infrastructure” vehicles, ahead of Cerebras’ planned Q2 2026 IPO and a $10 billion compute deal with OpenAI.([mlq.ai](https://mlq.ai/news/benchmark-capital-commits-225m-to-cerebras-in-ai-chip-funding-boost/))
Nvidia eyes $20B stake in OpenAI mega‑raise
Nvidia is close to investing about $20 billion in OpenAI as part of the ChatGPT maker's latest funding round, Reuters reported on February 4, 2026. The deal, still not final, would be Nvidia's largest-ever single investment and part of an OpenAI raise that could reach $100 billion.
Singapore Pours S$1B into Public AI Push
Singapore’s digital minister Josephine Teo has announced more than S$1 billion in funding for public AI research and talent development between 2025 and 2030. The programme will back new AI research centres, scholarships and high‑performance compute infrastructure, with the goal of tripling the domestic AI expert workforce to 15,000.([laregione.ch](https://www.laregione.ch/estero/estero/1899409/singapore-stanzia-oltre-un-miliardo-per-l-intelligenza-artificiale-tra-il-2025-e-il-2030))