Samwise Tech/AI/Robotics Newsletter
Sunday, April 19, 2026
Anthropic’s Claude Opus 4.7 Reclaims Top Spot on Global AI Benchmark Leaderboards
Anthropic released Claude Opus 4.7 on Thursday, narrowly reclaiming the lead on major AI benchmarks including SWE-Bench Verified, where it scores 80.9 percent. The new model improves on Opus 4.6 with sharper vision capabilities and a built-in self-verification system that lets it double-check its own outputs before responding. Anthropic simultaneously launched Claude Design, a new visual collaboration tool that allows users to co-create slides, prototypes, and one-pagers using natural language. Opus 4.7 is available immediately across all Claude products, the Claude API, Amazon Bedrock, and third-party cloud providers. A more powerful but restricted model, Mythos, remains limited to select enterprise cybersecurity partners.
Sources: VentureBeat
NVIDIA Releases Ising — First Open-Source AI Models Built to Accelerate Quantum Computing
NVIDIA launched the Ising model family on April 16, the first open-source AI models purpose-built to calibrate quantum processors and correct quantum errors. The family includes two components: Ising Calibration, a 35-billion-parameter vision-language model that analyzes quantum calibration data with minimal human oversight, and Ising Decoding, which corrects quantum errors up to 2.5 times faster and three times more accurately than conventional methods. Released under NVIDIA’s Open Model License on GitHub and Hugging Face, the models are already in use at Fermilab, Lawrence Berkeley National Laboratory, and Harvard’s engineering school. NVIDIA says Ising addresses one of the core obstacles on the path to fault-tolerant quantum computing.
Sources: NVIDIA Developer Blog
Snap Cuts 1,000 Jobs and 300 Open Roles as AI Now Generates 65 Percent of New Code
Snap announced Wednesday it is laying off approximately 1,000 employees — 16 percent of its global workforce — and eliminating more than 300 open positions, citing rapid advancement of artificial intelligence. CEO Evan Spiegel disclosed that AI systems now generate over 65 percent of Snap’s new code, reshaping how the company sizes its engineering teams. The restructuring will reduce annualized costs by more than 500 million dollars by mid-2026, with restructuring charges of 95 million to 130 million dollars expected in Q2. Affected U.S. employees will receive four months of severance pay, continued healthcare coverage, and accelerated equity vesting.
Sources: TechCrunch
Cadence and NVIDIA Partner to Close the Simulation-to-Reality Gap Slowing Robot Deployment
Cadence Design Systems and NVIDIA announced an expanded partnership on April 16 aimed at closing one of robotics’ most persistent problems: robots trained inside simulations frequently fail when deployed in the physical world. The collaboration integrates Cadence’s high-fidelity multiphysics simulation engines — which model fine-grained contact forces, friction, and materials — with NVIDIA’s Isaac open-source simulation libraries and Cosmos world models for generating synthetic robot training data. The combined stack runs from world-model training through physics simulation to real-world feedback, coordinated by AI agents. The companies also unveiled an integration with Jetson-based edge hardware, shortening the path from simulated training to physical robot deployment.
Sources: AI News
India Launches High-Level AI Governance Body to Map Job Disruption and Deployment Risks
India’s Ministry of Electronics and Information Technology announced the formation of the AI Governance and Economic Group on April 17, a high-level inter-ministerial body chaired by Union Minister Ashwini Vaishnaw. The AIGEG’s mandate is to coordinate national AI policy across ministries and regulators, prepare a ten-year roadmap for AI deployment, and identify which job categories face the greatest disruption risk from automation. A Technology and Policy Expert Committee will support the group with analysis of global regulatory trends. India’s move follows similar governance frameworks adopted this year by the EU, Japan, and the United States, reflecting growing concern about AI’s economic impact on labor markets.
Sources: MediaNama
Manycore Tech Surges 144 Percent on First Trading Day in World’s First Spatial Intelligence IPO
Chinese AI company Manycore Tech debuted on the Hong Kong Stock Exchange on April 17, closing at HKD 18.60 — a 144 percent premium over its IPO price of HKD 7.62 — in what is considered the world’s first listing explicitly positioned around spatial intelligence. Founded in 2015, Manycore builds AI tools for 3D content creation and world modeling, raising HKD 1.22 billion in an offering oversubscribed 1,591 times. The company, led by CEO Victor Huang, argues that spatial intelligence — AI systems capable of reasoning about and navigating physical environments — represents the next major frontier in AI development after language models.
Sources: Fortune
Northwestern Engineers Print Artificial Neurons That Communicate Directly with Living Brain Cells
Researchers at Northwestern University have developed flexible printed artificial neurons capable of activating living brain tissue, a step toward bioelectronics that communicate directly with the nervous system. The devices, built from nanoscale flakes of molybdenum disulfide and graphene deposited onto flexible substrates, generate lifelike electrical spike patterns that successfully triggered responses in mouse Purkinje neurons. Unlike earlier artificial neurons limited to simple on-off signals, these produce single spikes, continuous firing, and complex bursting patterns that match how real neurons communicate. The study, published April 15 in Nature Nanotechnology and led by professor Mark Hersam, has potential applications in neuroprosthetics for hearing, vision, and movement.
Sources: Nature Nanotechnology
Tech Pulse
Top Models: Claude Opus 4.5 (80.9%) | Claude Opus 4.6 (80.8%) | Gemini 3.1 Pro (80.6%)
Market: NVIDIA $4.60T | Apple $3.81T | Microsoft $2.85T
Curated by JD · samwise.agency
Leave a Reply
You must be logged in to post a comment.