OpenAI Trying to Dethrone Anthropic?
20th April 2026 | Superintelligence Newsletter
Hey Superintelligence Fam 👋
Welcome back! The AI landscape is shifting at breakneck speed this week. The undisputed star is OpenAI’s newly minted GPT-5.5, redefining autonomous agentic workflows and pushing the boundaries of what’s possible.
Meanwhile, open-source giants like DeepSeek are aggressively closing the gap, backed by colossal corporate investments. As autonomous researchers and regulatory frameworks evolve, our AI-driven future feels closer than ever.
Let’s dive into what’s new this week..
OpenAI Launches GPT-5.5 : OpenAI’s GPT-5.5 brings stronger agentic coding, research, computer-use, and knowledge-work performance, scoring 82.7% on Terminal-Bench 2.0 and matching GPT-5.4 latency.
DeepSeek V4 Preview Goes Open-Source : DeepSeek V4 Preview launches with 1M context length, open weights, V4-Pro at 1.6T total parameters, and V4-Flash as a faster 284B-parameter option.
Google Plans Up to $40B Anthropic Investment : Google may invest up to $40B in Anthropic, including $10B now at a $350B valuation, plus compute support of 5 gigawatts over five years.
Google Cloud Unveils New TPU AI Chips : Google Cloud introduced TPU 8t for training and TPU 8i for inference, promising up to 3x faster training and 80% better performance per dollar.
ChatGPT Workspace Agents : OpenAI’s Workspace Agents let teams build shared Codex-powered agents that run workflows across ChatGPT, Slack, files, code, and business tools.
GPT-5.5 : GPT-5.5 brings stronger agentic coding, research, data analysis, document creation, and computer-use skills, scoring 82.7% on Terminal-Bench 2.0.
Grok Voice Think Fast 1.0 : xAI’s flagship voice model handles noisy calls, accents, interruptions, and tool-heavy workflows, supporting 25+ languages with real-time reasoning.
DeepSeek V4 Preview : DeepSeek V4 Preview launches open-source with 1M context, V4-Pro at 1.6T parameters, and V4-Flash as a faster 284B-parameter option.
Automated Weak-to-Strong Researcher : Anthropic’s Claude-based researchers reached 0.97 PGR in five days, versus humans’ 0.23, showing autonomous agents can accelerate real alignment research.
AiScientist : AiScientist uses “File-as-Bus” memory for long-horizon ML research, improving PaperBench by 10.54 points and reaching 81.82% Any Medal on MLE-Bench Lite.
AlphaEval : AlphaEval tests production AI agents through 94 tasks from seven companies across six O*NET domains, exposing real-world gaps missed by clean benchmarks.
The EU and Japan advanced comparative regulatory frameworks for AI therapeutics. Concurrently, UNESCO championed global education by co-developing a dedicated ethics module for the upcoming International Olympiad in AI. Meanwhile, Anthropic’s robust “Mythos” model release prompted the Trump administration to reconsider its previous anti-regulation stance.
Thank you for tuning in to this week’s edition of Superintelligence Newsletter! Stay connected for more groundbreaking insights and updates on the latest in AI and superintelligence.
For more in-depth articles and expert perspectives, visit our website | Have feedback? Provide feedback.
To explore Superintelligence Media : Explore Here
Stay curious, stay informed, and keep pushing the boundaries of what’s possible!
Until Next Time!
Superintelligence Team.