SOCIAL MEDIA
Company@X
Sources AI at Meta / @AIatMeta Amazon Web Services / @awscloud Anthropic / @AnthropicAI Cursor / @cursor_ai Google / @Google Google Cloud Tech / @GoogleCloudTech Google …
Sources
- AI at Meta / @AIatMeta
- Amazon Web Services / @awscloud
- Anthropic / @AnthropicAI
- Cursor / @cursor_ai
- Google / @Google
- Google Cloud Tech / @GoogleCloudTech
- Google DeepMind / @GoogleDeepMind
- Grok / @grok
- Hugging Face / @huggingface
- Microsoft / @Microsoft
- OpenAI / @OpenAI
- OpenClaw🦞 / @openclaw
- Sequoia Capital / @sequoia
- Tesla / @Tesla
- Twitter / @a16z
- Waymo / @Waymo
- xAI / @xai
- Y Combinator / @ycombinator
Signal of the Day
OpenAI has officially merged ChatGPT and Codex into a unified platform interface, signaling a major strategic shift toward becoming a pure platform provider rather than competing with its customers in product categories. CEO Sam Altman declared that OpenAI aims to deliver AGI via a single interface for personal and enterprise use while offering robust APIs for custom development.
Key Announcements
Tesla · Source Tesla has officially announced the inauguration of its new Semi factory in Nevada, scheduled for next month. Separately, Tesla celebrated real-world performance updates, noting charging times for trips from Los Angeles to Silicon Valley have decreased to eight minutes.
Fastino AI · Source Fastino AI launched GLiNER 2.5, introducing a span-free model architecture for entity extraction and classification that scales linearly with document length and delivers a 24.75-point F1 score gain on XNLI. The release introduces five new capabilities, including long-context extraction, unlimited span lengths, and joint entity-relation extraction. Model weights have been made publicly available under an Apache 2.0 license across small, base, and multi variants.
Vernius Systems · Source Defense tech startup Vernius Systems launched radar seekers designed to turn standard drones into fully autonomous interceptor aircraft. The founders skipped their entire YC S26 batch in San Francisco to build and deploy directly alongside military operators in Ukraine. Within an eight-week timeframe, the company progressed from initial concept to field testing, securing $500 million in Letters of Intent (LOIs) and selling 45 units to a special forces unit.
Stoa Intelligence · Source Stoa launched Stoa Intelligence, a product designed to bring market pricing transparency to the opaque GPU rental and secondary markets. By aggregating private quotes and transaction activity, the platform delivers real-time hardware marks, configuration market levels, and fleet valuation data. This service directly addresses a major bottleneck in valuing collateral and securing GPU-backed hardware financing.
Archal · Source Archal (YC S26) launched API sandboxes specifically engineered to support autonomous AI coding and operational agents. The platform allows agents to programmatically spin up, inspect, and reset over 20 stateful environments, including Slack, Linear, and Datadog, for CI and evaluation testing. This provides crucial infrastructure for safely testing agent behaviors under real-world conditions.
Standard Machines · Source Standard Machines launched reinforcement learning (RL) environments designed to train and evaluate AI models on advanced silicon chip design. Led by Jacob Peake, the company is collaborating with leading frontier AI labs to push reasoning and physical design limits. This launch highlights the industry’s accelerate-to-automation shift in hardware engineering workflows.
OneTriangleAI · Source OneTriangleAI (YC S26) launched an ultra-low-cost, high-performance inference hosting service, debuting with hosting for DeepSeek-V4 Flash. Founded by MIT grads and veteran engineers, the company utilizes key-value (KV) cache optimizations to offer pricing at $0.15/M input and $0.35/M output tokens. This aggressive pricing signals intense cost competition at the foundation model deployment layer.
Also Noted
- Mundo AI (Source): Secured a $20 million Series A round to build multi-modal datasets, evaluations, and research to improve AI’s real-world perception across audio and video.
- Vetta (Source): Shipped a benchmark-tested agent task harness that reduces operational costs to $0.298 per finished long-horizon task while achieving the highest finish rate.
- Superset (Source): Launched Design Mode, allowing users or AI agents to click any element in its in-app browser and modify it through direct chat.
- Jcode bench (Source): Introduced an open, uncontaminatable coding benchmark utilizing iterative optimization of reference implementations instead of static training answers.
- Spectre Intelligence (Source): Launched a multi-strategy quantitative trading firm focused on semiconductors and biotech that deploys proprietary AI traders instead of human talent.
🎧 I can easily generate an audio briefing from these announcements if you want to listen to the highlights on the go.