Samwise Tech/AI/Robotics Newsletter
Friday, July 10, 2026
OpenAI Launches GPT-5.6 in Three Tiers as Every Major AI Lab Goes Public Simultaneously
OpenAI launched its GPT-5.6 model family Thursday with three pricing tiers — Sol ($5/$30 per million tokens), Terra ($2.50/$15), and Luna ($1/$6) — all available simultaneously across ChatGPT, Codex, and the API. Sol is OpenAI’s declared best coding model yet, claiming 2.8 percentage points above Anthropic’s Fable 5 on the Coding Agent Index while using less than half the output tokens at roughly one-third the cost. OpenAI also describes GPT-5.6 as its strongest cybersecurity model yet. The public launch marks the first day in AI history where OpenAI, Anthropic, and SpaceXAI each have a publicly accessible frontier model available at the same time.
Sources: TechCrunch Share ↗ ✉︎ Email 💬 Text
Humanoid Robots Controlled by Surgeons Perform World-First Gallbladder Surgery on Live Pigs
Surgeons remotely piloting Unitree G1 humanoid robots performed gallbladder removal surgery on living pigs in a preclinical trial published Thursday in the journal Nature, marking the first time humanoid robots have completed a minimally invasive surgical procedure on live animals. The Unitree G1 — 5 feet tall and weighing 60 pounds — could make robotic-assisted surgical care accessible in smaller hospitals and rural clinics lacking the resources for traditional surgical robots, which cost up to several million dollars and weigh roughly 1,800 pounds. Researchers say the approach could eventually allow surgeons to perform complex procedures remotely across geographic distances.
Sources: Ars Technica Share ↗ ✉︎ Email 💬 Text
How the Government Decided OpenAI’s Frontier Model Was Safe to Release
The 13-day government-coordinated preview for OpenAI’s GPT-5.6 Sol — during which access was limited to approximately 20 vetted partner organizations while government agencies evaluated safety — concluded Thursday with a full public launch. The voluntary process gave the NSA and other agencies time to run classified capability benchmarks and assess whether Sol crossed any classified thresholds. OpenAI characterized the delay as an anomaly, with the company saying such restrictions “shouldn’t be the norm.” The review establishes a template that may inform how future frontier models — including Grok 5 and Gemini 3.5 Pro — are handled ahead of a formal governance framework expected August 1.
Sources: TechCrunch Share ↗ ✉︎ Email 💬 Text
Meta’s Custom MTIA AI Chips Cleared Testing in Six Weeks, Enter TSMC Production in September
Meta’s latest MTIA (Meta Training and Inference Accelerator) chip cleared its hardware testing phase in approximately six weeks and is slated to enter manufacturing at Taiwan Semiconductor Manufacturing Company in September, targeting reduced GPU costs amid a global component shortage. Meta is working with Broadcom on chip design and sourcing RAM from Samsung, storage from SanDisk, and fiber-optic equipment from Sumitomo Electric. The four new MTIA chips, detailed publicly in March, will be used to train ranking and recommendation algorithms, support broader AI workloads, and handle inference for Meta’s consumer applications — part of the company’s broader push to reduce dependence on Nvidia GPUs.
Sources: TechCrunch Share ↗ ✉︎ Email 💬 Text
Nvidia’s Stock Falls 15% From Peak as Memory Replaces GPUs as AI’s Hottest Trade
Nvidia’s stock has fallen roughly 15% from its May peak even as projected revenue continues to grow, leaving the company trading below the S&P average on a price-to-earnings basis — an unusual position for the firm that created the modern AI compute market. Investors are shifting capital into memory companies instead, with Micron nearly tripling in value as DRAM emerges as the new data-center bottleneck following a partial easing of the GPU shortage. The shift illustrates how Nvidia’s success in resolving the GPU scarcity of recent years has moved the market premium to the next constraint in the AI infrastructure stack.
Sources: TechCrunch Share ↗ ✉︎ Email 💬 Text
Meta Enters AI Coding Market With Paid Muse Spark 1.1, Competing Against OpenAI and Anthropic
Meta launched Muse Spark 1.1 Thursday through a new Model API, entering the paid AI coding market for the first time as the company competes directly with OpenAI Codex and Anthropic’s agentic coding tools. The launch marks Meta beginning to charge for AI model access — a shift from the company’s prior strategy of free or open-weight releases. The model claims strong agentic and coding performance, with independent benchmark evaluations still pending. The move positions Meta alongside OpenAI and Anthropic in the rapidly expanding market for AI developer tools, with all three releasing major coding-focused AI updates on the same day.
Sources: TechCrunch Share ↗ ✉︎ Email 💬 Text
Google Opens AlphaEvolve AI Algorithm-Discovery Agent to All Cloud Customers
Google made AlphaEvolve generally available to all Google Cloud customers on the Gemini Enterprise Agent Platform Thursday, opening access to an AI system previously limited to private preview since December 2025. AlphaEvolve is a code optimization and algorithm discovery agent built on Gemini, designed to help solve complex computational problems. During its early access period it was tested across logistics, semiconductors, genomics, high-performance computing, and financial services. The tool uses large language models to iteratively propose and evaluate algorithmic solutions. The GA release arrives on the same day Google’s main frontier model competitors launched major AI updates.
Sources: Google Cloud Blog Share ↗ ✉︎ Email 💬 Text
Tech Pulse
Top Frontier Models: Claude Mythos 5 (95.5%) | Claude Fable 5 (95.0%) | Claude Opus 4.8 (88.6%) [SWE-bench Verified]
Top Open Source Models: GLM-5.2 (81.0%) | Kimi K2.5 (76.8%) | DeepSeek V4-Pro (67.2%) [Terminal-Bench / SWE-bench Verified]
Top Small Models (15–50B): Qwen3-30B-A3B (67.4%) | Mistral Large 2 (48B, 59.2%) | GLM-4-32B (55.8%) [SWE-bench Pro]
Top Edge Models (0–15B): Qwen3-8B (42.1%) | Phi-4 (14B, 37.8%) | Gemma 3 9B (31.4%) [SWE-bench Lite]
AI Leaders: Nvidia $4.8T | Alphabet $4.3T | Microsoft $2.8T
Robotics Leaders: Intuitive Surgical $230B | ABB $100B | Fanuc $25B
Curated by JD · samwise.agency