SUN, AUGUST 09, 2026
Independent · In‑Depth · Practitioner‑Tested
✎ News

AI News August 9 2026 — NVIDIA Open-Sources NOOA Agent Framework, Anthropic Launches $1.5B Enterprise JV

August 9: NVIDIA Labs open-sourced NOOA (Apache 2.0) — agent = one Python class, SWE-bench 82.2%, pip install nooa. Run in container/VM (AST checks are not containment). Anthropic launched Ode With Anthropic — $1.5B JV with Blackstone + H&F, 100 engineers, targeting mid-sized banks, health systems, and manufacturers with Claude in regulated sovereign deployments.

By AIToolsRecap August 9, 2026 5 min read 553 views
Home Articles News Claude AI AI News August 9 2026: NVIDIA NOOA Open-Source ...

AUGUST 9, 2026 — NVIDIA NOOA OPEN-SOURCE · ANTHROPIC $1.5B ENTERPRISE JV

  • NVIDIA NOOA open-sourced: Apache 2.0 Python agent framework — agent = one Python class (methods are actions, fields are state, docstrings are prompts). SWE-bench Verified 82.2%, CyberGym L1 86.8%, ARC-AGI-3 85.1%. Model-agnostic via LiteLLM. v0.0.8 alpha, Python 3.12-3.13. Security warning: AST checks are NOT containment — always run in container or VM. Full review + install guide →
  • Anthropic launches Ode With Anthropic: $1.5B JV with Blackstone and Hellman & Friedman now live with 100 engineers. Backed by Goldman, General Atlantic, Apollo, GIC, Sequoia. Targets mid-sized banks, health systems, manufacturers. Deploys Claude in regulated sovereign configurations. Launched same week as CGAO hire — coordinated enterprise + regulatory push. Full analysis →

Story 1 — NVIDIA NOOA: The Cleanest Agent Framework So Far

Per NVIDIA's technical blog, NOOA's key innovation is structural: an agent is a Python class, not a collection of prompt templates and callback graphs. Methods whose body is ... are completed by an LLM at runtime. Methods with normal bodies are deterministic Python. Everything — state, prompts, actions, contracts — lives in one class that can be tested, traced, and version-controlled like ordinary software. SWE-bench Verified 82.2% is a strong result for an alpha framework. The containment warning in the README is important: AST checks do not stop code that is actively trying to escape a sandbox. Run NOOA agents in a container, VM, or NVIDIA OpenShell. Install with pip install nooa. Full review →

Story 2 — Anthropic's Enterprise Push: Ode + CGAO + FLI C+

Ode With Anthropic is the implementation vehicle that turns Anthropic's model API into an enterprise product a bank compliance team can sign off on. The $1.5B JV with Blackstone and H&F provides the private equity distribution relationships that Anthropic's direct sales motion doesn't reach. 100 engineers are live deploying Claude at mid-sized banks, health systems, and manufacturers in data-sovereign, compliance-ready configurations. Combined with Tino Cuéllar's CGAO appointment (former CA Supreme Court Justice) and Anthropic's Politico-confirmed state-by-state AI safety law push, this week's Anthropic news tells one consistent story: the company is making a coordinated move into regulated enterprise as its primary growth market. The FLI C+ safety rating — the highest of any frontier lab — is the commercial differentiator in those conversations. Full analysis →

Tags
AI NewsGenerative AIAnthropicNvidiaAI Agents2026

Spot an inaccuracy?

We verify facts before publishing and correct errors promptly. If something in this article is wrong or outdated, let us know.

Report an error →