
GPT-5.6 Sol Review: Strong Model, Thin Access
OpenAI's GPT-5.6 Sol tops Terminal-Bench 2.1 at 91.9% with its multi-agent Ultra mode, but reward-hacking findings and government-gated access keep it out of reach for nearly everyone.
They summarize our coverage. We write it.
Newsletters like this one rebroadcast our headlines - often without the full review, the source reading, or the analysis underneath. Our weekly briefing sends the work they paraphrase, straight from the desk, before they get to it.
Free, weekly, no spam. One email every Tuesday. Unsubscribe anytime.

OpenAI's GPT-5.6 Sol tops Terminal-Bench 2.1 at 91.9% with its multi-agent Ultra mode, but reward-hacking findings and government-gated access keep it out of reach for nearly everyone.

The Trump administration lifted export controls on Anthropic's Fable 5 and Mythos 5 on June 30, restoring global access today while industry partners draft a four-dimension jailbreak severity framework.

OpenAI's GPT-5.6 family - Sol, Terra, and Luna - sets a new Terminal-Bench 2.1 record at 91.9% with subagent Ultra mode, but remains locked to ~20 government-vetted partners as of launch.

Tracking AI supply-chain attacks, agent exploits, prompt injection, model leaks, and the real-world incidents shaping AI security today.

Claude Mythos 5 is the full release of Anthropic's restricted Mythos family - same weights as Fable 5 but without safety classifiers for cybersecurity and biology, at $10/M input and $50/M output tokens.

The Commerce Department partially lifts its two-week export ban on Anthropic's most powerful model, clearing Mythos 5 for ~100 US companies and agencies while keeping Fable 5 blocked.

The intelligence agencies of five allied nations issued a joint statement warning that frontier AI will fundamentally transform offensive cybersecurity within months, not years - and that most organizations are not ready.

The Trump administration is requiring OpenAI to vet every GPT-5.6 customer individually before granting access, citing cybersecurity capabilities that rival Anthropic's restricted Mythos model.

OpenAI's GPT-5.5-Cyber found CVE-2026-8390 in Firefox's WebAssembly engine before Pwn2Own Berlin - five of six registered exploit entries withdrew.

OpenAI's GPT-5.5-Cyber is a cybersecurity-specialized fine-tune of GPT-5.5, restricted to vetted defenders through the Daybreak Cyber Partner Program and rated 85.6% on the CyberGym benchmark.

A Chinese cybercrime network sold $88/week phishing kits that used Google's own Gemini AI to generate fake sites impersonating banks, carriers, and government agencies at scale.

Claude Fable 5 is Anthropic's first publicly available Mythos-class model, with safety classifiers that fall back to Claude Opus 4.8 for high-risk requests across cybersecurity, biology, and chemistry.