OpenAI unveils GPT-5.6 Sol, Terra and Luna but keeps frontier AI locked to select partners under US government review
π/@OpenAI β’
Revision history
10 recorded changes
Want your article here?
Promote with Leviathan News$5/$30 per 1M tokens for Sol and a planned 750 tok/s Cerebras path in July put frontier agent loops within range for audit shops, protocol ops, and MEV research. Access gating is the part crypto should care about: weeks of model edge accruing to approved partners is exactly how a compliance moat forms before the market can benchmark the tool. Open weights and self-hosted coding agents just got a cleaner narrative.
Top comment by @Benthic
Explore the topic
More on GPT
OpenAI expands Daybreak with GPT-5.6-Cyber, a specialized model for authorized vulnerability research, exploit validation and advanced cyber defense
π/@Eric_Wallace_ Β·
OpenAI upgrades ChatGPT with GPT-5.6 Sol for Plus & Pro users, boosting Instant and deep reasoning, while GPT-5.6 Luna rolls out to Free & Go users
π/@OpenAI Β·
UK's AI Security Institute says Anthropic's Claude Mythos 5 and OpenAI's GPT-5.6 Sol engaged in potentially harmful online activity during a permissive cybersecurity evaluation
π/@AnthropicAI Β·
Moonshot's new Kimi K3 is the largest open weight model ever at 2.8T param, and scores the same as GPT-5.6 and Fable 5 on benchmarks. However, it is signficantly cheaper at $3/15 per M input/output, Sonnet 5 pricing
π/@cline Β·
OpenAI unveils GPT-Red, an automated AI red teamer that uses adversarial self-play to uncover prompt injection flaws and strengthen future models
π/@OpenAI Β·
Trump gives OpenAI the nod to roll out GPT-5.6 widely this Thursday after extra testing and government review.
Axios Β·
OpenAI expands Daybreak with GPT-5.6-Cyber, a specialized model for authorized vulnerability research, exploit validation and advanced cyber defense
π/@Eric_Wallace_ Β·
OpenAI upgrades ChatGPT with GPT-5.6 Sol for Plus & Pro users, boosting Instant and deep reasoning, while GPT-5.6 Luna rolls out to Free & Go users
π/@OpenAI Β·
UK's AI Security Institute says Anthropic's Claude Mythos 5 and OpenAI's GPT-5.6 Sol engaged in potentially harmful online activity during a permissive cybersecurity evaluation
π/@AnthropicAI Β·
Moonshot's new Kimi K3 is the largest open weight model ever at 2.8T param, and scores the same as GPT-5.6 and Fable 5 on benchmarks. However, it is signficantly cheaper at $3/15 per M input/output, Sonnet 5 pricing
π/@cline Β·
OpenAI unveils GPT-Red, an automated AI red teamer that uses adversarial self-play to uncover prompt injection flaws and strengthen future models
π/@OpenAI Β·
Trump gives OpenAI the nod to roll out GPT-5.6 widely this Thursday after extra testing and government review.
Axios Β·