OpenAI's GPT-5.6 — Sol, Terra and Luna — becomes the first frontier model to clear a customer-by-customer US government review before public release, with Sol hitting 750 tokens/sec on Cerebras
OpenAI shipped GPT-5.6 as three variants — Sol, Terra and Luna — and it is the first frontier model to pass a customer-by-customer US government review ahead of going public. Sol reached 750 tokens per second running on Cerebras hardware. A release paradigm and a throughput record in the same announcement.
The review mechanism is the precedent. A customer-by-customer government clearance before public release is a materially different release model from ship-then-restrict, and it establishes a template other labs will be measured against. It also creates a two-tier release landscape: models cleared through review, and models shipped without it.
The 750 tokens/sec figure on Cerebras is the other half. Throughput at that level changes what agentic workloads are economically sensible — the constraint on long-running agents has been wall-clock time as much as token cost. Set against Opus 5 halving its price on the same week, both major labs moved on agent economics simultaneously from different directions.
ThursdAI — July 2026 AI Releases → · Dentro AI — AI News July 2026: Key Events & Releases →