News
Research Report
The AI Transformation 100: Concrete Ideas for Fixing How We Work
Key Takeaways
- OpenAI introduces its new Sol, Terra and Luna models.
- New layered safeguards address cyber and biology misuse risks.
- Business, cybersecurity and science users may see lower costs and improved workflow efficiency.
OpenAI is pushing reasoning models deeper into enterprise workflows with a tiered lineup that undercuts its own prior pricing and adds new misuse controls.
The company began a limited preview of its GPT-5.6 family — Sol, Terra and Luna — spanning flagship, mid-tier and budget capability levels. General availability is planned in the coming weeks, with initial access restricted to trusted partners through the API and Codex.
The GPT-5.6 Family: Capability Breakdown
OpenAI claims improvements across coding, biology and cybersecurity workflows.
| Model | Key Capabilities |
|---|---|
| GPT-5.6 Sol | Flagship model with max reasoning effort and subagent ultra mode |
| GPT-5.6 Terra | Balanced everyday model; OpenAI claims 2x cheaper than GPT-5.5 |
| GPT-5.6 Luna | Fast, lowest-cost model in the GPT-5.6 family |
According to OpenAI, Terra delivers performance competitive with GPT-5.5 at half the cost, while Luna targets the company's lowest price point. Sol introduces a new "max reasoning effort" mode and an "ultra mode" that delegates work to subagents for complex tasks.
As part of ongoing engagement with the US government, OpenAI said it previewed the models' capabilities ahead of launch. However, the company added, "We don't believe this kind of government access process should become the long-term default. It keeps the best tools from users, developers, enterprises, cyber defenders and global partners who need them."
Pricing & Availability for the GPT-5.6 Family
Currently in preview, the GPT-5.6 models are only available through the API and Codex to a select group of “trusted” partners and organizations. However, according to the OpenAI, the models will soon become more broadly available to those using ChatGPT, Codex and the API.
| Feature | GPT-5.6 Sol | GPT-5.6 Terra | GPT-5.6 Luna |
|---|---|---|---|
| Input Price | $5 per 1M tokens | $2.50 per 1M tokens | $1 per 1M tokens |
| Output Price | $30 per 1M tokens | $15 per 1M tokens | $6 per 1M tokens |
| Cache Writes | 1.25x uncached input rate | 1.25x uncached input rate | 1.25x uncached input rate |
| Cache Reads | 90% cached-input discount | 90% cached-input discount | 90% cached-input discount |
| Prompt Caching | Explicit cache breakpoints; 30-minute minimum cache life | Explicit cache breakpoints; 30-minute minimum cache life | Explicit cache breakpoints; 30-minute minimum cache life |
| Cerebras Availability | Launching in July at up to 750 tokens per second for select customers | Not specified | Not specified |
Next-Gen LLMs Reshape Enterprise AI
Enterprise AI is moving from prompt-response tools to models that reason, plan and act, with real implications for how work gets done.
Major providers, including OpenAI, Anthropic, Google and DeepSeek, have shifted toward reasoning models built for math, coding and domain-specific tasks rather than simply scaling compute. Context windows are expanding in parallel, with Anthropic's Claude Opus 4.6 reaching the one-million-token threshold via opt-in beta.
Where traditional large language models respond to prompts, agentic systems plan, reason, remember and act toward defined goals with minimal human input. Despite these systems’ autonomous capabilities, experts recommend enterprises implement continuous AI purple teaming, third-party validation and autocorrection layers.
Recent OpenAI News
OpenAI has completed six acquisitions since mid-2024 across collaboration, analytics, governance, security and agent deployment.
Key deals include the $1.1 billion purchase of Statsig, the sub-$400 million acquisition of Neptune for model governance, and the acquisition of Ona to bring persistent cloud environments to its Codex agent platform.
On infrastructure, Nvidia committed up to $100 billion as OpenAI's primary chip supplier, a $38 billion AWS cloud contract diversified compute beyond Microsoft and the company joined the $500 billion Stargate data center venture with SoftBank and Oracle.
On the product side, GPT-5 launched for all ChatGPT users in August 2025, followed by a ChatGPT App Store in December, GPT-5.4 with a one-million-token context window in March 2026 and a $4 billion enterprise deployment unit in May.
Financially, OpenAI closed a $122 billion private funding round in April 2026 — the largest in history — anchored by Amazon, Nvidia and SoftBank, against roughly $2 billion in monthly revenue.