GPT-5.6 Sol
OpenAI's flagship GPT-5.6 family (Sol, Terra, Luna) targeting frontier coding, agentic tasks, and cybersecurity — currently in limited preview.
Updated 2026-06-27
Overview
GPT-5.6 Sol is the top tier of OpenAI's newest flagship model family, announced June 26, 2026 alongside two siblings: Terra (the balanced mid-tier) and Luna (the cost-efficient tier). The pitch is frontier-level reasoning aimed squarely at three areas OpenAI is leaning into hard — software engineering, long-horizon agentic workflows, and cybersecurity — where the company claims new state-of-the-art results on benchmarks including Terminal-Bench.
The three-tier structure is the practical story here. Sol is the heavyweight at $5 input / $30 output per million tokens, Terra slots in at $2.50 / $15, and Luna undercuts both at $1 / $6 — so teams can route easy calls to Luna and reserve Sol for the hard problems. This mirrors the now-standard frontier-lab playbook (think Anthropic's Opus/Sonnet/Haiku split), and the per-token economics put Sol in the same premium bracket as other flagship reasoning models rather than commodity territory.
The notable wrinkle: GPT-5.6 launched in limited preview after what OpenAI describes as coordination with the US government, a framing tied to the model's cybersecurity capabilities. That gating matters — at announcement this is not a model you can simply switch your production traffic to, and the cybersecurity angle cuts both ways (stronger defensive tooling, but also dual-use concern that explains the cautious rollout). Anyone evaluating it should treat availability, rate limits, and final pricing as subject to change until general availability.
Key features
Three-tier family
Sol (flagship), Terra (balanced), and Luna (efficient) share the GPT-5.6 generation but trade capability against cost, letting you route requests by difficulty instead of paying flagship rates for every call.
Agentic & coding focus
OpenAI positions the family around long-horizon agentic tasks and software engineering, claiming new state-of-the-art on Terminal-Bench — a benchmark measuring how reliably a model completes real terminal/coding workflows.
Cybersecurity capability
Major reported gains on cybersecurity benchmarks are a headline feature and the stated reason for the gov-coordinated limited preview — a dual-use capability OpenAI is rolling out cautiously.
Tiered API pricing
Per-million-token pricing scales with tier: Sol at $5/$30, Terra at $2.50/$15, Luna at $1/$6 (input/output), so cost-sensitive workloads can drop to Luna without leaving the family.
Pricing
| Plan | Price | What's included |
|---|---|---|
| Luna (efficient) | $1 / $6 per 1M tokens | Cost-efficient tier for high-volume or simpler tasks (input / output pricing). |
| Terra (balanced) | $2.50 / $15 per 1M tokens | Mid-tier balancing capability and cost for general production workloads. |
| Sol (flagship) | $5 / $30 per 1M tokens | Top-tier model for the hardest coding, agentic, and reasoning tasks. |
Cost-efficient tier for high-volume or simpler tasks (input / output pricing).
Mid-tier balancing capability and cost for general production workloads.
Top-tier model for the hardest coding, agentic, and reasoning tasks.
Pros & cons
Pros
- ✓Three-tier family lets you match model cost to task difficulty instead of overpaying
- ✓Claimed state-of-the-art on Terminal-Bench, a meaningful real-world coding/agentic benchmark
- ✓Luna tier ($1/$6 per 1M tokens) is genuinely cheap for a current-generation model
- ✓Strong focus on agentic, long-horizon workflows rather than just chat
Cons
- ×Launched in limited preview — not broadly available, so you can't reliably build production on it yet
- ×Sol's $5/$30 pricing is premium; costs add up fast on token-heavy agentic loops
- ×Cybersecurity/dual-use framing and gov coordination signal tighter usage gating than a normal release
- ×Benchmark claims are vendor-reported at launch; independent verification is still pending
How it compares
| Tool | Best for | Pricing | Score |
|---|---|---|---|
| GPT-5.6 Sol | — | API usage-based: Sol $5/$30, Terra $2.50/$15, Luna $1/$6 per 1M tokens | 9.1/10 |
| ChatGPT vs ChatGPT → | — | Free tier + Plus $20/mo + Pro $200/mo | 9.5/10 |
| Claude vs Claude → | — | Free tier + Pro $20/mo + Team $30/mo/user | 9.5/10 |
| Gemini vs Gemini → | — | Free tier + Advanced $19.99/mo | 9.2/10 |
Compare head-to-head
Related reading
GPT-5.6 Sol, Terra, Luna: OpenAI's Limited Preview
OpenAI's GPT-5.6 family ships as a gated preview after a US government request. Here's what Sol, Terra, and Luna actually are.
GPT-5.6 Sol vs GPT-5.5: What the Benchmarks Say
OpenAI previewed GPT-5.6 Sol on June 26. Here's how its reported benchmarks stack up against GPT-5.5 — and what to trust.
GPT-5.6 Sol: What Shipped, Pricing & Who It's For
OpenAI's June 26 GPT-5.6 launch splits into three tiers — Sol, Terra, Luna. Here's what shipped, the pricing logic, and who each is for.
Ready to try GPT-5.6 Sol?
Head to the official site to start with GPT-5.6 Sol — pricing and plans are listed above.
Visit GPT-5.6 Sol