Sol
Flagship
Most capable model for complex reasoning, extended coding, agent-driven workflows, and security-focused applications.
$5 / $30 per 1M tokens
GPT-5.6 Series ยท Solana ยท pump.fun
Mid-tier balanced model designed for everyday production workloads with a balance of capability, speed, and cost.
OpenAI introduced GPT-5.6 on June 26, 2026 โ a next-generation model series named after celestial bodies. The number identifies the generation; Sol, Terra, and Luna identify durable capability tiers that advance on their own cadence.
Flagship
Most capable model for complex reasoning, extended coding, agent-driven workflows, and security-focused applications.
$5 / $30 per 1M tokensBalanced ยท Mid-Tier
Strong performance with efficiency for large-scale production environments โ reliable results across high volumes without flagship overhead.
$2.50 / $15 per 1M tokensFast & Affordable
Strong capability at OpenAI's lowest cost โ optimized for speed and affordability when latency and budget matter most.
$1 / $6 per 1M tokensTerra sits at the sweet spot of the GPT-5.6 lineup โ competitive performance to GPT-5.5 while being twice as cheap. Built for teams that need production-grade reliability at scale.
Terra is engineered for everyday production workloads โ the kind of high-volume tasks where organizations need dependable output without paying flagship prices. It delivers the reliability teams expect when shipping real products.
At $2.50 input and $15 output per million tokens, Terra offers competitive performance to GPT-5.5 at half the price. For teams processing millions of tokens daily, that efficiency compounds into meaningful savings.
GPT-5.6 Terra ships with layered safeguards matched to its capability tier โ model-level refusals, real-time misuse classifiers, account-level review, and differentiated access to preserve legitimate work while constraining misuse.
On ExploitGym benchmarks, Terra demonstrates strong improvements in cyber capabilities as reasoning increases โ part of a family-wide step change in security-relevant skills, with safeguards calibrated to each tier.
GPT-5.6 introduces explicit cache breakpoints and a 30-minute minimum cache life. Cache reads receive a 90% discount โ making Terra even more economical for repeated or templated workloads.
During the limited preview, GPT-5.6 models are available through the API and Codex to trusted partners. Broader availability across ChatGPT, Codex, and the API is planned for the coming weeks.
OpenAI positions Terra as the workhorse of the GPT-5.6 family. Where Sol pushes the frontier on Terminal-Bench, GeneBench, and ExploitBench โ Terra brings that generation's improvements to the workloads most teams actually run every day.
Per 1 million tokens. Terra delivers flagship-adjacent capability at mid-tier economics.
| Model | Tier | Input | Output | Best For |
|---|---|---|---|---|
| Sol | Flagship | $5.00 | $30.00 | Complex reasoning, agents, security |
| Terra | Balanced | $2.50 | $15.00 | Production workloads at scale |
| Luna | Fast | $1.00 | $6.00 | Speed & lowest cost |
Cache writes billed at 1.25ร uncached input rate ยท Cache reads receive 90% discount ยท 30-minute minimum cache life
Community token celebrating GPT-5.6 Terra โ OpenAI's balanced mid-tier model. Trade on pump.fun, join the community on X, and stay tuned as Terra goes from limited preview to broad availability.
2j6CApLFvzSaRZ21bAStanY5qKAJwkuit6yn7eQbpump
OpenAI previews Sol, Terra, and Luna with limited API and Codex access for trusted partners.
Community token goes live on Solana. Join the movement around the balanced mid-tier model.
OpenAI plans general availability across ChatGPT, Codex, and the API for Sol, Terra, and Luna.