Cutoff Data
GPT-5.6 Sol vs GPT-5.6 Terra
What each model actually costs and how fast it answers, measured from our own billed API calls rather than the published rate cards.
Last reading 8/11/2026, 10:17:21 AM UTC
| Measure | GPT-5.6 Sol | GPT-5.6 Terra |
|---|---|---|
| Provider | OpenAI | OpenAI |
| Input / 1M | — | — |
| Output / 1M | — | — |
| Cost, 10M in + 2M out | — | — |
| Context | — | — |
| Median response | 1.33s | 1.17s |
| Uptime, 7 days | 28.3% | 28.3% |
| Benchmark score | — | 90.9% |
| Cost per task | — | — |
| Readings, 7 days | 92 / 0 | 92 / 21 |
Highlighted figures are the better of the two on that measure. The monthly cost line assumes 10M input and 2M output tokens at the prices we observed.
Keep going
Full readings for GPT-5.6 Sol and GPT-5.6 Terra, or see every model on the model index, status board and benchmark table.
How this is measured
Both models get the same prompt through the same gateway account every hour. We record the latency, the billed cost and the tokens consumed, then derive an observed price per million tokens from those calls.
Benchmark figures come from a fixed suite of code-graded tasks re-run against both models. Scores are the share of available points earned.
These are one buyer's readings on one route. Enterprise pricing, batch discounts and cached-input rates will differ.