$5 / $30
per million tokens in / out, launched July 9, 2026 with the Sol / Terra / Luna family
Blog
GPT-5.6 Sol pricing: the frontier tier with an output premium
Sol is OpenAI's frontier tier at $5 input / $30 output per million tokens. The input price ties Claude Opus 5 exactly; the output price runs 20% higher, and since generation-heavy work is output-heavy, that gap is bigger than it looks on the rate card.
Skip the read: get a measured recommendation in a few quick questions. Run the Stack Finder
+20%
Sol's output premium over Claude Opus 5's $25. Output is where generation workloads live
1.05M
token context window across the GPT-5.6 family, with 128K max output
GPT-5.6 Sol costs $5 per million input tokens and $30 per million output as of July 2026. The input price ties Claude Opus 5; the output price runs 20% higher. Cache reads keep the 90% input discount, and cache writes cost 1.25x uncached input, about $6.25 per million.
How much does GPT-5.6 Sol cost?
$5 per million input tokens, $30 per million output (OpenAI pricing reference). Cache reads keep the 90% input discount; writing new cache costs 1.25x uncached input, so a Sol cache write runs $6.25/M. The family launched July 9, 2026 sharing a 1.05M-token context window and 128K max output.
Want this priced against your own workload? Run the Stack Finder or start with $100 in credits.
The output-token catch
Frontier comparisons usually stop at the tied $5 input price. The bills do not: code generation, long-form writing, and agentic tool use are output-heavy, and on output Sol charges $30 against Opus 5's $25. Applied to our measured dashboard task's token profile, the same generation runs about $0.0788 at Sol list versus the $0.1158 we metered on Opus 5 and $0.0724 on Opus 4.8: which also shows why per-model verbosity matters as much as the rate card. We cannot meter Sol directly yet (no OpenAI account on our bench); the Sol figure is list math on our measured token counts, labeled as such.
Sol vs the market
| Model | Input $/M | Output $/M | Position |
|---|---|---|---|
| Claude Fable 5 | $10.00 | $50.00 | premium tier above every frontier option |
| GPT-5.6 Sol | $5.00 | $30.00 | OpenAI frontier |
| Claude Opus 5 | $5.00 | $25.00 | the direct rival: tied input, cheaper output |
| GPT-5.6 Terra | $2.50 | $15.00 | the step down inside OpenAI |
Full ladder with measured per-task bills in the market breakdown.
When Sol is the right buy anyway
Price is one column. If your stack is OpenAI-native, your prompts are tuned to the family, and Sol clears failure cases the cheaper tiers do not, a 20% output premium is cheaper than a migration. The expensive mistake is the same one everywhere on this ladder: paying frontier rates on the large fraction of traffic that a mid tier handles identically. That allocation problem is the one we publish receipts on.
GPT-5.6 Sol pricing questions, answered
Is GPT-5.6 Sol more expensive than Claude Opus 5?
On output, yes: $30 versus $25 per million, with input tied at $5. On full bills it depends on each model's verbosity for your prompts, which is why measured task costs beat rate-card math.
What do Sol's caching discounts actually do?
Cache reads bill at 10% of input ($0.50/M), and cache writes cost 1.25x uncached input ($6.25/M). Stable system prompts and shared context make the read discount dominate; churning your cache makes the write premium eat it.
What is the difference between Sol, Terra, and Luna?
Capability tiers of the same July 2026 family: Sol is frontier ($5/$30), Terra mid ($2.50/$15), Luna volume ($1/$6), sharing the 1.05M context window. Most production mixes belong on Terra or Luna with escalation, not on Sol by default.
Is there a Sol equivalent to Anthropic's Fable 5?
No. OpenAI's ladder currently tops out at Sol; Fable 5 occupies a premium tier above every frontier option at $10/$50. Fable-class workloads have no direct OpenAI swap today.
Not sure which model fits?
The Stack Finder asks a few quick questions about your workload and gives you a straight recommendation. No account required.
Frontier-level answers without frontier rates on every request. $100 to verify.
Buy Pareto API credits