OpenAI Cuts GPT-5.6 Sol API Prices

Reuters reported that OpenAI also lowered prices for smaller models late last month: the mid-tier GPT-5.6 Terra received a 20% reduction and the lower-cost Luna model an 80% reduction. The Sol adjustment is therefore part of a sequence of reported price reductions across the GPT-5.6 lineup.

For comparison, Reuters reported that Anthropic lists Claude Fable 5 at $10 per million input tokens and $50 per million output tokens, while Claude Opus 5 is listed at $5 and $25. Those figures are list-price comparisons rather than a performance benchmark, and they do not account for differences in model capabilities, context limits, caching, batching, or enterprise discounts.

For teams whose workloads are dominated by short-context inference, the new Sol pricing lowers direct token costs by 20% for inputs and about 33% for outputs. Output-heavy applications, including coding agents, document-generation workflows, and multi-step agent systems, can be particularly sensitive to output-token pricing because generated reasoning and tool-use traces may substantially increase total inference consumption.

Across the model market, comparable price changes tend to make workload-level measurement more important than list-price comparisons alone. Practitioners evaluating providers generally need to compare task success rates, latency, context behavior, token consumption, and the cost of retries alongside per-token rates.