Friday, 2 October 2026

Where technology leaders come to think out loud

The BriefArtificial Intelligence

OpenAI launches GPT-6 Sol at half the price of GPT-5.6 Sol

Both new API models undercut their GPT-5.6 namesakes on every listed token rate, though GPT-5.6 Sol’s current rate is itself a promotional price due to run at least until 21 November

The facts
  • Cached input costs 10% of the standard input rate: $0.20 per million tokens for Sol and $0.01 for Luna.
  • Batch and Flex processing are priced at half of standard rates, and fast mode at twice.
  • Regional processing adds a 10% premium where available, and EU data residency is available only with standard processing.
  • On 25 September OpenAI fixed an image encoding bug that had degraded image understanding in both models.
What it means for buyers

Teams running production workloads on GPT-5.6 should plan a move, but on their own evaluations rather than the price sheet. A cheaper model only saves money if it holds quality on the organization’s own prompts, so rerun test suites, especially anything using images, before switching. Check the long-prompt surcharge against real context sizes, and ask suppliers who quoted on old token rates whether the saving reaches the invoice.

OpenAI released GPT-6 Sol and GPT-6 Luna in its application programming interface (API) on 22 September 2026. Sol, which OpenAI says is built for complex coding and agentic workflows, costs $2 (£1.51) per million input tokens and $10 (£7.55) per million output tokens at standard rates, according to the API changelog.

That is half the $4 (£3.02) input and $20 (£15.10) output that GPT-5.6 Sol has cost since 21 August, when OpenAI cut its input rate by 20% and its output rate by 33%. That lower rate is itself promotional, and OpenAI says it runs at least through 21 November.

Luna, which OpenAI describes as its most efficient model for focused, high-volume tasks, costs $0.10 (£0.08) input and $0.50 (£0.38) output, against $0.20 and $1.20 for GPT-5.6 Luna. Both new models have a 1.05 million-token context window and return up to 128,000 tokens. Prompts of more than 272,000 input tokens are charged at twice the input rate and 1.5 times the output rate.

Sources
  1. OpenAI, API changelog, entries for 21 August, 22 September and 25 September 2026. https://developers.openai.com/api/docs/changelog
  2. OpenAI, “GPT-6 Sol”, API model documentation, accessed 26 September 2026. https://developers.openai.com/api/docs/models/gpt-6-sol
  3. OpenAI, “GPT-6 Luna”, API model documentation, accessed 26 September 2026. https://developers.openai.com/api/docs/models/gpt-6-luna
  4. OpenAI, “GPT-5.6 Sol”, API model documentation, accessed 26 September 2026. https://developers.openai.com/api/docs/models/gpt-5.6-sol
  5. OpenAI, “GPT-5.6 Luna”, API model documentation, accessed 26 September 2026. https://developers.openai.com/api/docs/models/gpt-5.6-luna
AdvertisementZoomInfo
More from The BriefAll briefs
ServiceNow

ServiceNow forecast puts UK tech job growth at about 430,000

Elovade

Brigantia retires its name to trade as Elovade UK and Ireland

Learning and Work Institute

AI skills report puts the prize at £80bn and asks for a tax credit

European Commission

Brussels proposes data center ratings and consults on minimum standards

Get the briefs every week in The VETTDD Briefing.