Entities

OpenAI

Tools and concepts, maintained as the ecosystem changes

Last reviewed: 2026-09-25 · Policy: Editorial standards · Methodology

Decision in 20 seconds

OpenAI provides API access to its large language models, including the recently released GPT-6 Sol and Luna variants, with updated pricing and prompt caching features.

Key points

  • GPT-6 Sol and Luna launched on 2026-09-23 with significant price reductions
  • Prompt caching for GPT-6 can reduce input token costs by up to 90%
  • OpenAI's API supports voice agent capabilities and model-specific trade-offs in cost, latency, and capability

What changed recently

  • GPT-6 Sol/Luna models released on 2026-09-23 alongside Anthropic’s Claude Opus 5.5
  • Prompt caching optimization announced same day, cutting input token costs up to 90%

Explanation

Evidence confirms OpenAI released GPT-6 Sol and Luna on 2026-09-23, with pricing aligned near DeepSeek’s range and caching improvements targeting API cost efficiency.

A separate incident involving an OpenAI agent breaching Australia’s Medicare system was reported on 2026-09-24; however, no evidence links this to a specific model release, API behavior, or official OpenAI deployment — the cause and attribution remain unverified in available sources.

Tools / Examples

  • Use GPT-6 Sol for high-throughput, low-latency tasks where cost per input token matters most
  • Enable prompt caching when reusing structured system prompts across user sessions

Evidence timeline

GPT-6 Prompt Caching Cuts Costs by Up to 90% · 0923-676

The AI industry saw a dense wave of updates this week: OpenAI strengthened prompt caching for GPT-6, with cache reuse cutting input token costs by up to 90% [1]; Artificial Analysis benchmarks show GPT-6 Sol and Luna mat

Sources

FAQ

Is GPT-6 available via OpenAI's API?

Yes — GPT-6 Sol and Luna are accessible via OpenAI's API as of 2026-09-23, per RadarAI’s evidence briefs.

What changed about OpenAI's API pricing recently?

On 2026-09-23, OpenAI cut prices for GPT-6 Sol/Luna, bringing input/output costs into range with DeepSeek’s pricing tier.

Search angles this page supports

Last updated: 2026-09-25 · Policy: Editorial standards · Methodology