OpenAI's New GPT-6 Sol and Luna Models Bring Astra Improvements to Cheaper Tiers

OpenAI today launched GPT–6 Sol and GPT–6 Luna, 2 caller models that travel GPT–6 Astra. OpenAI says the caller models were developed utilizing methods akin to Astra's for improvements successful nonrecreational work, factuality, coding, machine use, and alignment.


GPT–6 Sol is for mundane work, and it costs little than Astra. GPT–6 Luna is an adjacent cheaper model. Luna is optimized for accelerated responses and higher volume, portion Sol has much reasoning capability.

The caller models outperform their predecessors connected OpenAI's benchmark tests, and GPT–6 Sol makes "about fractional arsenic galore mistakes" arsenic GPT–5.6 Sol. GPT–6 Sol shows improvements successful coding and machine usage compared to GPT–5.6 Sol, and it matches oregon beats Claude Fable 5.1 scores connected immoderate benchmarks.

OpenAI compared GPT–6 Sol to Opus 5 connected respective tests, but Anthropic released a caller Opus 5.5 model contiguous that outperforms Astra connected immoderate coding and cognition enactment benchmarks.

GPT–6 Sol and Luna are adopting the improved connection benignant OpenAI introduced with Astra for simplified conversations with little jargon and less low-value details. Answers should beryllium somewhat shorter without losing substance, and the caller models person little rates of misleading claims astir their coding enactment successful OpenAI's alignment testing.

OpenAI is lowering token prices for Sol and Luna compared to their 5.6 counterparts, and punctual caching has been improved. OpenAI says agents volition reuse much context, respond faster, and get 90 percent discounts connected cached input-token reads.

GPT–6 Sol is priced astatine $2 per cardinal input tokens and $10 per cardinal output tokens, down by half. GPT–6 Luna costs $0.10 per cardinal input tokens and $0.50 per cardinal output tokens, down from $0.20 and $1.20, respectively. GPT–6 Sol costs fractional arsenic overmuch per token arsenic Opus 5.5, but Anthropic says Opus 5.5 uses less tokens per task than Opus 5, truthful it's not a consecutive comparison.

GPT–6 Sol and GPT–6 Luna are rolling retired successful ChatGPT Work and Codex starting contiguous for Plus, Pro, Business, Enterprise, and Edu users. GPT–6 Luna is disposable to Free and Go users successful the desktop app. GPT–6 Sol and Luna are not yet disposable successful Chat mode.
This article, "OpenAI's New GPT-6 Sol and Luna Models Bring Astra Improvements to Cheaper Tiers" archetypal appeared connected MacRumors.com

Discuss this article successful our forums