llm-stats.com's GPT-5.6 Luna model card lists concrete developer-facing specs for the Luna variant: pricing of $0.20 per million input tokens, $0.0200 per million cached input tokens, and $1.20 per million output tokens via OpenAI, alongside a 1.1M-token input context window and 128K-token output context window. It rec The card situates GPT-5.6 Luna within comparison slots against GLM-5.1, Gemini 3 Pro, Qwen3.7 Max, GPT-5, Qwen3.6 Plus, and Kimi K2.5, though several comparator scores are marked pending as of the September 3, 2026 page timestamp. It also serves as a working playground where developers can send requests billed at the l

CodeRabbit's July 9, 2026 engineering review of GPT-5.6 for coding agents and code review explicitly names Luna as part of the family, describing it as "the fastest and lowest-cost tier" and as the "low-reasoning lane for high-volume work where speed and cost are the main constraints." The piece characterizes the GPT-5 The review's coding benchmarks focus on Sol and Terra (Sol reportedly follows through better on messy repo tasks and finds more review issues; Terra is positioned as the cheaper lane to test for scoped work), with Luna serving as the speed/cost tier rather than a deep-reasoning evaluation target. The author concludes t

OpenAI's July 9, 2026 general-availability launch post introduces GPT-5.6 as a three-model family and explicitly names GPT-5.6 Luna as "our most cost-efficient model," positioning it alongside the flagship Sol and the balanced Terra. The post reports that GPT-5.6 Luna, together with Terra, outperforms Anthropic's Claud The same launch post frames GPT-5.6 family gains in terms of "more intelligence from every token" and introduces an "ultra" capability setting that coordinates multiple agents across parallel workstreams; GPT-5.6 Sol is described as achieving state-of-the-art results across coding, knowledge work, cybersecurity, and sc

OpenAI's June 26, 2026 preview post announces a limited preview of the GPT-5.6 series and explicitly names three variants: Sol (flagship), Terra (balanced everyday work), and Luna ("a fast and affordable model"). It states that Luna "brings strong capability at our lowest cost," establishing Luna's positioning as the p The preview describes a staggered rollout: OpenAI previewed its plans with the U.S. government and, at its request, is initially releasing the models only to a small group of trusted partners whose participation has been shared with the government, with general availability planned in the coming weeks. It shares early