OpenAI
OpenAI brings GPT-6 Sol and Luna to everyday AI work
The new models aim to make stronger coding and agent capabilities faster and more affordable across the API and Codex.
Share this news
OpenAI has introduced GPT-6 Sol and GPT-6 Luna as faster, lower-cost additions to the GPT-6 family. The company says the models bring improvements in coding, computer use, factuality, and professional work to tasks that do not need the full depth of Astra.
Both models are available through the API and Codex. OpenAI says their API prices are 50% lower than the promotional prices of their GPT-5.6 counterparts, while improved prompt caching can further reduce the cost of repeated context in long-running agents.
Two different jobs, one release
OpenAI positions Sol for complex coding and agent workflows and Luna for focused, high-volume work. The model pages list standard prices of $2 input and $10 output per million tokens for Sol, versus $0.10 input and $0.50 output for Luna. Both list a 1.05-million-token context window, but a larger window does not make a poorly scoped task cheaper or more reliable.
The launch article reports improvements on coding and agent evaluations and compares cost per task with rival models. Those benchmark comparisons come from OpenAI; they are useful directional evidence, not a substitute for testing your own prompts, tools, latency, and failure cases. The API changelog confirms the model IDs and availability in the Responses and Chat Completions APIs.
What I would benchmark
Run the same real task set through both models. Track finished-task quality, retries, tool calls, wall time, and total bill, not only token price. A cheap call that needs five retries can be the expensive choice. Sol may make sense for ambiguous, multi-step work; Luna may win when the task is narrow and repeated at scale. The right answer is in your workload, not in the model name.
Sources and further reading
Share this news
Related Items
- NewsOpenAI launches Dots, agents that keep working after you log off, and GPT-6.1 Sol at a fifth of Astra's priceEach dot runs on GPT-6 Astra with its own cloud computer and connections to more than 4,000 apps, and asks before sensitive actions. GPT-6.1 Sol costs $2 and $10 per million tokens and matches Astra on DeepSWE. Both arrived a day after OpenAI cancelled GPT-6.1 Astra for acting beyond what users had authorized.
- NewsOpenAI and Synopsys build GPT-Synopsys, a model that runs chip design tools on its ownEngineers give it a design goal and the model runs Synopsys' EDA tools, reads the results, makes changes and iterates until the design checks out. OpenAI licenses the tools and the two share revenue, but there is no launch date, price or benchmark yet.
- NewsZ.ai's GLM-Legal nearly matches Gemini 4 Argon on legal agent work, at a seventieth of the costThe legal model, built on GLM-5.3-Flash with reinforcement learning across tens of thousands of legal environments, passes 19.2% of the tasks in Harvey's Legal Agent Benchmark, seventh on Vals AI's leaderboard at 15 cents a task. It is a gated beta for law firms, with no published price and no open weights for now.
Comments
Questions, corrections, or your own take are all welcome. Sign in with GitHub to join in.