大模型网摘GPT-6.1 Terra API Quick Reference: $1/$5 Pricing, the Sol/Luna Price Gap, and How to Choose
Following our lineup piece, this one is purely about access: how GPT-6.1 Terra is priced, how it differs from its siblings, and which workloads it suits. All pricing figures come from the numbers OpenAI disclosed with the October 26 release.
Pricing and the price coordinate
GPT-6.1 Terra costs $1 per million input tokens and $5 per million output tokens. The coordinate within OpenAI's lineup makes it concrete: that is almost half the price of GPT-5.6 Terra and GPT-6.1 Sol, and around 10x the price of GPT-6 Luna. In other words, Sol and Terra are separated by roughly a 2x gap, while Terra and Luna sit about an order of magnitude apart. The 5:1 output-to-input ratio continues the industry-standard structure of pricier outputs.
Channel: available in Foundry
GPT-6.1 Terra has been callable in Microsoft Foundry since October 27. Model strings, inference parameters and rate limits follow the documentation of the platform you access it through; as of this writing we have not verified availability across third-party aggregators one by one, so confirm with your platform before procurement.
Capability boundaries in three sentences
OpenAI's Ted Sanders gave a comment on Hacker News that works as a selection rule of thumb: probably better than Astra and Sol at technical communication; similar on vision and computer use but much faster; a bit worse on most other things. The release announcement likewise scopes Terra to the communication and computer-based tasks that fill the workday. It is not an all-around substitute — it is a clearly specialized mid-tier model.
Three load-based paths
First, high-volume repetitive computer use and form-filling (office automation, RPA-style tasks): mid pricing plus high speed makes Terra a good volume runner. Second, hard problems that need deep reasoning: keep those on Astra or Sol — saving pennies and paying in retries is a losing trade. Third, very high throughput on simple tasks: Luna's price band fits better. Whichever path you take, stacking cache hits and off-peak windows remains the universal cost lever — see our DeepSeek off-peak pricing guide for the method.
Cross-vendor reference points
In the wider market, Terra's mid-tier band now competes mainly with DeepSeek V4.1-Flash (built around agent cost), GLM-5.3-Flash and Kimi's lightweight tiers. For comparison frameworks, see V4.1-Flash vs GPT-6.1 Sol and GLM-5.3-Flash; for DeepSeek's official two tiers, the deepseek-flash and deepseek-v4-pro quick reference lists model names and off-peak pricing; and for the lineup background, read the GPT-6.1 Terra release overview.