OpenAI launches GPT-6.1 Terra, a cheaper mid-tier model for visual reasoning and computer use
2026-10-26
OpenAI released GPT-6.1 Terra, described by OpenAI as a fast, efficient model combining visual reasoning and computer use with clearer writing, a step up from GPT-5.6 Terra. OpenAI's Ted Sanders described it, per Simon Willison, as a midrange option: better than GPT-6 Astra and GPT-6.1 Sol at technical communication, similar to them on vision and computer use but much faster, and "a bit worse on most other things." It is priced at $1/million input tokens and $5/million output tokens, roughly half the price of GPT-5.6 Terra and GPT-6.1 Sol, and about 10x the price of GPT-6 Luna.
Significance 3: A mid-tier release with a real, disclosed price cut (roughly half the prior Terra/Sol pricing) is a meaningful pricing data point, but it is a mid-tier rather than flagship release that per OpenAI's own account trails the flagship line on most capabilities.
Implications · machine-drafted, not owner judgment
A cheaper, faster mid-tier option widens OpenAI's price ladder beneath GPT-6 Astra/Sol, giving developers a lower-cost default for high-volume vision/computer-use workloads such as browser agents, which pressures competitors' mid-tier pricing (e.g. Gemini 3.8 Flash, Claude Haiku 5.5) even though GPT-6-generation models aren't yet tracked verdict entities.
- Independent benchmarking of GPT-6.1 Terra against Claude Haiku 5.5 and Gemini 3.8 Flash on cost-adjusted quality
- Whether OpenAI's GPT-6 line is added as formally tracked model entities with verified scores
Sources
- curator GPT-6.1 Terra retrieved 2026-10-27
Affected axes
Cost · Output speed