GPT-6.1 Terra vs GPT-6.1 Sol: API pricing, caching, and long prompts
Terra is the cheaper model on both input and output, so the question is whether Sol's results are worth 2 times the cost of a half cached call. Both have a long prompt tier, so check the full request again above 272,000 input tokens on Terra and 272,000 on Sol.
Terra costs 50% less than Sol on input and 50% less on output, so every call you can move down a tier keeps that difference. GPT-6.1 Terra is OpenAI's mid tier for everyday agentic, coding and writing work. GPT-6.1 Sol sits above it. The price gap only means something if Terra can do the job you have in mind. If it passes your tests, the gap is money you keep on every call. If it fails a share of tasks, the retries and the review time on those tasks decide whether Sol was the better buy. This page prices both on the same requests, so you can see how big the gap really is before you run that test.
See what the step up from Terra to Sol adds to the bill, from cache reads to the long prompt tiers.
By TechCompare · Updated
OpenAI: GPT-6.1 Terra
OpenAI's own price
- Input
- $1
- Output
- $5
- Cached
- $0.05
per 1M tokens
Price checked 2026-11-07
OpenAI: GPT-6.1 Sol
OpenAI's own price
- Input
- $2
- Output
- $10
- Cached
- $0.10
per 1M tokens
Price checked 2026-11-07
Cost Comparison
Based on 100,000 input tokens (50% cached), 5,000 output tokens, and 100 requests.Prices come live from OpenRouter at each maker's own price where it's listed. Temporary deals are left out.
Side-by-side specs
| Spec | GPT-6.1 Terra | GPT-6.1 Sol |
|---|---|---|
| Standard input per 1M tokens | $1.00 (better on this spec) | $2.00 |
| Standard output per 1M tokens | $5.00 (better on this spec) | $10.00 |
| Cache reads per 1M tokens | $0.05 (better on this spec) | $0.10 |
| 100K input + 5K output, no cache hits | $0.125 (better on this spec) | $0.25 |
| 100K input + 5K output, half the input cached | $0.0775 (better on this spec) | $0.155 |
| 500K input + 4K output, no cache hits | $1.03 (better on this spec) | $2.06 |
| Input / output on long prompts, per 1M | $2.00 / $7.50 above 272,000 input | $4.00 / $15.00 above 272,000 input |
| Context window | 1,050,000 tokens | 1,050,000 tokens |
| Batch API input / output per 1M | $0.50 / $2.50 | $1.00 / $5.00 |
How they differ
Input costs $1 per million tokens on Terra and $2 on Sol. Output costs $5 and $10, and cache reads cost $0.05 and $0.10. A request with 100,000 input tokens and 5,000 output tokens, no cache hits, costs $0.125 on Terra and $0.25 on Sol. With half the input read from cache it costs $0.0775 and $0.155, and with all of the input cached it costs $0.03 and $0.06. Over 10,000 of the half cached calls the totals are $775 and $1,550. Both models move the whole request to a higher rate once the prompt gets long, at 272,000 input tokens on Terra and 272,000 on Sol. Above those lines Terra bills $2 for input and $7.50 for output, and Sol bills $4 and $15. A document request with 500,000 input tokens and 4,000 output tokens and no cache hits costs $1.03 on Terra and $2.06 on Sol, with the matching tier applied to each. These examples leave out cache writes and tool fees. The embedded chart follows live OpenRouter rates, which can depend on provider routing, so treat the tables here as a way to see the shape of the gap.
Verdict
Run a fixed set of your own tasks through both models before you move any traffic. Count the attempts each model needs for an accepted result, along with the output tokens and the time a person spends fixing answers. On a half cached call Sol costs 2 times what Terra costs. That means Sol only comes out ahead on cost per finished task if Terra needs more than 2 attempts on average to reach an accepted result, ignoring review time. If Terra gets there in fewer, the gap stays in your favor. Most real workloads sit between the two extremes. A common setup sends routine calls to the mid tier and keeps the step up for the cases that fail a check. Price that split, not just the two ends. At 100,000 input and 5,000 output tokens with half the input cached, sending 9,000 of 10,000 calls to Terra and 1,000 to Sol costs $697.50 for the Terra share and $155 for the Sol share.
Which should you pick?
Choose GPT-6.1 Terra
Pick Terra when your tests show it finishes coding, agent and writing tasks to your standard. It suits high volume work where a small per call gap adds up, such as an agent that makes many turns on a stable prompt prefix or a pipeline that runs a fixed prompt over thousands of documents. Terra has Batch API prices here, which suits a nightly job that can wait for its results. A run of 20,000 documents with 6,000 input and 800 output tokens each and no cache hits costs $100 through batch, against $200 at standard rates. Keep a record of the tasks it fails, since those are the ones to route upward.
Official Terra model referenceChoose GPT-6.1 Sol
Pick Sol when the failures on Terra cost you more than the price gap. That happens when a wrong answer reaches a customer, when a person has to review every result anyway, or when a task needs several passes on Terra to get right. An agent session of 40 turns with 150,000 input tokens each, 90 percent of it cached and 2,000 output tokens per turn costs $2.54 on Sol, against $1.27 on Terra. Sol also has Batch API prices here, so work that can wait has a discounted path. Keep prompts under 272,000 input tokens where you can, since crossing that line moves the whole request to the long prompt rates.
Official Sol model reference