GPT-6.1 Sol Pricing: Cost per Agent Task
GPT-6.1 Sol is priced at $2 per 1M input tokens, $0.10 per 1M cached input tokens and $10 per 1M output tokens. A 20-step coding agent task with 40k tokens of mostly cached context costs about $0.47, so 1,000 tasks a month runs roughly $470.
Last updated: October 2026
Key takeaways
- GPT-6.1 Sol lists at $2 input, $0.10 cached input and $10 output per 1M tokens.
- OpenAI says it gets near GPT-6 Astra intelligence at a fifth of Astra's standard token prices.
- Cached input is half the price it was on GPT-6 Sol, which helps long agent loops most.
- Output still dominates the bill on short tasks, so measure output tokens per task.
- Compare cost per completed task, not cost per token, before you switch.
How much does GPT-6.1 Sol cost per million tokens?
OpenAI launched GPT-6.1 Sol at DevDay as an upgrade to GPT-6 Sol, aimed at agentic coding, computer use and professional work. The listed rates are below. See the full catalog on the OpenAI provider pricing page.
| Token type | Price per 1M tokens | |---|---| | Input | $2 | | Cached input | $0.10 | | Output | $10 |
The cached rate is described as halved versus GPT-6 Sol, which implies the earlier cached rate was $0.20. That is an inference from the announcement, so verify it before you rely on it in a contract or a forecast.
What does a GPT-6.1 Sol agent task cost?
Say a coding agent finishes a task in 20 steps. Each step sends a 40k-token context, of which 35k is cached and 5k is new, and returns 1k output tokens. These inputs are illustrative.
| Line item | Tokens per task | Rate | Cost per task (illustrative) | |---|---|---|---| | Cached input | 700k | $0.10 | $0.07 | | Uncached input | 100k | $2 | $0.20 | | Output | 20k | $10 | $0.20 | | Total | | | $0.47 |
At 1,000 tasks a month that is about $470. At the older $0.20 cached rate the same task would cost $0.54, so the cache cut saves about 13 percent here. That is useful but not dramatic. The bigger levers are fewer steps per task and shorter outputs, which is the same lesson we found in our post on GPT-5.6 Sol, Terra and Luna pricing.
Is GPT-6.1 Sol cheaper than GPT-6 Astra?
OpenAI's claim is near-Astra intelligence at a fifth of Astra's standard token prices. If that holds, Astra's implied standard rates would be around $10 input and $50 output per 1M tokens, which would put the same agent task near $2.35 instead of $0.47. These Astra figures are implied from the one-fifth claim, not a published rate card, so check them before you act.
The practical question is not the headline ratio. It is whether Sol finishes your task as reliably as Astra on your own prompts. A cheaper model that needs two attempts is not cheaper. For a framework, read why cost per task beats cost per token.
How does GPT-6.1 Sol compare with other frontier models?
Google's Gemini 4 Argon is reported at an identical $2 and $10 per 1M tokens at launch, rising later, so the contest is about reliability and ecosystem more than sticker price. Anthropic is the other obvious rival. Use the OpenAI vs Anthropic comparison to put the list prices side by side at your own token mix.
What about speed and multi-agent delegation?
GPT-6.1 Sol supports multi-agent delegation in the Responses API. Delegation can raise capability, but each sub-agent call is another set of input and output tokens. Budget the delegation depth, not just the top-level prompt, or your cost per task will drift upward unnoticed.
The source is OpenAI's own post, Introducing GPT-6.1 Sol, which has the capability claims.
Takeaway: a lower sticker price helps only when your cost per completed task actually falls, so test on your workload.
To test your own mix of input, cached and output tokens, open the LLM cost calculator.
Frequently asked questions
How much does GPT-6.1 Sol cost?
GPT-6.1 Sol is priced at $2 per 1M input tokens, $0.10 per 1M cached input tokens and $10 per 1M output tokens. That is the standard rate listed at launch.
What is the GPT-6.1 Sol API price per million tokens for cached input?
Cached input is $0.10 per 1M tokens, which OpenAI describes as half the cached price of GPT-6 Sol. It rewards long, repeated contexts such as agent loops.
Is GPT-6.1 Sol better than GPT-6 Astra?
OpenAI claims near-Astra intelligence at a fifth of Astra's standard token prices. Whether it matches Astra on your task depends on your prompts, so run a side-by-side test.
How do I estimate GPT-6.1 Sol cost for my app?
Count average input, cached input and output tokens per task, multiply by the three rates, then multiply by monthly task volume. Add a buffer for retries and multi-agent calls.
FAQ schema (JSON-LD)
Place this inside a <script type="application/ld+json"> tag in the page head.
More from the blog
LLM EconomicsJevons Paradox in AI: Why Cheaper Tokens Mean Bigger Bills
Jevons paradox in AI means that when tokens get cheaper, people use so many more that total spend rises. If price per task halves from $0.10 to $0.05, volume must double just to hold spend flat, and a tripling lifts spend 50 percent.
LLM EconomicsOpenRouter Pricing: What a Gateway Really Costs
OpenRouter pricing is the underlying model rate plus whatever platform fee the gateway adds, so the real question is net cost. Say a 5 percent fee on $10,000 of monthly tokens adds $500, and routing 15 percent of spend to cheaper models saves $1,500, a net gain of $1,000.
LLM EconomicsMoE Cost per Token: Why Active Parameters Rule
For a mixture-of-experts model, compute per token follows active parameters, while the GPUs you must rent follow total parameters. A model with 40B active out of 400B total can be cheaper per token than a dense 70B at high utilization, and more expensive when traffic is thin.
Pricing math, in your inbox.
One short note a week on AI pricing, token economics, and margin. No spam, unsubscribe anytime.