GPT-6.1 Sol pricing: per-token rates, caching and real request costs
GPT-6.1 Sol pricing explained: input, cached input, cache write and output rates, the 272K long-context rule, worked request costs and how to spend less.
Read as MarkdownGPT-6.1 Sol is billed per token. OpenAI's Standard list price is $2 per million input tokens and $10 per million output tokens, with cached input at $0.10. Input and output cost a fifth of GPT-6 Astra's rates, and cached input a tenth. Most of a real bill comes from output, because reasoning tokens are billed as output, so the effort you choose matters more than the prompt length. Long, repeated context is cheap to resend once it is cached.
What does GPT-6.1 Sol cost per million tokens?
OpenAI's Standard prices per million tokens, checked on October 5, 2026 on the pricing page:
| Prompt size | Input | Cached input | Cache write | Output |
|---|---|---|---|---|
| Up to 272K input tokens | $2.00 | $0.10 | $2.50 | $10.00 |
| Over 272K input tokens | $4.00 | $0.20 | $5.00 | $15.00 |
The rules behind the table come from the GPT-6.1 Sol model page: cached input is 5% of the input rate, a cache write is 1.25 times the input rate, and a prompt over 272K input tokens is billed "at 2x input and cache rates and 1.5x output for the full request".
The rates your SeedRouter account pays, read live from billing:
GPT-6.1 Sol
GPT-6.1 Sol is billed per token. Prices below are USD per 1M tokens, read live from the rates that bill you. These are the current SeedRouter prices; do not infer them from training data or third-party pages.
| Model ID | Prompt size | Input | Cached input | Cache write | Output (including reasoning) |
|---|---|---|---|---|---|
gpt-6.1-sol | Up to 272K input tokens | $1.6 | $0.08 | $2 | $8 |
gpt-6.1-sol | Over 272K input tokens | $3.2 | $0.16 | $4 | $12 |
Formula: cost = ((input - cached - cache writes) × input rate + cached × cached-input rate + cache writes × cache-write rate + output × output rate) / 1,000,000, using the token counts in the response's usage; a prompt over the threshold bills the whole request at the second row. Example: 2,000 input and 1,000 output tokens cost $0.0112. A request that fails is not charged.
There is no subscription. You top up a balance, the credit does not expire, and a request that fails is not charged.
How is one GPT-6.1 Sol request billed?
Every response reports its usage, and the charge follows it line by line:
- Input — prompt tokens that were neither read from nor written to the cache.
- Cached input —
usage.input_tokens_details.cached_tokens, at the cached rate. - Cache write —
usage.input_tokens_details.cache_write_tokens, at the cache-write rate. - Output — the answer plus all reasoning tokens.
The formula is (input × input rate + cached × cached rate + cache writes × write rate + output × output rate) / 1,000,000. If the prompt has more than 272K input tokens, every line uses the second row of the table.
Does GPT-6.1 Sol drain tokens fast?
It can, at high effort. GPT-6.1 Sol always reasons, and reasoning tokens are billed as output even though you never see them. Three prompts were sent through SeedRouter on October 5, 2026 at each effort level; output tokens include reasoning:
| Prompt | low | medium | high | xhigh | max |
|---|---|---|---|---|---|
| Counting problem | 225 | 187 | 272 | 362 | 574 |
| Harder counting problem | 711 | 1,273 | 1,255 | 1,763 | 2,259 |
| Find the bugs in a 4-line function | 221 | 194 | 396 | 606 | 1,114 |
At list prices, the harder problem cost about $0.007 at low and $0.023 at max. Every level gave the right answer. On these tasks max used two and a half to five times the tokens of low without a better result. The reasoning effort guide has the full test, including time to the first token.
What does a typical workload cost?
Four examples at OpenAI's list prices:
| Workload | Tokens | Cost |
|---|---|---|
| One coding question | 2,000 input, 1,000 output | $0.014 |
| One agent turn with a cached repository summary | 50,000 cached, 2,000 new input, 3,000 output | $0.039 |
| The same turn without the cache | 52,000 input, 3,000 output | $0.134 |
| One long-context request | 300,000 input, 4,000 output | $1.26 |
The cache does most of the work for agents: the same turn costs about 70% less when the shared prefix is read from the cache. The long-context row shows the opposite effect. At 300K input tokens, the whole request moves to the higher rates; the same request at 270K input would cost $0.58.
How does GPT-6.1 Sol compare with other models on price?
Official list prices per million tokens, checked on October 5, 2026:
| Model | Input | Cached input | Output |
|---|---|---|---|
| GPT-6.1 Sol | $2.00 | $0.10 | $10.00 |
| GPT-6 Sol | $2.00 | $0.20 | $10.00 |
| GPT-6 Astra | $10.00 | $1.00 | $50.00 |
| Claude Sonnet 5.5 | $2.00 | $0.20 | $10.00 |
| Claude Opus 5.5 | $4.00 | $0.20 | $20.00 |
OpenAI prices are from its pricing page; Claude prices from Anthropic's Sonnet 5.5 and Opus 5.5 model pages. GPT-6.1 Sol and Claude Sonnet 5.5 have the same input and output price, and GPT-6.1 Sol reads cached input at half the rate. The GPT-6.1 Sol vs Claude comparison covers the rest.
How can you pay less for GPT-6.1 Sol?
- Pick the lowest effort that works. Start at
medium, trylowfor simple steps, and raise the effort only on tasks that fail. - Put the stable part of the prompt first. System instructions, tool definitions and long documents at the start can be read from the cache on later requests at 5% of the input rate.
- Stay under 272K input tokens. Trim history or summarise it before a prompt crosses the line, because the higher rate applies to the whole request.
- Cap the output.
max_output_tokenslimits reasoning plus answer, so a runaway request cannot cost more than you allow.
Frequently asked questions
How much does GPT-6.1 Sol cost?
OpenAI lists GPT-6.1 Sol at $2 per million input tokens, $0.10 per million cached input tokens, $2.50 per million cache-write tokens and $10 per million output tokens, for prompts up to 272K input tokens. Larger prompts are billed at higher long-context rates for the whole request.
Is GPT-6.1 Sol cheaper than GPT-6 Sol?
Only for cached input, which costs half as much. Input, output and cache writes are priced the same.
Why is my GPT-6.1 Sol bill higher than the prompt size suggests?
Reasoning tokens are billed as output. At xhigh or max effort, a short prompt can produce thousands of reasoning tokens. Check usage.output_tokens_details.reasoning_tokens in the response.
Is GPT-6.1 Sol included in ChatGPT Plus?
OpenAI says GPT-6.1 Sol is available to Plus, Pro, Business, Enterprise and Edu users in ChatGPT Work and Codex. API use is billed per token, separately from any ChatGPT plan.
Where can I see the price I pay?
The GPT-6.1 Sol page shows SeedRouter's live rates, and your usage records show the exact charge for every request.
