OpenAI announced GPT-6.1 Sol as an upgrade to GPT-6 Sol. For standard text tokens, GPT-6.1 Sol costs $2.00 ($2) per million input tokens and $10.00 ($10) per million output tokens. Compare that to the standard price of GPT-6 Astra, which costs $10.00 ($10) per million input tokens and $50.00 ($50) per million output tokens. That places Sol at exactly one-fifth (20%) of Astra's standard token prices. Meanwhile, GPT-6 Luna provides high-volume processing at $0.10 per million input tokens and $0.50 per million output tokens.
For small business operators evaluating API expenses, Sol provides near-Astra benchmark performance at a lower cost for complex coding, computer use, and professional tasks. Cached input further lowers repetitive prompt expenses to $0.10 per million tokens.
GPT-6.1 Sol Pricing Compared With GPT-6 Astra and Luna
According to the GPT-6.1 Sol model documentation, standard API pricing per million tokens is:
Standard input: $2.00
Cached input: $0.10
Cache writes: $2.50
Standard output: $10.00
Cached input tokens are priced at 5% of the uncached input token rate ($0.10 per million), while cache writes are billed at 1.25 times the uncached input rate ($2.50 per million).
In comparison, GPT-6 Astra lists standard pricing of $10.00 per million input tokens, $1.00 per million cached input tokens, $12.50 per million cache writes, and $50.00 per million output tokens. GPT-6 Luna lists $0.10 per million input tokens, $0.01 per million cached input tokens, $0.125 per million cache writes, and $0.50 per million output tokens.
Model | Standard input / 1M tokens | Cached input / 1M tokens | Cache writes / 1M tokens | Standard output / 1M tokens |
|---|---|---|---|---|
GPT-6 Luna | $0.10 | $0.01 | $0.125 | $0.50 |
GPT-6.1 Sol | $2.00 | $0.10 | $2.50 | $10.00 |
GPT-6 Astra | $10.00 | $1.00 | $12.50 | $50.00 |
All three models feature a 1,050,000-token context window and a 128,000-token maximum output limit. Across each model, prompts with more than 272,000 input tokens (272K) are priced at 2 times input and cache rates and 1.5 times output rates for the full request. Batch and Flex processing options offer prices 50% lower than standard rates, while Fast mode is priced at 2 times standard rates. In Codex, OpenAI notes that GPT-6.1 Sol Ultrafast offers up to 8x faster token generation compared to standard speed, and regional processing adds a 10% premium where available.
Developers access the model using the identifier gpt-6.1-sol. Tool calling is supported through the Responses API for tools including web search, file search, computer use, and code interpreter. GPT-6.1 Sol is available to Plus, Pro, Business, Enterprise, and Edu users in ChatGPT Work and Codex, but is not yet available in Chat.
DeepSWE v1.1 and Task Benchmark Performance
OpenAI evaluated GPT-6.1 Sol across several benchmark tasks to measure quality relative to Astra and earlier models. In the release announcement, OpenAI reported the following verified evaluation results:
Coding: On DeepSWE v1.1, which evaluates complex software-engineering tasks in real codebases, GPT-6.1 Sol matches GPT-6 Astra at roughly one-fifth of the cost, while eclipsing GPT-6 Sol's best score by 6.4 percentage points at a lower reasoning effort and cost.
Computer use: On OSWorld 2.0 (offline set), which evaluates computer-use workflows, GPT-6.1 Sol comes within 2.1 percentage points of Astra's score at maximum reasoning effort at roughly one-seventh the cost per task. It also outperforms GPT-6 Sol by 7 percentage points at less than half the cost.
Document understanding: On GDP.pdf, which tests answering questions using complex PDF documents, GPT-6.1 Sol approaches GPT-6 Astra's performance at roughly one-fifth the cost per task, scoring higher than Opus 5.5 with fallbacks at less than half the cost per task.
Multi-step business workflows: On AutomationBench, which tests agents using 47 tools across sales, marketing, operations, support, finance, and HR, GPT-6.1 Sol scores 2.2 percentage points above Opus 5.5 at medium reasoning effort at roughly one-third the cost, and 4.8 percentage points above GPT-6 Sol at the same setting.
Factuality: On deliberately difficult error-inducing prompts, Sol reduces the share of responses containing a factual error from 11.4% to 7.7% at low reasoning effort (a reduction of approximately 32%), remaining within 1.9 percentage points of Astra's error rate at less than one-fifth the cost per task.
For small business applications, these results indicate that Sol delivers performance close to Astra on software engineering, computer navigation, and document workflows while charging standard token rates that are 80% lower ($2.00 versus $10.00 for input, and $10.00 versus $50.00 for output).
Prompt Caching and Small Business Workload Costs
To review what a small business would pay in practice, operators can calculate costs based on input volume, output volume, and whether prompts reuse cached context.
Hypothetical typical monthly workload: Imagine a small business deploying an automated customer triage or back-office document processing workflow that uses 5,000,000 input tokens and 1,000,000 output tokens per month. Assume every request stays under the 272,000-input-token limit, runs on standard processing without speed premiums, and uses no tool fees. Using the pricing formula cost = (tokens / 1,000,000) * rate:
GPT-6 Luna:
Input cost: (5,000,000 / 1,000,000) * $0.10 = $0.50
Output cost: (1,000,000 / 1,000,000) * $0.50 = $0.50
Total monthly cost: $0.50 + $0.50 = $1.00
GPT-6.1 Sol:
Input cost: (5,000,000 / 1,000,000) * $2.00 = $10.00
Output cost: (1,000,000 / 1,000,000) * $10.00 = $10.00
Total monthly cost: $10.00 + $10.00 = $20.00
GPT-6 Astra:
Input cost: (5,000,000 / 1,000,000) * $10.00 = $50.00
Output cost: (1,000,000 / 1,000,000) * $50.00 = $50.00
Total monthly cost: $50.00 + $50.00 = $100.00
Model | Input cost (5M tokens) | Output cost (1M tokens) | Total monthly cost |
|---|---|---|---|
GPT-6 Luna | $0.50 | $0.50 | $1.00 |
GPT-6.1 Sol | $10.00 | $10.00 | $20.00 |
GPT-6 Astra | $50.00 | $50.00 | $100.00 |
On this uncached workload, GPT-6.1 Sol costs $20.00 per month, compared to $100.00 on GPT-6 Astra and $1.00 on GPT-6 Luna.
Prompt caching significantly reduces input costs when workflows repeat system prompts, instructions, or schemas across multiple calls. Sol's cached input rate of $0.10 per million tokens is 95% lower than its standard $2.00 input rate. Cache writes cost $2.50 per million tokens (1.25 times the uncached rate).
Hypothetical prompt caching workload: Imagine an agent workflow where a business writes 1,000,000 reusable input tokens to cache, reads that cached context across 19 subsequent requests (totaling 19,000,000 cached input tokens), and generates 100,000 output tokens. At documented standard Sol rates:
Cache write cost: (1,000,000 / 1,000,000) * $2.50 = $2.50
Cached read cost: (19,000,000 / 1,000,000) * $0.10 = $1.90
Output cost: (100,000 / 1,000,000) * $10.00 = $1.00
Total cost with prompt caching: $2.50 + $1.90 + $1.00 = $5.40
Hypothetical uncached comparison: Imagine processing those same 20,000,000 input tokens and 100,000 output tokens entirely without caching:
Uncached input cost: (20,000,000 / 1,000,000) * $2.00 = $40.00
Output cost: (100,000 / 1,000,000) * $10.00 = $1.00
Total cost without prompt caching: $40.00 + $1.00 = $41.00
Net savings from caching: $41.00 - $5.40 = $35.60
