GPT-6.1 Sol Pricing: $2 / $10, With Cached Input Halved
OpenAI's new Sol keeps the old rates, halves the price of cached input and scores close to GPT-6 Astra on independent tests.
OpenAI released GPT-6.1 Sol on 29 September 2026, a week after GPT-6 Sol, at $2 per million input tokens and $10 per million output tokens. Those rates are unchanged, but cached input now costs $0.10 per million, half GPT-6 Sol's $0.20. OpenAI describes it as offering "near-Astra performance" at a lower cost, and now points GPT-6 Sol users to the new model.
GPT-6.1 Sol at a glance
| API model ID | gpt-6.1-sol |
|---|---|
| Input | $2.00 per 1M tokens |
| Output | $10.00 per 1M tokens |
| Cached input | $0.10 per 1M tokens (95% off input) |
| Cache writes | $2.50 per 1M tokens |
| Batch and Flex | 50% off |
| Fast mode | 2× standard prices |
| Long prompts | Above 272K input tokens, the whole request costs 2× input and cache rates and 1.5× output |
| Context window | 1.05M tokens |
| Maximum output | 128K tokens |
Regional processing adds 10% where available. The live figures, with the rest of the GPT-6 line, are on our GPT-6 pricing page.
How it compares with the rest of GPT-6
| Model | Input / output per 1M | Cached input |
|---|---|---|
| GPT-6 Astra | $10.00 / $50.00 | $1.00 |
| GPT-6.1 Sol | $2.00 / $10.00 | $0.10 |
| GPT-6 Sol | $2.00 / $10.00 | $0.20 |
| GPT-6 Luna | $0.10 / $0.50 | $0.01 |
GPT-6.1 Sol costs a fifth of Astra on input and output, and a tenth on cached input.
Performance
OpenAI has not published benchmark scores for GPT-6.1 Sol. On the independent Artificial Analysis Intelligence Index, it scored 52 on 2 October, up from GPT-6 Sol's 48 and one point behind GPT-6 Astra's 53. It is also economical: Artificial Analysis spent an average of $0.72 per task running its index on GPT-6.1 Sol, against $7.62 for Claude Sonnet 5.5, which scored 56 but wrote about six times as many output tokens. Both were tested at their highest effort setting.
It is slower than its predecessor. Artificial Analysis measured 64 output tokens per second, against 84 for GPT-6 Sol and 139 for Claude Sonnet 5.5. Independent scores are a guide, not a substitute for testing on your own tasks.
What it costs in practice
The cheaper cache matters most for agents, which resend the same context at every step. For a coding agent doing 400 tasks a month (about 750M input tokens with 85% served from cache, and 8M output tokens), GPT-6.1 Sol costs $369 a month, against $433 on GPT-6 Sol, $433 on Claude Sonnet 5.5 and $2,163 on GPT-6 Astra.
For a chat workload without caching, such as 10M input and 2M output tokens a month, nothing changes: it costs $40.00, the same as GPT-6 Sol and Claude Sonnet 5.5.
Compare GPT-6.1 Sol with Claude Sonnet 5.5 or with Claude Opus 5.5.
Should you switch?
- From GPT-6 Sol: yes, for most work. The rates are the same or lower, and independent scores are higher. Check two changes first: GPT-6.1 Sol does not support the
noneorminimalreasoning settings, and tool calling works only through the Responses API, not Chat Completions. - From GPT-6 Astra: test GPT-6.1 Sol on your hardest tasks. If it holds up, it costs 80% less.
- If you need fast replies: GPT-6.1 Sol generates output more slowly than GPT-6 Sol, so latency-sensitive apps should measure before moving.
Sources: OpenAI model page, OpenAI changelog, Artificial Analysis.