GPT-6.1 Sol vs GPT-6 Sol cost.
OpenAI released GPT-6 Sol on 22 September and GPT-6.1 Sol a week later at the same headline price. The difference is in the cache.
- Input gap
- same
- Output gap
- same
- Cached gap
- 2×
- Rates verified
- 2026-10-02
Current-model note: Both models are current. GPT-6 Sol was released on 22 September 2026 and GPT-6.1 Sol on 29 September 2026.
| Current model | Input / 1M | Cached / 1M | Cache write / 1M | Output / 1M | Context |
|---|---|---|---|---|---|
| GPT-6.1 SolOfficial pricing ↗ | $2.00 | $0.10 | $2.50 | $10.00 | 1,050,000 |
| GPT-6 SolOfficial pricing ↗ | $2.00 | $0.20 | $2.50 | $10.00 | 1,050,000 |
Rates verified 2026-10-02. The browser may refresh them from the live catalogue. Provider pricing and tiers remain authoritative.
GPT-6.1 Sol: exact local BPE tokenization
GPT-6 Sol: exact local BPE tokenization
PDF, DOCX, code, text, and data files are extracted locally and applied to both models.
GPT-6.1 Sol: documented formula
GPT-6 Sol: documented formula
Prompt text, document contents, and image pixels stay in the browser.
Text, code, PDF, DOCX, images, audio, or video · 12 MB per file
Your input, measured across providers.
Choose a provider card to inspect every available model.
Workload & Scaling Planner +
What five real workloads actually cost
Headline per-token prices rarely decide a bill. Request shape does. These figures are calculated from the verified rates for GPT-6.1 Sol and GPT-6 Sol, including any long-context tier that applies once a request crosses its threshold.
| Workload | Input | Output | GPT-6.1 Sol | GPT-6 Sol | Difference |
|---|---|---|---|---|---|
| Short chat turn A typical assistant exchange. | 1,000 | 500 | $0.0070 | $0.0070 | about equal |
| RAG answer Five retrieved chunks plus a question. | 12,000 | 800 | $0.032 | $0.032 | about equal |
| Code review A medium pull request with surrounding files. | 60,000 | 2,000 | $0.140 | $0.140 | about equal |
| Whole-document analysis A long report or contract read in one call. | 300,000 | 4,000 | $1.26 (tier) | $1.26 (tier) | about equal |
| Full-context load Filling most of a one-million-token window. | 900,000 | 4,000 | $3.66 (tier) | $3.66 (tier) | about equal |
"(tier)" marks a request that crossed a long-context threshold, so it is billed above the headline rate. Output length is the assumption most worth changing for your own case: paste a real prompt into the calculator above to replace these with your numbers.
The cached-input rate is the number most comparisons miss
Agents, chat threads and RAG pipelines resend the same system prompt and context on every call. Those repeated tokens bill at the cached rate, not the headline rate, so a model can be cheaper on paper and more expensive in production. Here the gap is 2.0x: GPT-6.1 Sol reads cached tokens at $0.100 per million.
| Model | Fresh input / 1M | Cached input / 1M | Discount | Break-even |
|---|---|---|---|---|
| GPT-6.1 Sol | $2.00 | $0.100 | 95% | 1 read |
| GPT-6 Sol | $2.00 | $0.200 | 90% | 1 read |
Break-even assumes a cache write costs about 1.25x the base input rate, which is what OpenAI and Anthropic currently document for a short time-to-live. It answers one question: how many times a cached prefix must be re-read before caching is cheaper than paying full price each call. Above that count, every further read saves the discount shown.
Identical sticker, half the cached rate
Both list $2 per million input tokens and $10 per million output, and both charge $2.50 per million to write to the cache. The difference is the read: $0.10 per million on GPT-6.1 Sol against $0.20 on GPT-6 Sol.
The same halving applies above the long-context threshold of 272,000 input tokens, where both models move to $4 input and $15 output, and cached reads cost $0.20 against $0.40.
A twentieth of the input price is the cheapest cached read OpenAI publishes. For an agent that resends a 100,000-token prefix on every turn, the cached tokens are most of the bill, so halving them is close to halving the invoice.
What the independent numbers say
On version 4.3.2 of the Artificial Analysis Intelligence Index, GPT-6.1 Sol scored 52 and GPT-6 Sol scored 48. Cost per index task was $0.72 against $1.06, so the newer model scored higher and cost less to run.
Artificial Analysis now marks GPT-6 Sol as superseded by GPT-6.1 Sol. Unless you are pinned to a snapshot for reproducibility, there is little reason to start new work on GPT-6 Sol.
The caching rules changed too
On GPT-5.6 and later, OpenAI charges 1.25 times the input rate to write a prefix into the cache. Earlier models wrote for free, so an older cost model will understate these.
The cache lifetime on these models is a single 30-minute setting, where earlier models offered a short in-memory cache or 24 hours. A workload that reused a cache across a day needs rechecking.
Our <a href="/prompt-caching-cost/">prompt caching comparison</a> prices a twenty-turn agent loop on both, including the write.
Model the workload, not the marketing price.
If your workload resends a long prefix, GPT-6.1 Sol is cheaper for the same headline price, because cached input costs half as much. If every request is fresh, the two cost the same and the choice turns on quality.
Review calculation methodology →Measure against the model you use
Other comparisons worth running
Compare approaches, not just models
Frequently asked questions
What is the difference between GPT-6.1 Sol and GPT-6 Sol?+
The headline price is identical at $2 per million input tokens and $10 output. GPT-6.1 Sol halves cached input to $0.10 per million, and it scores higher on the Artificial Analysis Intelligence Index, 52 against 48.
Should I migrate from GPT-6 Sol?+
For cache-heavy workloads, yes: the cached rate is half. For workloads with no reuse, the two cost the same. Artificial Analysis lists GPT-6 Sol as superseded by GPT-6.1 Sol.
Does OpenAI charge to write to the cache?+
Yes, on GPT-5.6 and later. A cache write costs 1.25 times the input rate, which is $2.50 per million on both of these models. Models before GPT-5.6 wrote for free.
What happens above 272,000 input tokens?+
Both models move to a long-context rate of $4 per million input and $15 output. Cached reads stay half as expensive on GPT-6.1 Sol: $0.20 against $0.40.