GPT-5 → GPT-5.6 Sol: cost and quality on real tasks
OpenAI shuts down gpt-5 on 2026-12-11 and names GPT-5.6 Sol as its replacement (OpenAI's deprecations page). After that date, requests to GPT-5 fail.
On the six tasks below, GPT-5.6 Sol costs 2.1–3.1 times as much per request at list prices. So the forced switch raises your real cost per result unless GPT-5.6 Sol works first time more often than GPT-5 did.
GPT-5.6 Sol's price is promotional until 2026-11-30, before GPT-5 shuts down, and OpenAI's pricing page doesn't say what it costs after that. The tables use the promotional price.
Price per million tokens
Standard list prices, checked 2026-11-12. Prices change often, so check the source before you rely on them.
| Model | Input | Cached input | Output | Source |
|---|---|---|---|---|
| GPT-5 (current)openai/gpt-5 | $1.25 | $0.125 | $10.00 | OpenAI's pricing page |
| GPT-5.6 Sol (new)openai/gpt-5.6-sol | $4.00 | $0.40 | $20.00 | OpenAI's pricing page |
- Note on GPT-5: API shutdown 2026-12-11; replacement gpt-5.6-sol. Reasoning model: thinking tokens are billed as output.
- Note on GPT-5.6 Sol: Promotional pricing until 2026-11-30 per OpenAI; the page does not state the post-promo price. Long context (>272K input tokens): $8 in / $0.80 cached / $30 out.
Cost per 1,000 results on six common tasks
What 1,000 requests cost at list price, for six task shapes from different kinds of app. The token counts are illustrative, not measured: your prompts may be longer or shorter, and the switch test below shows your real ones.
| Task, and what goes in and out | Tokens in / out | GPT-5 | GPT-5.6 Sol | Change | Break-even: GPT-5.6 Sol costs no more per result that works if it works first time at least |
|---|---|---|---|---|---|
| Support bot answerInstructions, three help-centre passages and the customer's question in; a short answer out. | 3,000 / 300 | $6.75 | $18.00 | 2.7 times as much | 2.7 times as often as GPT-5 |
| Invoice field extractionInstructions and one invoice's text in; about 15 fields as JSON out. | 2,500 / 250 | $5.63 | $15.00 | 2.7 times as much | 2.7 times as often as GPT-5 |
| Lead scoringScoring rules and one lead's form answers in; a score and a one-line reason out. | 1,200 / 100 | $2.50 | $6.80 | 2.7 times as much | 2.8 times as often as GPT-5 |
| Email reply draftInstructions and the email thread so far in; a reply of about 300 words out. | 1,500 / 400 | $5.88 | $14.00 | 2.4 times as much | 2.4 times as often as GPT-5 |
| TranslationInstructions and about 500 words of text in; the same text in another language out. | 900 / 800 | $9.13 | $19.60 | 2.1 times as much | 2.2 times as often as GPT-5 |
| Search re-rankingA search query and 20 result snippets in; the ids of the best five, in order, out. | 4,000 / 50 | $5.50 | $17.00 | 3.1 times as much | 3.1 times as often as GPT-5 |
GPT-5 bills thinking tokens as output. These task shapes leave them out, so real costs per request are higher than shown.
Real cost per result: when a switch really saves
Real cost per result is what you actually pay for each result that worked, counting the failed tries. A result that fails gets tried again, and you pay for both tries.
So the price per request is only half the answer. GPT-5.6 Sol only costs less per result that works if it works first time at least 2.2 times as often as GPT-5 on translation, and 3.1 times as often on search re-ranking. That's the new cost divided by the current cost, shown for each task in the last column above.
GPT-5.6 Sol costs more per request on every task here. If GPT-5 already works first time more than 46% of the time, GPT-5.6 Sol can't cost less per result that works, however well it does.
Example, with a made-up rate
Say 8 in 10 of GPT-5's support bot answers work first time. You pay $6.75 for 1,000 tries and get 800 that work, a real cost of $8.44 per 1,000 support bot answers that work. GPT-5.6 Sol costs $18.00 for 1,000 tries, so it would need 214% of its support bot answers to work first time to match. That's impossible, so here it costs more per result that works even if every try works.
Quality: test it on your own requests
This page can't tell you whether GPT-5.6 Sol handles your requests as well as GPT-5, and we don't publish quality scores. Quality depends on your prompts and your inputs: a model that answers short English questions well can still fail on long invoices, odd layouts or other languages.
The free model switch test runs your prompt and up to 20 of your real inputs on both models with your own key, through each provider's own API or OpenRouter. You see the answers side by side with the tokens and cost of each, and mark which ones worked.
Test both models on your promptsWhat to compare
- Format: does JSON still parse with the same field names, for extraction and scoring?
- Length and tone: are support answers and email drafts as short and polite as before?
- Refusals: does GPT-5.6 Sol decline requests that GPT-5 handled?
- Hard inputs: long documents, other languages and messy formatting, not just the easy cases.
Questions
Is GPT-5.6 Sol cheaper than GPT-5?
No: per request it costs 2.1–3.1 times as much on six common tasks. GPT-5 lists at $1.25 per million input tokens and $10.00 per million output tokens; GPT-5.6 Sol lists at $4.00 and $20.00 (GPT-5: price checked 12 nov 2026; unchanged since we started tracking it; GPT-5.6 Sol: price checked 12 nov 2026; unchanged since we started tracking it). Per result that works, it depends on how often each model works first time on your inputs: GPT-5.6 Sol only costs less per result that works if it works first time at least 2.2 times as often as GPT-5 on translation, and 3.1 times as often on search re-ranking.
When does GPT-5 stop working?
OpenAI shuts down gpt-5 on 2026-12-11, according to its deprecations page. Requests that name it will fail after that, so test the replacement on your real requests and switch before then.
Will GPT-5.6 Sol work as well as GPT-5 for my app?
Nobody can tell you that from a price table or a general benchmark, because it depends on your prompts and inputs. Run 10 to 20 of your real requests through both models with the free model switch test, using your own provider key, and compare the answers side by side before you switch any users.