GPT-5.5 — Specs, Pricing & Limits
Codex signed in with ChatGPT lost GPT-5.5 on 2026-10-14; OpenAI names GPT-6.1 Sol as the replacement on paid plans. Codex with your own API key keeps it.
Specs
| Status | Priced |
|---|---|
| Context window | 1,050,000 tokens |
| Max output | 128,000 tokens |
| Knowledge cutoff | Dec 01, 2025 |
| Reasoning effort levels | none, low, medium (default), high, xhigh |
| Long-context rule | >272,000 input tokens: 2x input, 1.5x output (full request) |
| Cache writes | No additional charge — a written token bills as ordinary input |
| Codex with ChatGPT sign-in | Retired 2026-10-14 — replacement: GPT-6.1 Sol on Plus, Pro, Business, Enterprise and Edu; GPT-6 Luna in the desktop app on Free and Go. |
Pricing per 1M tokens, every published service tier
| Standard — short context | in $5.00 · cached $0.50 · writes — · out $30.00 |
|---|---|
| Standard — long context | in $10.00 · cached $1.00 · writes — · out $45.00 |
| Batch — short context | in $2.50 · cached $0.25 · writes — · out $15.00 |
| Batch — long context | in $5.00 · cached $0.50 · writes — · out $22.50 |
| Flex — short context | in $2.50 · cached $0.25 · writes — · out $15.00 |
| Flex — long context | in $5.00 · cached $0.50 · writes — · out $22.50 |
| Fast — short context | in $12.50 · cached $1.25 · writes — · out $75.00 |
Rate limits by usage tier
| Free | Not supported |
|---|---|
| Build | 5,000 RPM · 1,000,000 TPM · 3,000,000 batch queue |
| Launch | 10,000 RPM · 4,000,000 TPM · 200,000,000 batch queue |
| Grow | 15,000 RPM · 40,000,000 TPM · 15,000,000,000 batch queue |
Long-context rate limits (requests over 272,000 input tokens, in place of the table above)
| Free | Not supported |
|---|---|
| Build | 500 RPM · 1,000,000 TPM · 40,000,000 batch queue |
| Launch | 2,000 RPM · 10,000,000 TPM · 200,000,000 batch queue |
| Grow | 8,000 RPM · 20,000,000 TPM · 2,000,000,000 batch queue |
GPT-5.5 is no longer a Codex model for most people. On October 14, 2026, OpenAI retired it from ChatGPT, ChatGPT Work and Codex on every plan — consumer, Business, Enterprise and Edu — while leaving it on the API. If you use Codex signed in with ChatGPT, the replacement OpenAI names is GPT-6.1 Sol on paid plans (it named GPT-6 Sol when it announced the retirement) and GPT-6 Luna in the desktop app on Free and Go. If you use Codex with your own API key, nothing changed on that date.
On the API it keeps its old prices, and that is now its weakest point. It was priced like GPT-5.6 Sol when both launched; since Sol's promotional cut and the GPT-6 releases, several newer models with the same context window and the same rate-limit group cost less on both input and output. GPT-5.5's knowledge cutoff also trails every GPT-6 model. The remaining reason to call it is a workload whose behavior you've validated on this exact model and don't want to re-test yet.
Two specifics carry over from its generation. It charges nothing extra for cache writes — a written token bills as ordinary input, the rule for every model before GPT-5.6 — and its long-context rule is worded "for the full session for standard, batch, and flex," which matches the published long bands: there is no long-context row for its Fast tier. Its reasoning effort runs from none to xhigh, with no max. And its page publishes a separate rate-limit table for requests over 272K input tokens, shown above beside the Standard one.
The tables above are computed directly from this site's facts module: pricing across every published service tier, the long-context bands, and both rate-limit tables.
Verified 2026-10-28 against https://developers.openai.com/api/docs/models/gpt-5.5.
Checked: https://developers.openai.com/api/docs/models/gpt-5.5 · https://developers.openai.com/api/docs/pricing · https://learn.chatgpt.com/docs/models