Migrating from gpt-realtime to gpt-realtime-2.1
What changes when you move from gpt-realtime to gpt-realtime-2.1: input price 0%, output price +50%, context window 32,000 → 128,000 tokens.
- From
- gpt-realtime
- Status
- Deprecated
- Shutdown
- Jan 20, 2027
- Time left
- 97 days
- Status
- Active
- Shutdown
- none announced
- Time left
- —
Prices
USD per 1M tokens, standard tier| Type | gpt-realtime | gpt-realtime-2.1 | Change |
|---|---|---|---|
| Input | $4.00first seen October 1, 2026 · source | $4.00first seen October 1, 2026 · source | 0% |
| Cached input | $0.40first seen October 1, 2026 · source | $0.40first seen October 1, 2026 · source | 0% |
| Output | $16.00first seen October 1, 2026 · source | $24.00first seen October 1, 2026 · source | +50% |
At 1M input and 250K output tokens a month: $8.00 on gpt-realtime, $10.00 on gpt-realtime-2.1 (+25%).
Other tiers, prompt-size bands and modalities are on each model's page.
Limits
Tokens| Limit | gpt-realtime | gpt-realtime-2.1 | Change |
|---|---|---|---|
| Context window | 32,000 tokensin place when tracking began (October 7, 2026) · source | 128,000 tokensin place when tracking began (October 7, 2026) · source | +300% |
| Max output tokens | 4,096 tokensin place when tracking began (October 7, 2026) · source | 32,000 tokensin place when tracking began (October 7, 2026) · source | +681.3% |
Features and parameters
Compare modelsSame on both
- Function calling: supported on both.gpt-realtime: in place when tracking began (October 7, 2026) · source · gpt-realtime-2.1: in place when tracking began (October 7, 2026) · source
- Prompt caching: supported on both.gpt-realtime: in place when tracking began (October 7, 2026) · source · gpt-realtime-2.1: in place when tracking began (October 7, 2026) · source
- Batch API: not supported on both.gpt-realtime: in place when tracking began (October 7, 2026) · source · gpt-realtime-2.1: in place when tracking began (October 7, 2026) · source
- Fine-tuning: not supported on both.gpt-realtime: in place when tracking began (October 7, 2026) · source · gpt-realtime-2.1: in place when tracking began (October 7, 2026) · source
Not published for gpt-realtime: Structured outputs, Image input, PDF input, Reasoning / thinking, Built-in web search, temperature parameter, top_p parameter, logprobs.
Not published for gpt-realtime-2.1: Structured outputs, Image input, PDF input, Built-in web search, temperature parameter, top_p parameter, logprobs.
Aliases
All aliasesNo aliases point at either model.
On cloud platforms
All platform dates| Model | Platform | Platform ID | Shutdown | Time left |
|---|---|---|---|---|
| gpt-realtime | Microsoft Foundryversion 2025-08-28 | gpt-realtime | Mar 2, 2027Active | 138 days |
| gpt-realtime-2.1 | Microsoft Foundryversion 2026-07-07 | gpt-realtime-2.1 | Jun 25, 2027Active | 253 days |
Read daily from each platform's own pages. The status above uses each company's own date. Time left counts to 00:00 UTC on the shutdown date, which is the evening before in ET.
Checklist
- Change the model string from
gpt-realtimetogpt-realtime-2.1. - Re-check costs: gpt-realtime-2.1 costs more than gpt-realtime (output +50%).
Counting to 7:00 pm ET on January 19, 2027, which is 00:00 UTC on January 20, 2027.
Compared from AIWire's published data when this page loaded. Each fact links to the page it came from. Model pages: gpt-realtime, gpt-realtime-2.1.