guides/priority-processing.md +0 −83 deleted
File Deleted View Diff
1# Priority processing
2
3> For the complete documentation index, see [llms.txt](/llms.txt). Markdown versions of documentation pages are available by appending `.md` to the page URL.
4
5Priority processing delivers significantly lower and more consistent latency compared to Standard processing while keeping pay-as-you-go flexibility.
6
7Priority processing is ideal for high-value, user-facing applications with regular traffic where latency is paramount. Priority processing should not be used for data processing, evaluations, or other highly erratic traffic.
8
9## Configuring Priority processing
10
11Requests to the Responses or Completions endpoints can be configured to use Priority processing through either a request parameter, or a Project setting.
12
13To opt-in to Priority processing at the request level, include the [`service_tier=priority`](https://platform.openai.com/docs/api-reference/responses/create#responses-create-service_tier) parameter for Completions or Responses.
14
15Create a response with priority processing
16
17```bash
18curl https://api.openai.com/v1/responses \
19 -H "Authorization: Bearer $OPENAI_API_KEY" \
20 -H "Content-Type: application/json" \
21 -d '{
22 "model": "gpt-5.6",
23 "input": "What does 'fit check for my napalm era' mean?",
24 "service_tier": "priority"
25 }'
26```
27
28```javascript
29import OpenAI from "openai";
30
31const openai = new OpenAI();
32
33const response = await openai.responses.create({
34 model: "gpt-5.6",
35 input: "What does 'fit check for my napalm era' mean?",
36 service_tier: "priority",
37});
38
39console.log(response);
40```
41
42```python
43from openai import OpenAI
44
45client = OpenAI()
46
47response = client.responses.create(
48 model="gpt-5.6",
49 input="What does 'fit check for my napalm era' mean?",
50 service_tier="priority",
51)
52print(response)
53```
54
55
56To opt in at the Project level, navigate to the Settings page, select the General tab under Project, then change the Project Service Tier to Priority. Once configured on the project, requests that don't specify a `service_tier` will default to Priority. Note that requests for the project will be gradually transitioned to Priority over time.
57
58The `service_tier` field in the [Responses](https://platform.openai.com/docs/api-reference/responses/object#responses/object-service_tier) or [Completions](https://platform.openai.com/docs/api-reference/chat/object#chat/object-service_tier) response objects will contain which service tier was used to process the request.
59
60## Rate limits and ramp rate
61
62**Baseline limits**
63
64Priority consumption is treated like Standard for rate‑limit accounting. Use your usual retry and backoff logic. For a given model, the rate limit is shared between Standard and Priority processing.
65
66**Ramp rate limit**
67
68If your traffic ramps too quickly, some Priority requests may be downgraded to Standard and billed at Standard rates. If the ramp rate limit is exceeded, the response will show service_tier="default". Currently, the ramp rate limit may apply if you’re sending at least 1 million TPM and >50% TPM increase within 15 minutes.
69
70To avoid triggering the ramp rate limit, we recommend:
71
72- Ramp gradually when changing models or snapshots
73- Use feature flags to shift traffic over hours, not instantly.
74- Avoid large ETL or batch jobs on Priority
75
76## Usage considerations
77
78- Per token costs are billed at a premium to standard - see [pricing](https://developers.openai.com/api/docs/pricing) for more information.
79- Cache discounts are still applied for priority processing requests.
80- Priority processing applies for multimodal / image input requests as well.
81- Requests handled with priority processing can be viewed in the dashboard using the "group by service tier" option.
82- See the [pricing page](https://developers.openai.com/api/docs/pricing) for which models currently support Priority processing.
83- Long context, fine-tuned models and embeddings are not yet supported.