GPT-5.5

OpenAI flagship with automatic prefix caching above 1,024 tokens. A strong all-rounder and the default for auto.best when latency is not the constraint.

$5.00
Input / 1M tokens
$30.00
Output / 1M tokens
$0.50
Cache read · −90%
1.1M
Context window

At a glance.

ProviderOpenAI
Familygpt-5
Released2026-04
LicenseProprietary
Context window1.1M tokens
Max output128K tokens
Modalitiestext, image, pdf
Tool callingYes
Reasoning modeYes
Cachingautomatic
Batch discount50% off

What reuse looks like here.

GPT-5.5 · agent trafficper request
Total input100%
Candidate reuse61%
Realized reuse54%
Capture rate88%
240ms
Warm TTFT · −67% vs cold
138
Output tokens / sec
Reuse economics

What you actually pay once caching works.

At a typical 55% prefix reuse, a million input tokens on GPT-5.5 effectively costs $2.52 instead of $5.00 - blending to roughly $9.39 with a 25% output share. Background work drops a further 50% on the batch tier.

Estimate it for your workload
Best for
generalcodingmultimodal

Routes through these aliases:

Same OpenAI client, this model.

python
from openai import OpenAI

client = OpenAI(base_url="https://api.zumik.ai/v1", api_key="zk_live_...")

r = client.responses.create(
    model="gpt-5-5",          # or an alias like auto.best
    input="Draft a fix for the failing test.",
)
print(r.usage.input_tokens_cached)   # confirm reuse

GPT-5.5, answered.

How much does GPT-5.5 cost?

GPT-5.5 is $5.00 per million input tokens and $30.00 per million output tokens through Zumik. Cache reads are $0.50 per million, a 90% discount on input.

What is GPT-5.5's context window?

GPT-5.5 supports a 1.1M-token context window with up to 128K output tokens.

Does GPT-5.5 support prompt caching?

Yes. OpenAI uses Automatic prefix caching caching. In the Zumik corpus, GPT-5.5 shows a median cache capture of 88% on agent workloads.

Which Zumik aliases route to GPT-5.5?

GPT-5.5 is a candidate for the auto.best, auto.balanced, code.balanced aliases, selected when it wins under current routing policy.

Run GPT-5.5 with reuse measured.

Point an OpenAI client at Zumik and see exactly how much of this model's input you are reusing.