Skip to main content
gpt-5-4-nano is the smallest tier of the GPT-5.4 family — the cheapest way to get GPT-flavored reasoning on the gateway. The right pick for very-high-volume traffic (classification, routing, short structured extraction) where unit cost dominates and the task fits a small prompt. Pricing: 0.20/1Minput,0.20 / 1M input, 1.25 / 1M output, $0.02 / 1M cache read — see the rate card.

​
Protocols

ProtocolPath
OpenAI Chat CompletionsPOST https://llm.bytespike.ai/v1/chat/completions
OpenAI ResponsesPOST https://llm.bytespike.ai/v1/responses

​
Quickstart

curl https://llm.bytespike.ai/v1/chat/completions \
  -H "Authorization: Bearer $BYTESPIKE_API_KEY" \
  -H "content-type: application/json" \
  -d '{
    "model": "gpt-5-4-nano",
    "reasoning_effort": "low",
    "messages": [
      { "role": "user", "content": "Classify: refund / billing / technical / other.\n\nMy invoice charged twice." }
    ]
  }'

​
Capabilities

CapabilitySupported
Chat Completions✅
Responses API✅
Streaming (SSE)✅
Vision✅
Tool use✅ parallel
JSON mode✅
Structured outputs✅
Reasoning effort✅ (low / medium)
Web search—
Context window128K tokens

​
When to use

  • Very-high-volume classification, routing, structured extraction — the cheapest GPT tier when unit cost dominates.
  • Short-prompt agent loops — tool-use chains where each step is small and the gate is throughput.
  • Cost-floor fallback — when the task fits nano, prefer it over gpt-5-4-mini for the savings.
When not to use:
  • Hard reasoning that benefits from reasoning_effort: "high" — go to gpt-5-4 or gpt-5-5.
  • Web search needed — only gpt-5-2 and up have it.

​
Next