LLM Pricing
PricingLeaderboardsToolsProvidersReleasesGuides

© 2026 LLM Pricing

About
·Contact
·Privacy
·RSS
← Model list

GPT-5.4 nano

openai·openai/gpt-5.4-nano·GA·Closed·gpt-nano series
Cheapest GPT-5.4 lane for simple routing, extraction, and bulk automation

GPT-5.4 nano by openai is offered by 33 providers on this page. Public prices are shown for 32 of them. Its official list price is $0.20 per 1M input tokens and $1.25 per 1M output tokens.

Artificial Analysis rates it 39.7 on the Intelligence Index, with 56.1 for coding and 29.7 for agentic tasks. Against its lowest blended price of $0.41 per 1M, that is roughly 97 index points per dollar, which is the value ratio the leaderboards rank on. Median output speed is 177 tokens per second, with 2.9s to the first token. Running one task of the Artificial Analysis suite costs about $0.1491, which reflects how many tokens its reasoning consumes rather than the unit price alone.

The context window is 400,000 tokens at the reference host, but hosts report different limits, from 128,000 to 1,047,576, so the usable window depends on the provider you pick. It supports reasoning, tool use, and structured output. Accepted input modalities are Text, Image, and PDF. Providers report the capability flags inconsistently, so verify a specific feature against the host you plan to use. Its training knowledge cuts off at 2025-08-31.

Specs & pricing

Input / output per 1M tokens
Official price·OpenAI
$0.20 / $1.25
Blended $0.463 · Cache read $0.02
Lowest paid·PoeGateway
$0.18 / $1.10
Blended $0.41 · 1.1× spread
Context
400,000
Output limit
128,000
Knowledge cutoff
2025-08-31
Released / updated
2026-03-17 / 2026-03-17
Capabilities
✓ Reasoning✓ Tool use✓ Structured output✓ Temperature✓ Attachments
⚠ Providers report capability flags inconsistently
Modalities
TextImagePDF

Quality & performanceArtificial Analysis · Intelligence Index v4.1 · rep. tier xhigh

Intelligence39.7
Value97
Coding56.1
Agentic29.7
Output speed177 tok/s
TTFT2.9 s
Cost per task$0.1491
Value formulaIQ 39.7 ÷ min blended $0.410 = 97
Reasoning tier → intelligence / speed (higher tier = stronger but slower)
Non-ReasoningIQ 17.8 · 185 tok/s
mediumIQ 30.8 · 185 tok/s
xhighIQ 39.7 · 177 tok/s

Quality is independently evaluated by Artificial Analysis. Speed/latency are model-level medians.

Available at 33 providers32 with public prices

ProviderTierInputOutputCache readCache writeContextOutput limitStatus
PoeGateway$0.18$1.10$0.018—400,000128,000
NanoGPTGateway$0.20$1.25$0.02—400,000128,000
VivgridGateway$0.20$1.25$0.02—400,000128,000
ImpossiblGateway$0.20$1.25$0.02—400,000128,000
OrcaRouterGateway$0.20$1.25$0.02—400,000128,000
Azure Cognitive ServicesFirst-party$0.20$1.25$0.02—400,000128,000
PioneerGateway$0.20$1.25$0.02$0.201,047,576 ⚠128,000
RequestyGateway$0.20$1.25$0.02—400,000128,000
CrossModelGateway$0.20$1.25$0.02$0.20400,000128,000
FrogBot
gpt-5-4-nano
Gateway$0.20$1.25$0.02—400,000128,000
OpenAIOfficialFirst-party$0.20$1.25$0.02—400,000128,000

Sorted by blended price (input×0.75 + output×0.25) asc. The official channel always shows regardless of rank. Whether a gateway's low price is actually usable can't be verified.

Your usage cost

1Poe$71.56
2NanoGPT$80.90
3Vivgrid$80.90
4Impossibl$80.90
5OrcaRouter$80.90
6Azure Cognitive Services$80.90
26OpenAI · Official$80.90
Switch to Poe to save $9.34/mo (12%). The gap is small, so staying on the official channel is fine.
Note: this is a gateway; verify availability and rate limits yourself.

Benchmark16 items

NameConditionsScoreMetricSource
SWE-Bench Provariant: reasoning effort xhigh52.4resolve rateSource ↗
Terminal-Benchvariant: reasoning effort xhigh · v2.046.3accuracySource ↗
MCP Atlasvariant: reasoning effort xhigh56.1scoreSource ↗
Toolathlonvariant: reasoning effort xhigh35.5scoreSource ↗
τ²-Bench Telecomvariant: reasoning effort xhigh92.5accuracySource ↗
GPQA Diamondvariant: reasoning effort xhigh82.8accuracySource ↗
Humanity's Last Examvariant: with tools37.7accuracySource ↗
Humanity's Last Examvariant: without tools24.3accuracySource ↗
OSWorld-Verifiedvariant: reasoning effort xhigh39success rateSource ↗
MMMU Provariant: with Python69.5accuracySource ↗
MMMU Provariant: without tools66.1accuracySource ↗
OmniDocBenchvariant: reasoning effort none · v1.50.2419overall edit distanceSource ↗
OpenAI MRCRvariant: 8-needle, 64K-128K · vv244.2accuracySource ↗
OpenAI MRCRvariant: 8-needle, 128K-256K · vv233.1accuracySource ↗
Graphwalksvariant: BFS, 0-128K73.4accuracySource ↗
Graphwalksvariant: parents, 0-128K50.8accuracySource ↗

The same benchmark scores very differently across harness / dataset, so the qualifying conditions must be shown together.

Reasoning control

effort = noneeffort = loweffort = mediumeffort = higheffort = xhigheffort = maxbudget_tokenseffort = minimal

2 / 33 providers expose no reasoning control (reasoning_options: []).

Related models

Agentic Chat (GPT-5.4 Nano)same series$0 / $0OpenAI GPT-5.4 Nanosame series$0.20 / $1.25GPT-5 Nanosame series$0.05 / $0.40DeepSeek V4 Flashcheaper alternative$0.14 / $0.28GLM-5.3-Flashcheaper alternative$0.075 / $0.25MiMo-V2.5cheaper alternative$0.14 / $0.28

Price historyone sample accumulated per data sync

Input listOutput listMin blended
$1.25$02026-08-052026-08-132026-08-05 · Input list $0.202026-08-06 · Input list $0.202026-08-07 · Input list $0.202026-08-08 · Input list $0.202026-08-09 · Input list $0.202026-08-10 · Input list $0.202026-08-11 · Input list $0.202026-08-12 · Input list $0.202026-08-13 · Input list $0.202026-08-05 · Output list $1.252026-08-06 · Output list $1.252026-08-07 · Output list $1.252026-08-08 · Output list $1.252026-08-09 · Output list $1.252026-08-10 · Output list $1.252026-08-11 · Output list $1.252026-08-12 · Output list $1.252026-08-13 · Output list $1.252026-08-05 · Min blended $0.412026-08-06 · Min blended $0.412026-08-07 · Min blended $0.412026-08-08 · Min blended $0.412026-08-09 · Min blended $0.412026-08-10 · Min blended $0.412026-08-11 · Min blended $0.412026-08-12 · Min blended $0.412026-08-13 · Min blended $0.41
Data partly from models.dev (MIT) · OpenAI official docs ↗