modelbenchmark.io

GPT-5 Nano

OpenAI · gpt-5-nano

Compare

Tiny GPT-5 lane for routing, extraction, classification, and bulk jobs

Specification

openai listing

Context
400K
Max output
128K
Released
2025-08-07
Knowledge cutoff
2024-05-30
Retires
2026-12-11
Open weights
no
Input
text, image
Output
text

Price

US dollars per million tokens · openai list price

Input
$0.05
Output
$0.40
Cache read
$0.005
Cache write
$0.05

Batch and priority prices

US dollars per million tokens · per listing

PriceListing$/M
Batch inputazure$0.0275
Batch inputazure$0.025
Batch outputazure$0.22
Batch outputazure$0.20

Batch input is 45% below the interactive input price at the same listing.

Quality

11 benchmarks · 2 sources

Composite
33rd
percentile of 344 scored models
Rank
232 / 344
±0.05 sd
Evidence
11 × 2
3 or more benchmarks, from 2 or more sources
Effort range
9.5
points between effort settings
coding19th1/3 bench
reasoning37th3/4 bench
math42nd6/7 bench
knowledge5th1/1 bench

By reasoning effort

every setting placed on the same scale as the leaderboard

SettingCompositeBenchmarks behind it
medium50th7
high41st10
low20th5

Every score

one row per source and configuration — nothing averaged away

BenchmarkScoreConfigurationSourceRun
MATH level 595.2 ±0.5mediumEpoch AI2025-08-20
↳94.9 ±0.5highEpoch AI2025-08-20
Settings differ by 0.3 points on this benchmark.
OTIS Mock AIME 2024-202581.1 ±4.6highEpoch AI2025-10-31
↳74.2 ±5.2mediumEpoch AI2025-08-07
↳41.1 ±7.5low · 2 runsEpoch AI2026-07-13
Settings differ by 40.0 points on this benchmark.
GPQA diamond69.4 ±2.8highEpoch AI2025-10-30
↳67.4 ±2.7mediumEpoch AI2025-08-07
↳53.0 ±3.5low · 2 runsEpoch AI2026-07-13
Settings differ by 16.4 points on this benchmark.
FrontierMath-v114.1 ±13.3high · 2 runsEpoch AI2025-10-30
↳8.6 ±10.0medium · 2 runsEpoch AI2025-08-07
Settings differ by 5.5 points on this benchmark.
FrontierMath-Tier-4-2025-07-011.0 ±2.1medium · 2 runsEpoch AI2025-08-07
↳0.0 ±0.0high · 2 runsEpoch AI2025-10-30
Settings differ by 1.0 points on this benchmark.
Chess Puzzles22.0 ±4.2highEpoch AI2026-10-13
↳8.5 ±3.6low · 2 runsEpoch AI2026-10-13
Settings differ by 13.5 points on this benchmark.
Mystery Game Puzzles9.0 ±2.9mediumEpoch AI2026-08-27
↳8.0 ±2.7highEpoch AI2026-08-27
↳7.0 ±2.9low · 2 runsEpoch AI2026-08-27
Settings differ by 2.0 points on this benchmark.
SWE-Bench verified34.8medium · mini-SWE-agentSWE-bench—
FrontierMath-Tier-4-v22.4 ±2.4highEpoch AI2026-06-12
FrontierMath-Tiers-1-3-v220.0 ±2.4highEpoch AI2026-06-12
↳3.9 ±0.8low · 2 runsEpoch AI2026-08-27
Settings differ by 16.1 points on this benchmark.
SimpleQA Verified11.7 ±1.0highEpoch AI2026-08-10

The leading figure for a benchmark is the median across its configurations, so one heroic high-effort run cannot set the number.

Available from 32 hosts

HostIn $/MOut $/MCache rdCache wrContextOutputRetires
openaivendor$0.05$0.40$0.005—
400K
272K
128K2026-12-11
poe$0.0455$0.3636$0.0045—400K128K—
jiekou$0.0475$0.38$0.0047—400K128K—
abacus$0.05$0.40$0.005—400K128K—
aimlapi$0.05$0.40$0.005—400K128K—
azure
$0.05
$0.055
$0.40
$0.44
$0.005
$0.0055
—
400K
272K
128K2027-02-09
azure-cognitive-services$0.05$0.40$0.005—400K128K—
cloudflare-ai-gateway$0.05$0.40$0.005—128K128K—
databricks$0.05$0.40$0.005$0.05
400K
272K
128K—
edenai$0.05$0.40$0.005—400K128K—
fastrouter$0.05$0.40$0.005—400K128K—
helicone$0.05$0.40$0.005—400K128K—
impossibl$0.05$0.40$0.005—400K128K—
kilo$0.05$0.40$0.005—400K128K—
llmgateway$0.05$0.40$0.005—400K128K—
llmgateway-providers$0.05$0.40$0.005—400K128K—
lowrouter · auto$0.05$0.40$0.005—400K128K—
merge-gateway$0.05$0.40$0.005—400K128K—
nano-gpt$0.05$0.40$0.005—400K128K—
nearai$0.05$0.40$0.005—400K128K—
neon$0.05$0.40$0.01—400K128K—
oci · oci$0.05$0.40——272K128K—
ofox$0.05$0.40$0.01—400K128K—
opencode$0.05$0.40$0.005—400K128K—
openrouter$0.05$0.40$0.005—400K128K—
orcarouter$0.05$0.40$0.005—400K128K—
perplexity$0.05$0.40$0.005————
prime$0.05$0.40——400K128K—
replicate · replicate$0.05$0.40—————
sap-ai-core$0.05$0.40$0.005—400K128K—
vercel$0.05$0.40$0.005—400K128K—
cortecs$0.06$0.439$0.019—400K128K—

Price history

append-only observations · a listing writes a row only when its price moves

DateHostField$/M
2026-09-02azureCache read$0.01
2026-10-20azureCache read$0.005
2026-09-02azure-cognitive-servicesCache read$0.01
2026-10-20azure-cognitive-servicesCache read$0.005
2026-09-02neonCache read$0.005
2026-10-24neonCache read$0.01
2026-09-02ofoxCache read$0.01
2026-09-19ofoxCache read$0.008
2026-10-17ofoxCache read$0.01
2026-09-02openrouterCache read$0.005
2026-10-10openrouterCache read$0.005
2026-09-02jiekouInput$0.045
2026-10-22jiekouInput$0.0475
2026-09-02ofoxInput$0.05
2026-09-19ofoxInput$0.04
2026-10-17ofoxInput$0.05
2026-09-02openrouterInput$0.05
2026-10-10openrouterInput$0.05
2026-09-02poeInput$0.045
2026-10-22poeInput$0.0455
2026-09-02jiekouOutput$0.36
2026-10-22jiekouOutput$0.38
2026-09-02ofoxOutput$0.40
2026-09-19ofoxOutput$0.32
2026-10-17ofoxOutput$0.40
2026-09-02openrouterOutput$0.40
2026-10-10openrouterOutput$0.40
2026-09-02poeOutput$0.36
2026-10-22poeOutput$0.3636
2026-10-13aimlapiCache read$0.0688
2026-10-17aimlapiCache read$0.0065
2026-10-23aimlapiCache read$0.005
2026-10-13aimlapiInput$0.0756
2026-10-17aimlapiInput$0.065
2026-10-23aimlapiInput$0.05
2026-10-13aimlapiOutput$0.6052
2026-10-17aimlapiOutput$0.52
2026-10-23aimlapiOutput$0.40

Other listings of this model

same model, different key

More from OpenAI

most-hosted first