GPT-5.4 Nano (2026-03-17)
Mar 2026
GPT-5.4 Nano (2026-03-17)Compare inon
GPT-5.4 Nano (2026-03-17) measured on 9 of 109 Chat & Knowledge benchmarks9
BenchmarkByScoreRank
Knowledge2
Reasoning1
Math4
Legacy · saturated or no longer reported2
No benchmark matches
Alternatives?
ModelScoreCostPer taskWhat changes
CheaperClaude Haiku 5.5xhigh+13%-57%$0.12Remove sampling parameters+5BetterGPT-6 Lunamax+13%-57%$0.068Function calling only with none · Chat Completions+3
OpenDeepSeek V4.1 Flash 1017max+13%+14%$0.26Pass back reasoning_content · Chat Completions+4
CheaperGLM 5.3 FlashmaxOpen+7%-49%$0.28Defaults to max · Self-hosted+4CheaperGPT OSS 120BhighOpen+1%-87%$0.11Context: 400K → 131K+1ModelScoreCostPer taskWhat changes
Claude Haiku 5.5xhigh+13%-57%$0.12Remove sampling parameters+5GPT-6 Lunamax+13%-57%$0.068Function calling only with none · Chat Completions+3GLM 5.3 FlashmaxOpen+7%-49%$0.28Defaults to max · Self-hosted+4GPT OSS 120BhighOpen+1%-87%$0.11Context: 400K → 131K+1
Qwen3.7 Flashunstated-1%-88%$0.03 / $0.13ModelScoreCostPer taskWhat changes
GPT-6 Lunamax+13%-57%$0.068Function calling only with none · Chat Completions+3Claude Haiku 5.5high+11%-57%$0.079Remove sampling parameters+5GPT-5.6 Lunamax+12%same$0.18
Grok 4.3high+6%+238%$0.21GPT-5.6 Sollow+9%+1,630%$0.26ModelScoreCostPer taskWhat changes
DeepSeek V4.1 Flash 1017maxOpen+13%+14%$0.26Pass back reasoning_content · Chat Completions+4DeepSeek V4 Pro 0813maxOpen+12%+328%$0.67No image input+5
Kimi K3maxOpen+11%+1,197%$2.00Sampling parameters are fixed+3Kimi K2.6unstatedOpen+9%+270%$0.95 / $4Context: 400K → 256KKimi K2.7 CodeunstatedOpen+9%+270%$0.95 / $4Context: 400K → 256K+1ModelScoreCostPer taskWhat changes
GPT-6.1 Solmax+14%—$0.72Remove unsupported parameters+5GPT-6 Solmax+14%—$1.04GPT-6 Astramax+14%—$3.26Monitoring can return HTTP 403+2GPT-5.6 Solmax+14%—$1.99GPT-5.6 Terramax+14%—$1.40Prompts with >272K input tokensGPT-6 Lunamax+13%—$0.068Function calling only with none · Chat Completions+3GPT-5.6 Lunamax+12%—$0.18GPT-5.4 (2026-03-05)high+11%—$2.50 / $15All OpenAI models →History
5 versionsBenchmarks
in Chat & KnowledgeGPQA Diamond · Epoch78.5#72
OTIS Mock AIME 2024 202587.8#55
FrontierMath25.9#17
FrontierMath Tier 46.3#14
FrontierMath Tiers 1–3 (v2)44.9#50
FrontierMath Tier 4 (v2)12.2#48
SimpleQA Verified v111.7#76
Chess Puzzles36#36
FAQ
How much does GPT-5.4 Nano (2026-03-17) cost?›
$0.20 per 1M input tokens and $1.25 per 1M output tokens.
Is GPT-5.4 Nano (2026-03-17) free to use?›
There is no free API; the cheapest paid price is $0.20 per 1M input tokens.
How good is GPT-5.4 Nano (2026-03-17), and what is it best at?›
#76 of 84 in Chat & Knowledge.
Is GPT-5.4 Nano (2026-03-17) better than GPT-5.1 (2025-11-13)?›
On OTIS Mock AIME 2024 2025, GPT-5.4 Nano (2026-03-17) scores 87.8 vs 88.6 — about the same.
How can I use GPT-5.4 Nano (2026-03-17)?›
Through the OpenAI API.
What is the context window of GPT-5.4 Nano (2026-03-17)?›
400K tokens.
When was GPT-5.4 Nano (2026-03-17) released? Is there a newer version?›
Released 2026-03-17. It has been superseded by GPT-6 Luna.
Can GPT-5.4 Nano (2026-03-17) read images?›
It accepts text and images as input.
What is GPT-5.4 Nano (2026-03-17)?›
GPT-5.4 Nano (2026-03-17) is a language model from OpenAI, released Mar 2026.
What are the best alternatives to GPT-5.4 Nano (2026-03-17)?›
Claude Haiku 5.5 (cheaper), GPT-6 Luna (better), DeepSeek V4.1 Flash 1017 (open weights).