GPT-5.4 mini (xhigh)

By OpenAI · released 2026-03-17 (5 months) · snapshot 2026-08-17

Off the frontierNot on frontier at launch dayproprietary
Intelligence index
40.90
Cost to run the AA index
$1096
Price per 1M tokens (in / out)
$0.75 / $4.50
End-to-end time
8.86s
Better than…

Is GPT-5.4 mini (xhigh) on the Pareto frontier, and what beats it?

Off the frontier

No. Grok 4.6 (high) is at least as smart for the same money or less.

Next step up: Claude Opus 5 (Adaptive Reasoning, High Effort), smarter but dearer.

GPT-5.4 mini (xhigh) vs Grok 4.6 (high)GPT-5.4 mini (xhigh) vs Claude Opus 5 (Adaptive Reasoning, High Effort)
#ModelCreatorIntelligenceTotal costE2E totalOpenReleasedFrontier
1GPT-5.4 mini (xhigh)OpenAI40.90$10968.86s2026-03-17
2Grok 4.6 (high)SpaceXAI60.90$106845.0s2026-08-12Frontier
3Claude Opus 5 (Adaptive Reasoning, High Effort)Anthropic61.50$197424.9s2026-07-24Frontier

Open in the picker Daily check URL: 200 on the frontier, 409 once beaten

At a glance

GPT-5.4 mini (xhigh) today

Time to first answer6.26s
Output speed192 tok/s
Input modalitiesNot reported
0204060$10$20$50$100$200$500$1k$2k$5kCost to run the Artificial Analysis Intelligence Index (USD, log scale)Intelligence IndexCeleris-1 (Celeris): intelligence 12.40, $157Claude 4.5 Haiku (Reasoning) (Anthropic): intelligence 29.90, $471Claude 4.5 Sonnet (Reasoning) (Anthropic): intelligence 37.40, $1037Claude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback) (Anthropic): intelligence 62.10, $5455Claude Opus 4.7 (Adaptive Reasoning, Max Effort) (Anthropic): intelligence 55.00, $3743Claude Opus 4.8 (Adaptive Reasoning, Max Effort) (Anthropic): intelligence 57.30, $3752Claude Opus 5 (Adaptive Reasoning, Low Effort) (Anthropic): intelligence 52.50, $555Claude Opus 5 (Adaptive Reasoning, Medium Effort) (Anthropic): intelligence 58.60, $1116Claude Sonnet 4.6 (Adaptive Reasoning, Max Effort) (Anthropic): intelligence 48.40, $3355Claude Sonnet 5 (Adaptive Reasoning, Max Effort) (Anthropic): intelligence 55.30, $4011Claude Sonnet 5 (Non-reasoning, High Effort) (Anthropic): intelligence 42.60, $416DeepSeek R1 (Jan '25) (DeepSeek): intelligence 18.60, $403DeepSeek V3 (Dec '24) (DeepSeek): intelligence 14.20, $20.0DeepSeek V3 0324 (DeepSeek): intelligence 15.20, $35.2DeepSeek V4 Flash (Reasoning, High Effort) (DeepSeek): intelligence 39.00, $58.5DeepSeek V4 Flash (Reasoning, Max Effort) (DeepSeek): intelligence 42.10, $104DeepSeek V4 Flash 0731 (Reasoning, Max Effort) (DeepSeek): intelligence 51.80, $323DeepSeek V4 Pro (Reasoning, High Effort) (DeepSeek): intelligence 43.70, $109DeepSeek V4 Pro (Reasoning, Max Effort) (DeepSeek): intelligence 45.30, $176DeepSeek V4 Pro 0813 (Reasoning, Max Effort) (DeepSeek): intelligence 53.20, $605Gemini 2.5 Pro (Google): intelligence 25.90, $611Gemini 3.1 Flash-Lite (Google): intelligence 25.60, $93.8Gemini 3.1 Pro Preview (Google): intelligence 47.70, $812Gemini 3.5 Flash (high) (Google): intelligence 52.00, $1042Gemini 3.5 Flash-Lite (Google): intelligence 37.40, $153Gemini 3.6 Flash (high) (Google): intelligence 51.60, $727Gemma 4 26B A4B (Reasoning) (Google): intelligence 26.10, $59.5Gemma 4 31B (Non-reasoning) (Google): intelligence 22.30, $32.8GLM-4.6 (Reasoning) (Z AI): intelligence 29.30, $314GLM-4.7 (Reasoning) (Z AI): intelligence 34.50, $607GLM-5.1 (Reasoning) (Z AI): intelligence 41.00, $676GLM-5.2 (max) (Z AI): intelligence 52.60, $735GPT-4.1 mini (OpenAI): intelligence 14.80, $64.2GPT-4.1 nano (OpenAI): intelligence 9.60, $36.7GPT-5 (high) (OpenAI): intelligence 35.30, $789GPT-5 mini (high) (OpenAI): intelligence 25.80, $130GPT-5.1 (high) (OpenAI): intelligence 37.50, $782GPT-5.4 (xhigh) (OpenAI): intelligence 53.10, $2183GPT-5.4 nano (xhigh) (OpenAI): intelligence 39.70, $277GPT-5.5 (high) (OpenAI): intelligence 54.70, $1721GPT-5.5 (low) (OpenAI): intelligence 44.50, $374GPT-5.5 (medium) (OpenAI): intelligence 51.40, $939GPT-5.5 (Non-reasoning) (OpenAI): intelligence 35.80, $212GPT-5.5 (xhigh) (OpenAI): intelligence 56.30, $2778GPT-5.5 Instant (June 2026) (OpenAI): intelligence 29.20, $423GPT-5.6 Sol (low) (OpenAI): intelligence 50.70, $344GPT-5.6 Sol (max) (OpenAI): intelligence 60.90, $2823GPT-5.6 Sol (medium) (OpenAI): intelligence 55.60, $580GPT-5.6 Sol (Non-reasoning) (OpenAI): intelligence 41.90, $240GPT-5.6 Sol (xhigh) (OpenAI): intelligence 59.00, $1525GPT-5.6 Terra (high) (OpenAI): intelligence 50.10, $395GPT-5.6 Terra (low) (OpenAI): intelligence 41.30, $130GPT-5.6 Terra (max) (OpenAI): intelligence 56.60, $1390GPT-5.6 Terra (medium) (OpenAI): intelligence 46.80, $192GPT-5.6 Terra (Non-reasoning) (OpenAI): intelligence 34.60, $98.5GPT-5.6 Terra (xhigh) (OpenAI): intelligence 52.80, $590gpt-oss-120b (high) (OpenAI): intelligence 24.10, $95.0gpt-oss-120b (low) (OpenAI): intelligence 14.90, $24.2gpt-oss-20b (high) (OpenAI): intelligence 15.20, $32.7Grok 4.3 (high) (SpaceXAI): intelligence 37.90, $288Grok 4.3 (Non-reasoning) (SpaceXAI): intelligence 25.00, $201Grok 4.5 (high) (SpaceXAI): intelligence 55.80, $579Grok Build 0.1 0616 (SpaceXAI): intelligence 40.70, $377Hy3 (Tencent): intelligence 42.20, $93.6Inkling (xhigh) (Thinking Machines): intelligence 42.30, $698Inkling Small (Thinking Machines): intelligence 41.20, $193Kimi K2.5 (Reasoning) (Kimi): intelligence 36.00, $304Kimi K2.6 (Kimi): intelligence 45.10, $841Kimi K2.7 Code (Kimi): intelligence 43.00, $545Kimi K3 (low) (Kimi): intelligence 48.30, $283Kimi K3 (max) (Kimi): intelligence 59.70, $2425Ling 3.0 Flash (InclusionAI): intelligence 37.80, $72.8Llama 4 Maverick (Meta): intelligence 14.50, $35.9Llama 4 Scout (Meta): intelligence 10.30, $11.6LongCat 2.0 (LongCat): intelligence 34.00, $442Magistral Medium 1.2 (Mistral): intelligence 18.00, $920Magistral Small 1.2 (Mistral): intelligence 11.50, $301Mercury 2 (Inception): intelligence 21.90, $97.5MiMo-V2.5 (Xiaomi): intelligence 38.00, $25.2MiMo-V2.5-Pro (Xiaomi): intelligence 42.90, $98.9MiniMax-M2.7 (MiniMax): intelligence 38.90, $143MiniMax-M3 (MiniMax): intelligence 45.40, $205Ministral 3 14B (Mistral): intelligence 11.20, $119Ministral 3 3B (Mistral): intelligence 7.10, $91.6Ministral 3 8B (Mistral): intelligence 9.00, $114Mistral Large 3 (Mistral): intelligence 15.90, $71.1Mistral Medium 3.1 (Mistral): intelligence 14.70, $134Mistral Medium 3.5 (Mistral): intelligence 30.40, $883Mistral Small 3.1 (Mistral): intelligence 14.90, $29.8Mistral Small 3.2 (Mistral): intelligence 10.70, $129Mistral Small 4 (Reasoning) (Mistral): intelligence 19.70, $87.7Muse Glimmer (high) (Meta): intelligence 35.10, $105Muse Spark 1.1 (xhigh) (Meta): intelligence 53.20, $547Nemotron 3 Super 120B A12B (Reasoning) (NVIDIA): intelligence 25.70, $254Nemotron 3 Ultra 550B A55B (Reasoning) (NVIDIA): intelligence 38.30, $534Nemotron 3.5 Lightning (NVIDIA): intelligence 23.60, $59.7NVIDIA Nemotron 3 Nano 30B A3B (Reasoning) (NVIDIA): intelligence 14.50, $39.4Qwen3 235B A22B 2507 (Reasoning) (Alibaba): intelligence 19.90, $610Qwen3 30B A3B 2507 (Reasoning) (Alibaba): intelligence 14.60, $152Qwen3 Coder Next (Alibaba): intelligence 21.30, $250Qwen3 Next 80B A3B (Reasoning) (Alibaba): intelligence 16.90, $400Qwen3.5 122B A10B (Non-reasoning) (Alibaba): intelligence 28.20, $221Qwen3.5 122B A10B (Reasoning) (Alibaba): intelligence 32.80, $444Qwen3.5 35B A3B (Non-reasoning) (Alibaba): intelligence 24.30, $228Qwen3.5 397B A17B (Reasoning) (Alibaba): intelligence 34.30, $527Qwen3.5 9B (Reasoning) (Alibaba): intelligence 21.80, $240Qwen3.6 27B (Non-reasoning) (Alibaba): intelligence 31.30, $359Qwen3.6 27B (Reasoning) (Alibaba): intelligence 37.70, $667Qwen3.6 35B A3B (Non-reasoning) (Alibaba): intelligence 24.60, $530Qwen3.6 35B A3B (Reasoning) (Alibaba): intelligence 32.10, $505Qwen3.6 Plus (Alibaba): intelligence 40.50, $526Qwen3.7 Max (Alibaba): intelligence 46.70, $1064Qwen3.7 Plus (Alibaba): intelligence 39.40, $322Qwen3.8 2.4T A95B (Alibaba): intelligence 57.70, $1672Qwen3.8 Max (Alibaba): intelligence 58.10, $1741Ring-2.6-1T (InclusionAI): intelligence 31.70, $455Solar Pro 3 (Upstage): intelligence 14.50, $158Solar Pro 4 (Upstage): intelligence 41.60, $321Step 3.7 Flash (StepFun): intelligence 30.90, $320Trinity Large Thinking (Arcee AI): intelligence 18.70, $255GPT-5.6 Luna (Non-reasoning) (OpenAI): intelligence 26.80, $10.2GPT-5.6 Luna (low) (OpenAI): intelligence 33.90, $13.7GPT-5.6 Luna (medium) (OpenAI): intelligence 38.90, $21.1GPT-5.6 Luna (high) (OpenAI): intelligence 47.00, $55.0GPT-5.6 Luna (xhigh) (OpenAI): intelligence 50.10, $95.1Gemini 3.7 Flash (low) (Google): intelligence 50.90, $165GPT-5.6 Luna (max) (OpenAI): intelligence 52.30, $172Gemini 3.7 Flash (medium) (Google): intelligence 53.40, $278Gemini 3.7 Flash (high) (Google): intelligence 56.00, $485Muse Spark 1.2 (xhigh) (Meta): intelligence 56.80, $639GPT-5.6 Sol (high) (OpenAI): intelligence 57.30, $955Grok 4.6 (high) (SpaceXAI): intelligence 60.90, $1068Claude Opus 5 (Adaptive Reasoning, High Effort) (Anthropic): intelligence 61.50, $1974Claude Opus 5 (Adaptive Reasoning, Xhigh Effort) (Anthropic): intelligence 62.50, $2909Claude Opus 5 (Adaptive Reasoning, Max Effort) (Anthropic): intelligence 63.10, $3836GPT-5.4 mini (xhigh) (OpenAI): intelligence 40.90, $1096GPT-5.4 mini (xhigh)GPT-5.6 Luna (Non-reasoning)Grok 4.6 (high)Claude Opus 5 (high)Claude Opus 5 (max)

1 of 3 effort levels plotted. The remaining levels lack intelligence or benchmark cost data.

GPT-5.4 mini (xhigh)Same release, other effort levelsToday's Pareto frontierEvery other model
Up and left is better. Dotted line: today's Pareto frontier. Hover a point for its numbers, click it for its page.

Intelligence: Artificial Analysis Intelligence Index. Cost: total cost to run that benchmark, so reasoning tokens count. API id: gpt-5-4-mini.

Matchup

Compare head to head

Suggested: what beats it, its effort levels, and frontier models near its price.

Compare GPT-5.4 mini (xhigh) with measured models

Release

What the GPT-5.4 mini release did to the frontier on 2026-03-17

None of its 3 effort levels reached the frontier at launch.

Closest miss: GPT-5.4 mini (xhigh), 6.8 points below Gemini 3.1 Pro Preview at 1.3× the cost.

Release page: launch-day frontier and every effort level

Deciding

Should you use GPT-5.4 mini (xhigh)?

Should you use GPT-5.4 mini (xhigh)?

Only when it fits your setup: Grok 4.6 (high) is at least as smart for the same money or less, so GPT-5.4 mini (xhigh) is off the value frontier today.

What is better and cheaper than GPT-5.4 mini (xhigh)?

Grok 4.6 (high) — at least as smart for the same money or less.

How much does GPT-5.4 mini (xhigh) cost?

$0.75/$4.50 in/out per 1M tokens, and $1096 to run the full Artificial Analysis Intelligence Index, the honest all-in number for reasoning models.

Is GPT-5.4 mini (xhigh) on the Pareto frontier?

No: Grok 4.6 (high) beats it on intelligence, cost, or both.

Frontier membership by axis pair

Other axis pairs where GPT-5.4 mini (xhigh) is Pareto-optimal.

Y axisX axisOn the frontier
AA intelligence indexAA total cost to run (USD)
AA intelligence indexEnd-to-end total seconds
AA intelligence indexOutput speed tokens/sec
Agents: Accept: text/markdown or ?format=md returns this page as Markdown. JSON: /api/v1/models/gpt-5-4-mini · Markdown: /api/v1/models/gpt-5-4-mini?format=md · llms.txt · agent skill