GPT-5.4vsGPT-5.6 Terra
GPT-5.4 | GPT-5.6 Terra | |
|---|---|---|
| Benchmarks | ||
Nonsense detection BullshitBench v2Given a confidently-worded but nonsensical prompt, does the AI spot that it makes no sense and push back — instead of playing along and inventing an answer? The score is how often it clearly called out the nonsense. Higher is better. | 48% | 53%Best |
Prompt injection robustness Gray Swan IPI · k = 1Attackers hide malicious instructions inside content the AI reads — a web page, an email, a document — and try to hijack what it does. Gray Swan's indirect prompt injection benchmark measures how often such an attack succeeds when the attacker gets a single try. Lower is better. | — | 5.4% |
Prompt injection robustness Gray Swan IPI · k = 10Attackers hide malicious instructions inside content the AI reads — a web page, an email, a document — and try to hijack what it does. This variant gives the attacker 10 tries and counts an attack as successful if any of them works. Lower is better. | — | 26% |
Prompt injection robustness Gray Swan IPI · k = 15Attackers hide malicious instructions inside content the AI reads — a web page, an email, a document — and try to hijack what it does. This variant gives the attacker 15 tries and counts an attack as successful if any of them works. Lower is better. | — | 30.4% |
Agentic coding CursorBench v3.2Cursor's own test of harder, real-world coding tasks inside a code editor, on the refreshed v3.2 task set. Scores aren't comparable with v3.1. Higher is better. | — | 64.9% |
Next.js coding Next.js EvalsVercel's open eval of how well AI coding agents build and migrate real Next.js apps — measured as the share of tasks the agent completes successfully. Higher is better. | 83% | — |
Agentic computer work Frontier-Bench v0.1A hard, ever-evolving set of real computer tasks — coding, system administration, data work, and more — that an AI agent has to complete on its own. Run by the Harbor / Laude Institute team as the successor to Terminal-Bench (v0.1 is the first release of the task set). The score is the share of tasks solved. Higher is better. | — | 20.8% |
Agentic terminal coding Terminal-Bench 2.1Can the AI work in a command-line terminal — running commands and finishing technical setup tasks the way a developer would? Higher is better. | — | 84.3% |
Agentic terminal coding Terminal-Bench 2.0Can the AI work in a command-line terminal — running commands and finishing technical setup tasks the way a developer would? (Version 2.0 of the test.) Higher is better. | 75.1% | — |
Software engineering Expert-SWE (Internal)OpenAI's private set of expert-level software-engineering problems. Higher is better. | 68.5% | — |
General tool use ToolathlonTests how well the AI uses everyday real-world tools and apps to get things done. Higher is better. | 54.6% | — |
Web browsing BrowseCompCan the AI browse the web and track down hard-to-find answers? Higher is better. | 82.7% | — |
Cybersecurity CyberGymTests the AI on cybersecurity challenges — finding and exploiting software weaknesses inside a safe sandbox. Higher is better. | 79% | — |
Advanced math FrontierMath · Tier 1–3Very hard, research-level math problems. Tiers 1–3 are the (still extremely difficult) lower tiers. Higher is better. | 47.6% | — |
Advanced math FrontierMath · Tier 4Very hard, research-level math problems. Tier 4 is the hardest — close to what professional research mathematicians tackle. Higher is better. | 27.1% | — |
Science GPQA DiamondGraduate-level science questions in biology, physics, and chemistry — hard enough that subject-matter PhDs score around 65%. Higher is better. | 92.8% | — |
Agentic computer use OSWorld-VerifiedCan the AI actually operate a computer — clicking, typing, and using real apps — to finish tasks on its own? Higher is better. | 75% | — |
Knowledge work GDPval (win/tie rate)How often the AI's work matches or beats a human expert's on real knowledge-work tasks. Higher is better. | 83% | — |
Community preference Arena Elo (Text)Real people chat with two anonymous AIs side by side and vote for the answer they prefer. Votes become a chess-style Elo rating on arena.ai — it measures which AI people actually like, not test scores. Higher is better. | 1476Best | 1467 |
Community preference (code) Arena Elo (Code)Like the text arena, but people vote on which AI writes better code. The votes become a chess-style Elo rating on arena.ai. Higher is better. | 1462 | 1526Best |
| Overview | ||
| Company | OpenAI | OpenAI |
| Release date | Mar 5 2026 | Jun 26 2026 |
| Access | Proprietary | Proprietary |
Which is better: GPT-5.4 or GPT-5.6 Terra?
GPT-5.6 Terra leads GPT-5.4 on 2 of the 3 benchmarks they both report (BullshitBench v2, Arena Elo (Text), Arena Elo (Code)). GPT-5.4 shipped 113 days before GPT-5.6 Terra, so benchmark comparisons should account for the intervening progress.
Published specifications for these two models are limited — see each model page for the latest details.
On BullshitBench v2, GPT-5.6 Terra leads at 53% vs GPT-5.4 at 48%. On Arena Elo (Text), GPT-5.4 leads at 1476 vs GPT-5.6 Terra at 1467. On Arena Elo (Code), GPT-5.6 Terra leads at 1526 vs GPT-5.4 at 1462.
Frequently asked questions
GPT-5.4 was released by OpenAI on Mar 5 2026.
GPT-5.6 Terra was released by OpenAI on Jun 26 2026.
Other comparisons
GPT-5.4vsClaude Opus 5GPT-5.6 TerravsClaude Opus 5GPT-5.4vsGemini 3.6 FlashGPT-5.6 TerravsGemini 3.6 FlashGPT-5.4vsMuse GlimmerGPT-5.6 TerravsMuse GlimmerGPT-5.4vsGrok 4.5GPT-5.6 TerravsGrok 4.5GPT-5.4vsDeepSeek-V4-Flash-0731GPT-5.6 TerravsDeepSeek-V4-Flash-0731GPT-5.4vsMistral Medium 3.5GPT-5.6 TerravsMistral Medium 3.5