BenchLeader
Menu
Best LLMsChoose a modelLeaderboardsModelsBenchmarksProvidersReleases9

GPT-5.4 Pro vs GPT-5.6 Terra

Full profiles: GPT-5.4 Pro (OpenAI) and GPT-5.6 Terra (OpenAI): every benchmark, price and rank.

Updated · Data from 9 sources

As of 25 Oct 2026, GPT-5.4 Pro and GPT-5.6 Terra (max) are neck and neck on quality (63.3 vs 64.4 on the BenchLeader Index, inside the ±6.0 margin of error), and GPT-5.6 Terra (max) costs 15× less, which makes it the better buy for most workloads. Pick GPT-5.4 Pro for knowledge (74 vs 57); pick GPT-5.6 Terra for instruction following (70 vs 64).

GPT-5.4 Pro benchmarks OpenAI GPT-5.6 Terra benchmarks max OpenAIOpen in the comparison tool

Which to pick for…

The pick's numbers come first. Category scores on the BenchLeader scale (50 is the average model, 15 points one standard deviation). High confidence means a lead of half a standard deviation or more; under one point is called either.

Use casePickWhyConfidence
ReasoningGPT-5.4 Pro73 vs 70 category scoreMedium
KnowledgeGPT-5.4 Pro74 vs 57 category scoreHigh
Images and other mediaGPT-5.4 Pro70 vs 64 category scoreMedium
Following instructionsGPT-5.6 Terra70 vs 64 category scoreMedium
Lowest costGPT-5.6 Terra$4.50 vs $67.50 per 1M blended; agentic coding session $0.168 vs $2.52–
Fastest first answerGPT-5.4 Pro6.1 s (GPT-5.4 Pro) vs 101.0 s (GPT-5.6 Terra (max)) to the first answer token–
High volume on a budgetNeither flagshipA smaller model is usually the better fit: Cheapest good LLMs, Fastest LLMs–

Key differences

  • GPT-5.4 Pro and GPT-5.6 Terra (max) are neck and neck on quality (63.3 vs 64.4, within the ±6.0 margin).
  • GPT-5.4 Pro is stronger in knowledge, multimodal, reasoning.
  • GPT-5.6 Terra (max) is stronger in instruction following, agents & tools, coding, composite, long context, maths.
  • GPT-5.6 Terra (max) is 15× cheaper ($4.50 vs $67.50 per 1M blended).
  • GPT-5.6 Terra (max) streams 17.9× faster (126 vs 7 tokens per second).

At a glance

SpecGPT-5.4 ProGPT-5.6 Terra
DeveloperOpenAIOpenAI
Released4 Mar 20269 Jul 2026
Context window (tokens)1,050,0001,050,000
Compared atGPT-5.4 ProGPT-5.6 Terra (max)
Open weightsNoNo
Price in / out / cached input per 1M$30.00 / $180.00 / –$2.00 / $12.00 / $0.200

Category by category

Category scores on the BenchLeader scale: 50 is the average evaluated model, 15 points is one standard deviation. A gap under one point is called level.

CategoryGPT-5.4 ProGPT-5.6 Terra (max)Leader
Reasoning72.570.1GPT-5.4 Pro
Coding–65.1not comparable
Agents & tools–58.9not comparable
Maths–68.9not comparable
Knowledge73.657.2GPT-5.4 Pro
Instruction following63.869.9GPT-5.6 Terra
Multimodal69.863.7GPT-5.4 Pro
Long context–66.6not comparable
Composite–76.6not comparable

Where sources disagree

Across the 0 benchmarks both have, GPT-5.4 Pro leads 0, GPT-5.6 Terra leads 0.

Overall indexGPT-5.4 ProGPT-5.6 Terra (max)Leader
BenchLeader Index63.364.4neck and neck

Each publisher runs its own tasks, harnesses and effort settings, so their overall verdicts can differ; BenchLeader reconciles them on one scale. † measured at another effort setting of the same model (the publisher did not test the one compared here).

Price, speed and latency

List prices blended at three input tokens per output token. Per-task prices use the token counts on the methodology page; suite cost is what Artificial Analysis paid to run its whole Intelligence Index.

MeasureGPT-5.4 ProGPT-5.6 Terra (max)
Blended price per 1M tokens$67.50$4.50
Input / output per 1M$30.00 / $180.00$2.00 / $12.00
Cached input per 1M–$0.200
Price per task: chat reply$0.066$0.0044
Price per task: summarize a 30-page report$0.468$0.031
Price per task: code edit$0.450$0.030
Price per task: agentic coding session$2.52$0.168
Price per task: structured extraction$0.096$0.0064
Cost to run a full benchmark suite–$1.40
Output speed7 tok/s126 tok/s
Time to first answer6.1 s101.0 s
Time to a complete answer–105.0 s

Each model is compared at its best-ranked reasoning-effort setting. GPT-5.4 Pro scores 63.3 and takes 6 s to its first answer token; GPT-5.4 Pro (xhigh) scores 61.9 in 6 s. GPT-5.6 Terra (max) scores 64.4 and takes 101 s to its first answer token; GPT-5.6 Terra (xhigh) scores 62.8 in 7 s, GPT-5.6 Terra (high) scores 61.3 in 2 s. Higher effort means more thinking before the answer: better scores, longer waits and more tokens billed. How reasoning effort changes scores and cost.

Every benchmark

121 benchmarks, grouped by category; each name links to the board and its leaders. † measured at another effort setting of the same model, shown for reference and not counted as a win.

MetricPublisherGPT-5.4 ProGPT-5.6 Terra (max)
BenchLeader IndexBenchLeader63.364.4
Instruction following scoreBenchLeader63.869.9
Knowledge scoreBenchLeader73.657.2
Multimodal scoreBenchLeader69.863.7
Reasoning scoreBenchLeader72.570.1
Agents & tools scoreBenchLeader–58.9
Coding scoreBenchLeader–65.1
Composite scoreBenchLeader–76.6
Long context scoreBenchLeader–66.6
Maths scoreBenchLeader–68.9
Blended price $/MBenchLeader$67.50$4.50
Output speedBenchLeader7 tok/s126 tok/s
Time to first answerBenchLeader6.1 s101.0 s
Context windowBenchLeader1,050,000 tokens1,050,000 tokens
Reasoning
GPQA DiamondEpoch AI Benchmarking Hub94.6%†93.3%
Humanity's Last ExamScale AI / CAIS44.3%–
SimpleBenchSimpleBench74.1%48.9%†
Arena Hard PromptsArena (LMArena)–1490†
LiveBench 2.0 ReasoningLiveBench–90.6%†
LiveBench 1.0 ReasoningLiveBench–90.6%
GPQA Diamond (AA)Artificial Analysis–92.5%
Humanity's Last Exam (AA)Artificial Analysis–42.9%
GPQA Diamond (Vals)Vals AI–90.9%†
Kagi LLM BenchmarkKagi LLM Benchmark–51.3%†
ARC-AGI-1ARC Prize94.5%†96.5%
ARC-AGI-2ARC Prize83.3%†83.9%
ARC-AGI-3ARC Prize–0.8%
CritPtArtificial Analysis30.0%†30.0%
Chess PuzzlesEpoch AI Benchmarking Hub59.0%†47.0%
Mystery Game PuzzlesEpoch AI Benchmarking Hub–35.0%
LMCAEpoch AI Benchmarking Hub–55.0%
DTBenchEpoch AI Benchmarking Hub–93.3%
Coding
SciCodeSciCode–55.0%
WeirdMLWeirdML57.4%†78.3%†
FrontierCodeCognition–41.3%†
Arena CodingArena (LMArena)–1518†
Arena WebDevArena (LMArena)–1521†
LiveBench 2.0 CodingLiveBench–86.2%†
LiveBench 1.0 CodingLiveBench–78.3%
Show all 121 benchmarks
MetricPublisherGPT-5.4 ProGPT-5.6 Terra (max)
Coding
LiveCodeBenchVals AI–85.9%†
IOIVals AI–87.6%
SWE-bench (Vals)Vals AI–95.4%
Code MigrationVals AI–47.8%
ProgramBenchVals AI–0.5%
Vibe Code Bench 1-100Vals AI–14.8%
Vibe Code Bench v1.1Vals AI–74.6%
Surface Evolver BenchEpoch AI Benchmarking Hub–83.8%†
DeepSWE v1.1Epoch AI Benchmarking Hub–69.6%
CursorBenchCursor–41.3%
ALE-BenchEpoch AI Benchmarking Hub–1951.4
Terminal-Bench 4.0 (AA)Artificial Analysis–35.4%
Terminal-Bench 2.1 (AA)Artificial Analysis–88.0%
Startrise FrontendStartrise–66.9%†
Agents & tools
Terminal-BenchTerminal-Bench–21.5%
APEX-AgentsMercor–58.2%
Arena AgentArena (LMArena)–0.2†
LiveBench Long AgenticLiveBench–48.4%†
LiveBench BrowserLiveBench–84.0%†
LiveBench Agentic CodingLiveBench–55.0%
Terminal-Bench HardArtificial Analysis–57.6%
τ²-Bench Telecom (AA)Artificial Analysis–86.3%
Terminal-Bench 2.1 (Vals)Vals AI–77.5%
GDPval-AA v2.1Artificial Analysis–47.6%
τ³-Banking (AA)Artificial Analysis–40.2%
ITBench SRE (AA)Artificial Analysis–51.0%
APEX-Agents (AA)Artificial Analysis–38.9%
CyberBenchVals AI–72.1%
Finance Agent v2Vals AI–54.4%
Harvey's Legal Agent Benchmark v1Vals AI–0.8%
Legal Research BenchVals AI–41.4%
SkillsBenchVals AI–58.9%
SREBenchVals AI–9.5%
Tax Agent BenchVals AI–30.0%
Terminal-Bench 4.0 (Vals)Vals AI–22.7%
Terminal-Bench ScienceVals AI–10.0%
BALROGBALROG–53.2%
Vending-Bench 2Andon Labs–7343.2†
AutomationBenchArtificial Analysis–59.6%
GDP.pdf (AA)Artificial Analysis–24.0%
EnterpriseOps-GymArtificial Analysis–38.5%
Terminal-Bench-Science 0.1 (AA)Artificial Analysis–13.8%
FACTS SearchGoogle–81.7%†
Maths
FrontierMath Tiers 1–3Epoch AI Benchmarking Hub82.5%†86.0%
FrontierMath Tier 4Epoch AI Benchmarking Hub58.5%†70.7%
OTIS Mock AIMEEpoch AI Benchmarking Hub–99.7%
ProofBenchVals AI–74.0%†
LiveBench MathematicsLiveBench–94.9%
MathArena ApexMathArena69.8%†–
Arena MathsArena (LMArena)–1483†
Knowledge
SimpleQA VerifiedEpoch AI Benchmarking Hub46.3%†43.2%
LiveBench 2.0 Data AnalysisLiveBench–79.3%†
LiveBench ResearchLiveBench–60.4%†
LiveBench 1.0 Data AnalysisLiveBench–79.3%
AA-OmniscienceArtificial Analysis–0.1
MMLU-ProVals AI–86.7%†
LegalBenchVals AI–85.1%
CorpFinVals AI–65.3%†
TaxEvalVals AI–76.2%†
MultiNRCScale AI SEAL62.3%–
Excel Modeling BenchmarkVals AI–66.2%
MedCodeVals AI–43.4%†
MedScribeVals AI–82.9%†
Public Benefits Bench v1.1Vals AI–62.4%
GDP.pdfEpoch AI Benchmarking Hub–24.7%†
AA-Omniscience: accuracyArtificial Analysis–46.8%
AA-Omniscience: non-hallucinationArtificial Analysis–12.1%
Instruction following
LiveBench LanguageLiveBench–82.9%
LiveBench Instruction FollowingLiveBench–64.6%
IFBenchArtificial Analysis–71.2%
MultiChallengeScale AI SEAL69.2%–
TutorBenchScale AI SEAL56.6%–
Arena Instruction FollowingArena (LMArena)–1464†
Human preference
Arena TextArena (LMArena)–1467†
EQ-Bench 4EQ-Bench–1224†
Arena Creative WritingArena (LMArena)–1429†
Arena Multi-turnArena (LMArena)–1472†
Arena Longer QueriesArena (LMArena)–1468†
Multimodal
Arena VisionArena (LMArena)–1271†
LiveBench DesignLiveBench–76.0%†
MMMU-ProArtificial Analysis–80.7%
VISTAScale AI SEAL53.9%–
MMMU-Pro (Vals)Vals AI–86.5%
MortgageTaxVals AI–67.3%†
SAGEVals AI–47.0%
Blueprint-Bench 2Epoch AI Benchmarking Hub–30.8%†
Long context
AA-LCRArtificial Analysis–83.0%
Arena DocumentArena (LMArena)–1477†
MLCRArtificial Analysis–27.2%
Composite
Epoch Capabilities IndexEpoch AI Benchmarking Hub159.6160.1†
LiveBench 2.0LiveBench–75.0%†
LiveBench 1.0LiveBench–77.9%
AA Intelligence Index v4.3.2Artificial Analysis–42.1
Vals Multimodal IndexVals AI–65.1%†
Safety & honesty
FORTRESSScale AI SEAL14.8%–
MASKScale AI SEAL91.7%–

Data as of 2026-10-25. Best configuration of each model; every score links to its source on the model pages.

What changed

  • GPT-5.6 Terra (max): 1 new result (Harvey's Legal Agent Benchmark v1).
  • GPT-5.4 Pro: 1 score revision.
  • GPT-5.6 Terra (max): 1 score revision.
  • GPT-5.6 Terra (max): BenchLeader Index 63.8 → 64.4.
  • GPT-5.6 Terra (max): 5 new results (AutomationBench, GDP.pdf (AA), MLCR…).
  • GPT-5.6 Terra (max): 1 new result (SREBench); 1 score revision.
  • GPT-5.6 Terra (max): 1 score revision.
  • GPT-5.4 Pro: BenchLeader Index 62.1 → 63.6.
  • GPT-5.4 Pro: 1 new result (MultiNRC).
  • GPT-5.4 Pro: BenchLeader Index 64.0 → 62.1.

Rankings: Best LLM overall · Best LLM for agents · Best cheap LLMs

GPT-5.4 Pro vs GPT-5.6 Terra: questions

Is GPT-5.4 Pro better than GPT-5.6 Terra?
GPT-5.4 Pro and GPT-5.6 Terra (max) are neck and neck on quality (63.3 vs 64.4, within the ±6.0 margin). The BenchLeader Index combines every independent quality benchmark; the two are within the margin of error as of 2026-10-25, so check the category scores for your use.
Which is cheaper, GPT-5.4 Pro or GPT-5.6 Terra, and will that change?
GPT-5.6 Terra is cheaper today: $4.50 against $67.50 per million tokens, blended at three input tokens per output token. No price change has been announced for either.
Which is better for long documents?
Both accept 1,050,000 tokens of context.
Which has lower latency?
GPT-5.4 Pro answers first: 6.1 s against 101.0 s to the first answer token, as Artificial Analysis measures it. Reasoning effort drives most of this: a higher setting thinks longer before it answers.
Which is faster, GPT-5.4 Pro or GPT-5.6 Terra?
GPT-5.6 Terra streams faster: 126 against 7 output tokens per second.
When was GPT-5.6 Terra released?
GPT-5.6 Terra was released on 9 Jul 2026.