Enter opens the highlighted result. Esc closes.

GPT-5.4 vs GPT-5.6 Terra

GPT-5.4 leads GPT-5.6 Terra on coding, 64.4 to 63.9 on the Vibe Coding Index, and GPT-5.6 Terra costs 1.3x less per output token ($12.00 vs $15.00 per million). Updated October 22, 2026; 96 benchmarks have a result for both.

Replace GPT-5.4 Replace GPT-5.6 Terra
GPT-5.4OpenAI
Coding
64.4
Overall
69.6
Output $/M
$15.00
50
GPT-5.6 TerraOpenAI
Coding
63.9
Overall
68.1
Output $/M
$12.00
45
vs boards won of 96 shared, 1 tied

GPT-5.4 or GPT-5.6 Terra: which to pick

GPT-5.4 and GPT-5.6 Terra are level for coding, 64.4 against 63.9 on the Vibe Coding Index, so price and context decide.

Pick GPT-5.6 Terra for output tokens 1.3x cheaper ($12.00 vs $15.00 per million) and a lower cost per solved task ($0.42 against $1.02). Pick GPT-5.4 for work measured by LMArena Text: coding or LMArena Text: software and IT, where it leads.

Model AGPT-5.4OpenAI64.4/1003 independent benchmarks$1.02 per solved taskModel BGPT-5.6 TerraOpenAI63.9/1006 independent benchmarks$0.42 per solved task
Shared results
96
GPT-5.4 only
14
GPT-5.6 Terra only
31
Like-for-like categories
8 of 9

Only benchmarks that tested both models count toward a pick; a result only one model has is shown but not compared.

Which one for your work

Each card is decided by one number both models carry. Where only one has it, the card says so and makes no call.

At a glance

Coding index
GPT-5.4 64.4 from 3 benchmarks, GPT-5.6 Terra 63.9 from 6 benchmarks.
Widest category
Math, 9.3 points apart on average over 6 shared benchmarks.
Widest single gap
FrontierMath, tier 4: GPT-5.4 49.0% against GPT-5.6 Terra 70.7%, 21.7 points apart on a 0 to 100 scale.
Output price
GPT-5.4 $15.00 against GPT-5.6 Terra $12.00 per million output tokens, 1.3x apart.
Context window
Both take 1.05M tokens.

Benchmark by benchmark

First the score: boards each model leads, per category, where both have a result. Then every board, a category to a fold; bars are each result on a 0 to 100 scale, so compare the two in a row, not one row with another.

Boards won, where both have a result
CategoryGPT-5.4BarsGPT-5.6 TerraTies
Coding2131 tie
Intelligence02
Agentic work16
Reasoning114
Math24
Knowledge83
Web design17
General256
All shared boards50451 tie
CodingLike-for-likeSix of these boards feed the Vibe Coding Index213
Coding benchmarks for GPT-5.4 and GPT-5.6 Terra
BenchmarkGPT-5.4GPT-5.6 TerraLeader
LiveBench codingNot measured86.2%Only GPT-5.6 Terra measured
LMArena WebDev14641522GPT-5.6 Terra
AA Coding Index71.176.7GPT-5.6 Terra
LiveCodeBench (Vals AI)84.1%85.9%GPT-5.6 Terra
SWE-bench Verified (Vals AI)78.2%95.4%GPT-5.6 Terra
Terminal-Bench 4 (Vals AI)Not measured22.7%Only GPT-5.6 Terra measured
Vibe Code Bench (Vals AI)67.4%74.6%GPT-5.6 Terra
Code migration (Vals AI)35.0%47.8%GPT-5.6 Terra
ProgramBench (Vals AI)0.5%0.5%Tie
IOI (Vals AI)Not measured87.6%Only GPT-5.6 Terra measured
Terminal-Bench 2.1 (Vals AI)Not measured77.5%Only GPT-5.6 Terra measured
Vibe Code Bench 1-100 (Vals AI)Not measured14.8%Only GPT-5.6 Terra measured
SkillsBench (Vals AI)51.7%58.9%GPT-5.6 Terra
CyberBench (Vals AI)Not measured72.1%Only GPT-5.6 Terra measured
SciCodeNot measured55.0%Only GPT-5.6 Terra measured
Terminal-Bench Hard (Artificial Analysis run)57.6%62.9%GPT-5.6 Terra
Terminal-Bench81.8%Not measuredOnly GPT-5.4 measured
FrontierCodeNot measured41.3%Only GPT-5.6 Terra measured
GSO-Bench31.4%Not measuredOnly GPT-5.4 measured
WeirdML77.7%78.3%GPT-5.6 Terra
MirrorCode15.6%Not measuredOnly GPT-5.4 measured
ALE-Bench1607.01951.4GPT-5.6 Terra
LMArena Text: coding14961485GPT-5.4
LMArena Text: software and IT14911482GPT-5.4
LMArena WebDev: React14541513GPT-5.6 Terra
LMArena WebDev: HTML14781526GPT-5.6 Terra
LMArena WebDev: from an image14361541GPT-5.6 Terra
SciCodeNot measured10.8%Only GPT-5.6 Terra measured
Neither model has a result on the other 11 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index02
Intelligence benchmarks for GPT-5.4 and GPT-5.6 Terra
BenchmarkGPT-5.4GPT-5.6 TerraLeader
AA Intelligence Index39.042.1GPT-5.6 Terra
Epoch Capabilities Index157.3160.1GPT-5.6 Terra
Agentic workLike-for-likePublic boards, shown as published, kept out of the index16
Agentic work benchmarks for GPT-5.4 and GPT-5.6 Terra
BenchmarkGPT-5.4GPT-5.6 TerraLeader
AA Agentic IndexNot measured43.2Only GPT-5.6 Terra measured
Design Arena: agents building web apps10591120GPT-5.6 Terra
Design Arena: agents building full-stack apps10171061GPT-5.6 Terra
Design Arena: agents building mobile apps10871111GPT-5.6 Terra
Design Arena: agents building Android apps10311062GPT-5.6 Terra
Design Arena: agents building Godot games1135Not measuredOnly GPT-5.4 measured
Design Arena: agents building gamesNot measured1083Only GPT-5.6 Terra measured
Design Arena: HTML slidesNot measured1058Only GPT-5.6 Terra measured
Design Arena: PowerPoint slidesNot measured1105Only GPT-5.6 Terra measured
LiveBench researchNot measured60.4%Only GPT-5.6 Terra measured
LiveBench browserNot measured84.0%Only GPT-5.6 Terra measured
LiveBench long agenticNot measured48.4%Only GPT-5.6 Terra measured
SREBench (Vals AI)Not measured9.5%Only GPT-5.6 Terra measured
Finance Agent (Vals AI)Not measured54.4%Only GPT-5.6 Terra measured
Tax Agent Bench (Vals AI)Not measured30.0%Only GPT-5.6 Terra measured
Legal Research Bench (Vals AI)Not measured41.4%Only GPT-5.6 Terra measured
Excel Modeling Benchmark (Vals AI)Not measured66.2%Only GPT-5.6 Terra measured
Terminal-Bench-Science (Vals AI)Not measured10.0%Only GPT-5.6 Terra measured
τ²-Bench Telecom87.1%86.3%GPT-5.4
APEX-Agents52.4%58.2%GPT-5.6 Terra
METR time horizons74.3%Not measuredOnly GPT-5.4 measured
BALROGNot measured53.2%Only GPT-5.6 Terra measured
DeepResearch Bench35.1%Not measuredOnly GPT-5.4 measured
PostTrainBench19.0%Not measuredOnly GPT-5.4 measured
Vending-Bench 2$6,144$7,343GPT-5.6 Terra
Neither model has a result on the other 16 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index114
Reasoning benchmarks for GPT-5.4 and GPT-5.6 Terra
BenchmarkGPT-5.4GPT-5.6 TerraLeader
GPQA Diamond91.7%90.9%GPT-5.4
LiveBench reasoningNot measured90.6%Only GPT-5.6 Terra measured
MMMU Pro (Vals AI)87.5%86.5%GPT-5.4
GPQA Diamond (Artificial Analysis run)92.0%92.5%GPT-5.6 Terra
Humanity's Last Exam (Artificial Analysis run)43.7%42.9%GPT-5.4
GPQA Diamond (Epoch AI run)93.3%93.3%GPT-5.6 Terra
ARC-AGI93.7%96.5%GPT-5.6 Terra
ARC-AGI-274.0%83.9%GPT-5.6 Terra
SimpleBenchNot measured48.9%Only GPT-5.6 Terra measured
Chess puzzles48.0%47.0%GPT-5.4
Mystery game puzzles37.0%35.0%GPT-5.4
EBR-bench25.4%Not measuredOnly GPT-5.4 measured
LMArena Text: hard prompts14851469GPT-5.4
LMArena Text: expert15071496GPT-5.4
LMArena Text: hard prompts in English14791465GPT-5.4
LMArena Vision: diagrams13271295GPT-5.4
LMArena Vision: homework13361312GPT-5.4
Kagi LLM Benchmark63.8%51.3%GPT-5.4
Neither model has a result on the other 4 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index24
Math benchmarks for GPT-5.4 and GPT-5.6 Terra
BenchmarkGPT-5.4GPT-5.6 TerraLeader
AIME96.7%Not measuredOnly GPT-5.4 measured
OTIS Mock AIME97.8%99.7%GPT-5.6 Terra
FrontierMath, tiers 1 to 378.6%86.0%GPT-5.6 Terra
FrontierMath, tier 449.0%70.7%GPT-5.6 Terra
ProofBench56.0%74.0%GPT-5.6 Terra
LMArena Text: math14881465GPT-5.4
LMArena Text: mathematical work14971478GPT-5.4
Neither model has a result on the other 6 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index83
Knowledge benchmarks for GPT-5.4 and GPT-5.6 Terra
BenchmarkGPT-5.4GPT-5.6 TerraLeader
MMLU Pro87.5%86.7%GPT-5.4
LegalBench (Vals AI)86.0%85.1%GPT-5.4
CaseLaw (Vals AI)63.8%Not measuredOnly GPT-5.4 measured
CorpFin (Vals AI)65.3%65.3%GPT-5.6 Terra
TaxEval (Vals AI)74.0%76.2%GPT-5.6 Terra
MortgageTax (Vals AI)68.3%67.3%GPT-5.4
MedQA (Vals AI)96.1%Not measuredOnly GPT-5.4 measured
MedCode (Vals AI)41.3%43.4%GPT-5.6 Terra
SimpleQA Verified45.1%43.2%GPT-5.4
LMArena Text: science14791452GPT-5.4
LMArena Text: medicine and healthcare14581430GPT-5.4
LMArena Text: legal and government14861461GPT-5.4
LMArena Text: business and finance14741448GPT-5.4
FACTS Benchmark SuiteNot measured52.0%Only GPT-5.6 Terra measured
Neither model has a result on the other 4 boards here
Web designLike-for-likePublic boards, shown as published, kept out of the index17
Web design benchmarks for GPT-5.4 and GPT-5.6 Terra
BenchmarkGPT-5.4GPT-5.6 TerraLeader
Design Arena: websites11981268GPT-5.6 Terra
Design Arena: UI components12121256GPT-5.6 Terra
Design Arena: game dev11941229GPT-5.6 Terra
Design Arena: data viz11961240GPT-5.6 Terra
Design Arena: 3D10651134GPT-5.6 Terra
Design Arena: all code categories11791241GPT-5.6 Terra
Design Arena: SVG11411219GPT-5.6 Terra
Design Arena: ASCII art11881177GPT-5.4
LiveBench designNot measured76.0%Only GPT-5.6 Terra measured
GeneralLike-for-likePublic boards, shown as published, kept out of the index256
General benchmarks for GPT-5.4 and GPT-5.6 Terra
BenchmarkGPT-5.4GPT-5.6 TerraLeader
LiveBench data analysisNot measured79.3%Only GPT-5.6 Terra measured
Public Benefits Bench (Vals AI)Not measured62.4%Only GPT-5.6 Terra measured
SAGE (Vals AI)43.3%47.0%GPT-5.6 Terra
MedScribe (Vals AI)77.5%82.9%GPT-5.6 Terra
Vals Multimodal IndexNot measured65.1%Only GPT-5.6 Terra measured
IFBench73.9%71.2%GPT-5.4
Long context reasoning (AA-LCR)82.0%83.0%GPT-5.6 Terra
DTBench94.4%93.3%GPT-5.4
LMCA52.0%54.9%GPT-5.6 Terra
CL-bench27.9%Not measuredOnly GPT-5.4 measured
Surface Evolver BenchNot measured83.8%Only GPT-5.6 Terra measured
Furniture assembly37.5%54.2%GPT-5.6 Terra
CL-bench Life21.7%Not measuredOnly GPT-5.4 measured
LMArena Text14691448GPT-5.4
LMArena Text: instruction following14691456GPT-5.4
LMArena Text: creative writing14391413GPT-5.4
LMArena Vision13031271GPT-5.4
LMArena Text: English14671449GPT-5.4
LMArena Text: non-English14651441GPT-5.4
LMArena Text: Chinese15201514GPT-5.4
LMArena Text: German14721460GPT-5.4
LMArena Text: Japanese14851454GPT-5.4
LMArena Text: Korean14491427GPT-5.4
LMArena Text: French14931466GPT-5.4
LMArena Text: Spanish14521451GPT-5.4
LMArena Text: Russian14801452GPT-5.4
LMArena Text: Polish14721446GPT-5.4
LMArena Text: multi-turn14821453GPT-5.4
LMArena Text: longer queries14731454GPT-5.4
LMArena Text: writing and language14581431GPT-5.4
LMArena Text: entertainment and media14381418GPT-5.4
LMArena Vision: English13011272GPT-5.4
LMArena Vision: Chinese13541337GPT-5.4
LMArena Vision: reading text13121278GPT-5.4
LMArena Vision: humor12841236GPT-5.4
LMArena Vision: recognizing things1251Not measuredOnly GPT-5.4 measured
LMArena Vision: creative writing12831248GPT-5.4
LMArena Document14711472GPT-5.6 Terra
Neither model has a result on the other 8 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for GPT-5.4 and GPT-5.6 Terra
BenchmarkGPT-5.4GPT-5.6 TerraLeader
Neither model has a result on 5 boards here

What each workload costs

Three fixed token mixes at each model's listed price, so per-token rates become a bill.

Chat turn

1K input + 500 output tokens

GPT-5.4$0.010Fits in one request
GPT-5.6 Terra$0.0080Fits in one request

GPT-5.6 Terra costs less

Repository review

50K input + 3K output tokens

GPT-5.4$0.170Fits in one request
GPT-5.6 Terra$0.136Fits in one request

GPT-5.6 Terra costs less

Agent loop

200K input + 10K output tokens

GPT-5.4$0.650Fits in one request
GPT-5.6 Terra$0.520Fits in one request

GPT-5.6 Terra costs less

Listed OpenRouter rates, without cache or batch discounts. A workload that does not fit a model's context window would need splitting.

Specification differences

The leading value in a row is marked; a row with a missing value names no leader. Our two rows are marked in red.

Specifications of GPT-5.4 and GPT-5.6 Terra
SpecificationOpenAI GPT-5.4OpenAI GPT-5.6 Terra
MakerOpenAIOpenAI
WeightsNo public weights listedNo public weights listed
ReleasedMarch 5, 2026July 9, 2026
Context window1.05M1.05M
Input $/Mtok$2.50$2.00
Output $/Mtok$15.00$12.00
Coding index64.463.9
Independent measurements behind it36
Cost per solved task$1.02$0.42
SourceOpenRouter listingOpenRouter listing

Questions

Is GPT-5.4 better than GPT-5.6 Terra for coding?

GPT-5.4 and GPT-5.6 Terra are level for coding, 64.4 against 63.9 on the Vibe Coding Index, so price and context decide. Pick GPT-5.6 Terra for output tokens 1.3x cheaper ($12.00 vs $15.00 per million) and a lower cost per solved task ($0.42 against $1.02). Pick GPT-5.4 for work measured by LMArena Text: coding or LMArena Text: software and IT, where it leads.

Which is cheaper, GPT-5.4 or GPT-5.6 Terra?

GPT-5.4 costs $2.50 per million input tokens and $15.00 per million output tokens; GPT-5.6 Terra costs $2.00 and $12.00. Per solved SWE-bench Verified task, GPT-5.4 costs $1.02 and GPT-5.6 Terra $0.42.

Which has the bigger context window, GPT-5.4 or GPT-5.6 Terra?

Neither: both take 1.05M tokens.

How many benchmarks is the GPT-5.4 vs GPT-5.6 Terra comparison based on?

96 benchmarks have a result for both models as of October 22, 2026. GPT-5.4's coding index rests on 3 independent benchmarks and GPT-5.6 Terra's on 6. Where only one model has a result, the page says so instead of naming a winner.

Compare next

Embed this comparison on your site

Free under CC BY 4.0, while the source link stays visible. Preview the frame · Terms and other embeds

<iframe src="https://llmbenchmark.ai/embed/compare/gpt-5-4-vs-gpt-5-6-terra" title="GPT-5.4 vs GPT-5.6 Terra, LLMBenchmark.ai" width="100%" height="360" style="border:0;max-width:100%" loading="lazy"></iframe>
<p style="font:13px/1.4 sans-serif;margin:6px 0 0">Source: <a href="https://llmbenchmark.ai/compare/gpt-5-4-vs-gpt-5-6-terra">LLMBenchmark.ai, GPT-5.4 vs GPT-5.6 Terra</a></p>