BenchLeader

GPT-5.4 vs GPT-6.2 Astra

Full profiles: GPT-5.4 (OpenAI) and GPT-6.2 Astra (OpenAI): every benchmark, price and rank.

Updated · Data from 14 sources

As of 19 Nov 2026, GPT-6.2 Astra is ahead of GPT-5.4 on quality (68.9 vs 63.0 on the BenchLeader Index), but GPT-5.4 is 2.8× cheaper at $5.63 vs $16.00 per 1M tokens blended. GPT-6.2 Astra leads in every category we measure both in. GPT-5.4 streams faster.

Key facts (as of 19 Nov 2026)

  • GPT-5.4 scores 63.0 on the BenchLeader Index and GPT-6.2 Astra scores 68.9 (BenchLeader, 19 Nov 2026).
  • GPT-5.4 costs $5.63 per 1M tokens blended (list price, 19 Nov 2026).
  • GPT-6.2 Astra costs $16.00 per 1M tokens blended (list price, 19 Nov 2026).
  • GPT-5.4 streams faster than GPT-6.2 Astra (Artificial Analysis, 7-day median, 19 Nov 2026).
  • Of the 15 benchmarks both models have, GPT-6.2 Astra leads 15 (BenchLeader, 19 Nov 2026).

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
CodingGPT-6.2 Astra76 vs 65 category score; Vibe Code Bench v1.1 89.9% vs 67.4%High
Agents & tool useGPT-6.2 Astra66 vs 63 category scoreMedium
MathsGPT-6.2 Astra73 vs 65 category score; FrontierMath Tiers 1–3 92.6% vs 78.6%High
ReasoningGPT-6.2 Astra73 vs 65 category score; GPQA Diamond 96.8% vs 93.3%High
KnowledgeGPT-6.2 Astra74 vs 58 category score; SimpleQA Verified 78.5% vs 45.1%High
Images and other mediaGPT-6.2 Astra72 vs 61 category score; MMMU-Pro 89.5% vs 78.4%High
Lowest costGPT-5.4$5.63 vs $16.00 per 1M blended; agentic coding session $0.210 vs $0.640–
Fastest first answerGPT-5.4104.6 s (GPT-5.4 (xhigh)) vs 202.7 s (GPT-6.2 Astra (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-6.2 Astra (max) leads on quality: 68.9 vs 63.0.
  • GPT-6.2 Astra (max) is stronger in agents & tools, coding, knowledge, maths, multimodal, reasoning.
  • GPT-5.4 (xhigh) is 2.8× cheaper ($5.63 vs $16.00 per 1M blended).
  • GPT-5.4 (xhigh) streams 1.3× faster (65 vs 49 tokens per second).

At a glance

SpecGPT-5.4GPT-6.2 Astra
DeveloperOpenAIOpenAI
Released5 Mar 202617 Nov 2026
Context window (tokens)1,050,0001,050,000
Compared atGPT-5.4 (xhigh)GPT-6.2 Astra (max)
Open weightsNoNo
Price in / out / cached input per 1M$2.50 / $15.00 / $0.250$8.00 / $40.00 / $0.400

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 (xhigh)GPT-6.2 Astra (max)Leader
Reasoning65.373.1GPT-6.2 Astra
Coding64.776.1GPT-6.2 Astra
Agents & tools63.066.4GPT-6.2 Astra
Maths64.772.5GPT-6.2 Astra
Knowledge58.474.2GPT-6.2 Astra
Instruction following72.4–not comparable
Multimodal60.571.7GPT-6.2 Astra
Long context65.7–not comparable
Composite–69.5not comparable

On one chart

Each domain on the index's own scale, where 50 is an average model. Click a domain to see the benchmarks behind it. Pick up to 5 models from the list.

2550 avg75ReasoningCodingAgents & toolsMathsKnowledgeMultimodalGPT-5.4, Reasoning: 65.3GPT-5.4, Coding: 64.7GPT-5.4, Agents & tools: 63.0GPT-5.4, Maths: 64.7GPT-5.4, Knowledge: 58.4GPT-5.4, Multimodal: 60.5GPT-6.2 Astra, Reasoning: 73.1GPT-6.2 Astra, Coding: 76.1GPT-6.2 Astra, Agents & tools: 66.4GPT-6.2 Astra, Maths: 72.5GPT-6.2 Astra, Knowledge: 74.2GPT-6.2 Astra, Multimodal: 71.7

Not drawn, measured for fewer than two of these models: Instruction following, Long context, Human preference.

Model, click to add or removeIndex

Where sources disagree

Across the 15 benchmarks both have, GPT-5.4 leads 0, GPT-6.2 Astra leads 15. Every overall index that rates both puts GPT-6.2 Astra ahead.

Overall indexGPT-5.4 (xhigh)GPT-6.2 Astra (max)Leader
BenchLeader Index63.068.9GPT-6.2 Astra

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. Cost, tokens and time per benchmark task are measured: Artificial Analysis’ averages over its whole Intelligence Index run.

MeasureGPT-5.4 (xhigh)GPT-6.2 Astra (max)
Blended price per 1M tokens$5.63$16.00
Input / output per 1M$2.50 / $15.00$8.00 / $40.00
Cached input per 1M$0.250$0.400
Price per task: chat reply$0.0055$0.015
Price per task: summarize a 30-page report$0.039$0.120
Price per task: code edit$0.037$0.108
Price per task: agentic coding session$0.210$0.640
Price per task: structured extraction$0.0080$0.024
Agentic coding session, input mostly cached$0.109$0.298
Cost per benchmark task (AA Index)–$4.00
Output tokens per task (AA Index)–41k
Of which reasoning–26k
Time per task (AA Index)–14 min
Output speed65 tok/s49 tok/s
Time to first answer104.6 s202.7 s
Time to a complete answer, last measured 13 Nov 2026193.1 s–

Each model is compared at its best-ranked reasoning-effort setting. GPT-5.4 (xhigh) scores 63.0 and takes 105 s to its first answer token; GPT-5.4 (low) scores 58.7 in 3 s, GPT-5.4 (high) scores 58.3 in 2 s. GPT-6.2 Astra (max) scores 68.9 and takes 203 s to its first answer token; GPT-6.2 Astra (xhigh) scores 67.1 in 67 s, GPT-6.2 Astra (medium) scores 65.6 in 4 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

177 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 (xhigh)GPT-6.2 Astra (max)
BenchLeader IndexBenchLeader63.068.9
Agents & tools scoreBenchLeader63.066.4
Coding scoreBenchLeader64.776.1
Instruction following scoreBenchLeader72.4–
Knowledge scoreBenchLeader58.474.2
Long context scoreBenchLeader65.7–
Maths scoreBenchLeader64.772.5
Multimodal scoreBenchLeader60.571.7
Reasoning scoreBenchLeader65.373.1
Composite scoreBenchLeader–69.5
Blended price $/MBenchLeader$5.63$16.00
Output speedBenchLeader65 tok/s49 tok/s
Time to first answerBenchLeader104.6 s202.7 s
Context windowBenchLeader1,050,000 tokens1,050,000 tokens
Reasoning
GPQA DiamondEpoch AI Benchmarking Hub93.3%96.8%
Humanity's Last ExamScale AI / CAIS36.2%–
Arena Hard PromptsArena (LMArena)1487†–
LiveBench 2.0 ReasoningLiveBench–91.6%
LiveBench 1.0 ReasoningLiveBench88.1%–
GPQA Diamond (AA)Artificial Analysis92.0%–
Humanity's Last Exam (AA)Artificial Analysis43.7%–
HLE-Diamond (AA)Artificial Analysis–72.2%
GPQA Diamond (Vals)Vals AI91.7%–
Kagi LLM BenchmarkKagi LLM Benchmark63.8%†–
ARC-AGI-1ARC Prize93.7%98.5%
ARC-AGI-2ARC Prize74.0%95.8%
ARC-AGI-3ARC Prize0.2%†99.8%
CritPtArtificial Analysis23.4%–
CritPt v2 hard subset (AA)Artificial Analysis–71.5%
MysteryMechanismVals AI–53.1%
HLE-Diamond (Scale)Scale AI SEAL–66.9%†
Arena ExpertArena (LMArena)1491†–
Chess PuzzlesEpoch AI Benchmarking Hub48.0%78.0%
Mystery Game PuzzlesEpoch AI Benchmarking Hub37.0%95.0%
CL-benchEpoch AI Benchmarking Hub27.9%–
CL-bench LifeEpoch AI Benchmarking Hub21.7%–
LMCAEpoch AI Benchmarking Hub52.0%–
DTBenchEpoch AI Benchmarking Hub94.4%–
ForecastBenchForecastBench62.7%†–
Show all 177 benchmarks
MetricPublisherGPT-5.4 (xhigh)GPT-6.2 Astra (max)
Coding
SWE-bench Verified (Epoch)Epoch AI Benchmarking Hub76.9%†–
SciCodeSciCode56.6%–
WeirdMLWeirdML77.7%–
GSO-BenchGSO-Bench31.4%–
Arena CodingArena (LMArena)1511†–
Arena WebDevArena (LMArena)1400†1781
LiveBench 2.0 CodingLiveBench–91.9%
LiveBench 1.0 CodingLiveBench77.5%–
LiveCodeBenchVals AI84.1%–
SWE-bench (Vals)Vals AI78.2%–
SWE-Bench ProScale AI SEAL59.1%–
Code MigrationVals AI35.0%–
ProgramBenchVals AI0.0%–
Vibe Code Bench v1.1Vals AI67.4%89.9%
Vibe Code Bench 1-100 v1.1Vals AI–55.2%
SWE-Bench Pro (private)Scale AI SEAL43.4%–
SWE Atlas v1.0: Codebase QnAScale AI SEAL40.8%–
SWE Atlas v1.0: RefactoringScale AI SEAL44.3%–
SWE Atlas v1.0: Test WritingScale AI SEAL44.4%–
SWE Atlas v1.1Scale AI SEAL–60.5%†
SWE Atlas v1.1: Codebase QnAScale AI SEAL–67.7%†
SWE Atlas v1.1: Test WritingScale AI SEAL–62.2%†
SWE Atlas v1.1: RefactoringScale AI SEAL–51.4%†
MirrorCodeEpoch AI Benchmarking Hub23.3%67.8%
DeepSWE v1.1Epoch AI Benchmarking Hub51.8%–
ALE-BenchEpoch AI Benchmarking Hub1607†–
AlgoTuneAlgoTune1.9†–
Terminal-Bench 4.0 (AA)Artificial Analysis–64.1%
Terminal-Bench 2.1 (AA)Artificial Analysis78.3%–
GameBench 2 OverallDuelLab GameBench 253.5†–
GameBench 2 by reasoning settingDuelLab GameBench 262.1–
Design Arena WebsiteDesign Arena–1339†
FrontierCode 2.0 (AA)Artificial Analysis–45.4%
Agents & tools
Terminal-BenchTerminal-Bench81.8%†62.1%
APEX-AgentsMercor52.4%†–
Arena AgentArena (LMArena)–15.2
LiveBench Long AgenticLiveBench–70.8%
LiveBench BrowserLiveBench–93.0%
LiveBench Agentic CodingLiveBench53.8%–
Terminal-Bench HardArtificial Analysis57.6%–
τ²-Bench Telecom (AA)Artificial Analysis87.1%–
MCP AtlasScale AI SEAL70.6%–
HiL-BenchScale AI SEAL9.7%†–
τ³-bench Bankingτ²-bench39.4%–
GDPval-AA v2.1Artificial Analysis1256–
τ³-Banking (AA)Artificial Analysis39.6%–
Analyst Agent (AA)Artificial Analysis–62.5%
APEX-Agents (AA)Artificial Analysis33.3%–
Finance Agent v2.1Vals AI–60.8%
Harvey's Legal Agent Benchmark v1.1Vals AI–16.3%
Harvey's Legal Agent Benchmark v1Vals AI0.0%–
Legal Research BenchVals AI–42.8%
Robo-benchVals AI–32.7%
SkillsBenchVals AI51.7%–
Tax Agent BenchVals AI–21.8%
Terminal-Bench 2.0 (Vals)Vals AI58.4%†–
Terminal-Bench 4.0 (Vals)Vals AI–63.1%
Terminal-Bench ScienceVals AI–71.4%
Arena SearchArena (LMArena)1196†–
Arena Agent Task CompletionArena (LMArena)–12.4
DeepResearch BenchDeepResearch Bench35.1%†–
PostTrainBenchEpoch AI Benchmarking Hub17.8%†–
METR Time HorizonsMETR74.3%–
GBAEvalEpoch AI Benchmarking Hub45.1%†–
Vending-Bench 2Andon Labs6144.2†–
SpireClimbEpoch AI Benchmarking Hub–41.3%
AutomationBenchArtificial Analysis–77.3%
GDP.pdf (AA)Artificial Analysis–29.8%
Harvey LAB-AA v1.1Harvey–15.0%
EnterpriseOps-GymArtificial Analysis–29.9%
AA-Briefcase v1.1Artificial Analysis–1643
Design Arena AgenticDesign Arena1034†1304
Terminal-Bench-Science 0.1 (AA)Artificial Analysis–78.1%
CWE-Bench (AA)Artificial Analysis–65.0%
DeepSec-Bench (AA)Artificial Analysis–46.8%
CyberGym end-to-end (AA)Artificial Analysis–0.0%
MKA (AA)Artificial Analysis–52.7%
DeepSearchQAGoogle63.7%†–
Maths
FrontierMath Tiers 1–3Epoch AI Benchmarking Hub78.6%92.6%
FrontierMath Tier 4Epoch AI Benchmarking Hub49.0%100.0%
OTIS Mock AIMEEpoch AI Benchmarking Hub95.3%100.0%
ProofBenchVals AI56.0%–
LiveBench MathematicsLiveBench94.2%–
AIME (Vals)Vals AI96.7%–
AIME 2026MathArena99.2%–
HMMT February 2026MathArena97.7%–
USAMO 2026MathArena95.2%–
MathArena ApexMathArena54.2%–
Autoform-HardVals AI–87.5%
Arena MathsArena (LMArena)1459†–
FrontierMath ErdősEpoch AI Benchmarking Hub–10.3%
Knowledge
SimpleQA VerifiedEpoch AI Benchmarking Hub45.1%78.5%
LiveBench 2.0 Data AnalysisLiveBench–84.1%
LiveBench ResearchLiveBench–91.5%
LiveBench 1.0 Data AnalysisLiveBench79.3%–
AA-Omniscience v1.1Artificial Analysis–50
MMLU-ProVals AI87.5%–
LegalBenchVals AI86.0%–
CorpFinVals AI65.3%–
TaxEvalVals AI74.0%–
MedQAVals AI96.1%–
PRBench FinanceScale AI SEAL45.6%†–
PRBench LegalScale AI SEAL44.4%†43.1%†
MultiNRCScale AI SEAL58.3%†–
BioMysteryBenchVals AI–80.7%
CaseLaw v2Vals AI63.8%–
Excel Modeling Benchmark v1.1Vals AI–76.0%
MedCodeVals AI41.3%–
MedScribeVals AI77.5%–
DrugDiscoveryBenchScale AI SEAL–65.5%†
EBR-benchEpoch AI Benchmarking Hub25.4%92.4%
AA-Omniscience v1.1: accuracyArtificial Analysis–68.6%
AA-Omniscience v1.1: non-hallucinationArtificial Analysis–40.9%
Instruction following
LiveBench LanguageLiveBench82.6%–
LiveBench Instruction FollowingLiveBench70.2%–
IFBenchArtificial Analysis74.0%–
MultiChallengeScale AI SEAL–73.2%†
Arena Instruction FollowingArena (LMArena)1460†–
Human preference
Arena TextArena (LMArena)1465†–
EQ-Bench 4EQ-Bench1258†–
EQ-Bench Creative Writing v3EQ-Bench1841†–
Arena Creative WritingArena (LMArena)1433†–
Arena Multi-turnArena (LMArena)1480†–
Arena Longer QueriesArena (LMArena)1477†–
Multimodal
Arena VisionArena (LMArena)1292†–
LiveBench DesignLiveBench–82.8%
MMMU-ProArtificial Analysis78.4%89.5%
VISTAScale AI SEAL50.9%†–
MMMU-Pro (Vals)Vals AI87.5%–
MortgageTaxVals AI68.3%–
Humanity's Sixth SenseScale AI SEAL–63.2%†
VTBScale AI SEAL29.2%†–
Blueprint-Bench 2Epoch AI Benchmarking Hub27.1%†–
Furniture AssemblyEpoch AI Benchmarking Hub37.5%89.2%
VIALS-AAHandshake AI–65.8%
AA-3DArtificial Analysis–1318
Long context
AA-LCR v1.1Artificial Analysis82.0%–
Arena DocumentArena (LMArena)1471†–
MLCRArtificial Analysis–39.4%
Composite
Epoch Capabilities IndexEpoch AI Benchmarking Hub157.9†174.2†
LiveBench 2.0LiveBench–84.9%
LiveBench 1.0LiveBench78.0%–
AA Intelligence Index v5.0.1Artificial Analysis–62.1
AA Intelligence Index v4.3.2Artificial Analysis39.0–
AA Progress Index v1Artificial Analysis311.6–
Vals IndexVals AI–67.7
Vals Finance IndexVals AI–57.3%
Vals RSI IndexVals AI15.5%29.0%
Vals Science IndexVals AI–68.4%
Vals Legal IndexVals AI–69.7%
AA Cyber IndexArtificial Analysis–37.3
Safety & honesty
MASKScale AI SEAL89.7%†–

Data as of 2026-11-19. Best configuration of each model; every score links to its source on the model pages.

DuelLab GameBench 2, release gb2-20261114-ca113f475f5e (https://benchmarks.duellab.org/releases/gb2-20261114-ca113f475f5e/), CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Scores mapped to BenchLeader model names; not endorsed by DuelLab.

What changed

  • GPT-6.2 Astra (max): BenchLeader Index 67.8 → 68.9.
  • GPT-6.2 Astra (max): 6 new results (VIALS-AA, Arena WebDev, Arena Agent…).
  • GPT-6.2 Astra (max): BenchLeader Index 67.2 → 67.8.
  • GPT-6.2 Astra (max): 2 new results (MKA (AA), Terminal-Bench).
  • GPT-5.4 (xhigh): 1 new result (GDPval-AA v2.1).
  • GPT-5.4 (xhigh): 1 new result (MirrorCode).
  • GPT-5.4 (xhigh): BenchLeader Index 63.7 → 63.2.
  • GPT-5.4 (xhigh): 1 score revision.
  • GPT-5.4 (xhigh): 1 new result (AA Progress Index v1).
  • GPT-5.4 (xhigh): 2 new results (Harvey's Legal Agent Benchmark v1, Vals RSI Index).

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

GPT-5.4 vs GPT-6.2 Astra: questions

Is GPT-6.2 Astra better than GPT-5.4?
GPT-6.2 Astra (max) leads on quality: 68.9 vs 63.0. The BenchLeader Index combines every independent quality benchmark; GPT-6.2 Astra (max) is ahead overall as of 19 Nov 2026, but check the category scores for your use.
GPT-5.4 vs GPT-6.2 Astra: which is cheaper, and will that change?
GPT-5.4 is cheaper today: $5.63 against $16.00 per million tokens, blended at three input tokens per output token. No price change has been announced for either.
Is GPT-5.4 or GPT-6.2 Astra better for coding?
GPT-6.2 Astra scores higher in coding (76 vs 65 on the category index, where 50 is average).
Which is faster, GPT-5.4 or GPT-6.2 Astra?
GPT-5.4 streams faster (65 vs 49 output tokens per second, Artificial Analysis 7-day median).
Is GPT-5.4 or GPT-6.2 Astra better for agents and tool use?
GPT-6.2 Astra scores higher in agentic tasks (66 vs 63 on the category index, where 50 is average).
More questions (4)
Which is better for long documents?
Both accept 1,050,000 tokens of context.
Which has lower latency?
GPT-5.4 answers first: 104.6 s against 202.7 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.
Do independent benchmarks agree on GPT-5.4 vs GPT-6.2 Astra?
Across the 15 benchmarks both have, GPT-5.4 leads 0, GPT-6.2 Astra leads 15. Every overall index that rates both puts GPT-6.2 Astra ahead.
When was GPT-6.2 Astra released?
GPT-6.2 Astra was released on 17 Nov 2026.