BenchLeader
Menu
Best LLMsChoose a modelCoding plansLeaderboardsModelsBenchmarksProvidersReleases7

GPT-6 Astra vs Grok 4.20

Full profiles: GPT-6 Astra (OpenAI) and Grok 4.20 (SpaceXAI): every benchmark, price and rank.

Updated · Data from 14 sources

As of 6 Nov 2026, GPT-6 Astra is ahead of Grok 4.20 on quality (68.2 vs 58.4 on the BenchLeader Index), but Grok 4.20 is 13× cheaper at $1.56 vs $20.00 per 1M tokens blended. GPT-6 Astra leads in every category we measure both in. Grok 4.20 streams ~2× faster.

Key facts (as of 6 Nov 2026)

  • GPT-6 Astra scores 68.2 on the BenchLeader Index and Grok 4.20 scores 58.4 (BenchLeader, 6 Nov 2026).
  • GPT-6 Astra costs $20.00 per 1M tokens blended (list price, 6 Nov 2026).
  • Grok 4.20 costs $1.56 per 1M tokens blended (list price, 6 Nov 2026).
  • Grok 4.20 streams ~2× faster than GPT-6 Astra (Artificial Analysis, 7-day median, 6 Nov 2026).
  • Of the 1 benchmarks both models have, GPT-6 Astra leads 1 (BenchLeader, 6 Nov 2026).
GPT-6 Astra benchmarks OpenAI Grok 4.20 benchmarks thinking SpaceXAIAdd a third model

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 Astra83 vs 54 category scoreHigh
Agents & tool useGPT-6 Astra69 vs 51 category scoreHigh
MathsGPT-6 Astra78 vs 54 category score; ProofBench 99.0% vs 14.0%High
ReasoningGPT-6 Astra86 vs 61 category scoreHigh
KnowledgeGPT-6 Astra56 vs 53 category scoreMedium
Chat quality (human preference)–Only Grok 4.20 has an Arena rating (66 category score)Not comparable
Lowest costGrok 4.20$1.56 vs $20.00 per 1M blended; agentic coding session $0.085 vs $0.800–
Fastest first answerGPT-6 Astra4.1 s (GPT-6 Astra) vs 19.9 s (Grok 4.20 (thinking)) 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 Astra leads on quality: 68.2 vs 58.4.
  • GPT-6 Astra is stronger in agents & tools, coding, composite, knowledge, maths, reasoning.
  • Grok 4.20 (thinking) is stronger in human preference, instruction following, long context, multimodal.
  • Grok 4.20 (thinking) is 13× cheaper ($1.56 vs $20.00 per 1M blended).
  • Grok 4.20 (thinking) streams 2.8× faster (102 vs 37 tokens per second).

At a glance

SpecGPT-6 AstraGrok 4.20
DeveloperOpenAISpaceXAI
Released2 Sep 202617 Feb 2026
Context window (tokens)1,050,0002,000,000
Compared atGPT-6 AstraGrok 4.20 (thinking)
Open weightsNoNo
Price in / out / cached input per 1M$10.00 / $50.00 / $1.00$1.25 / $2.50 / $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-6 AstraGrok 4.20 (thinking)Leader
Reasoning85.661.0GPT-6 Astra
Coding83.253.5GPT-6 Astra
Agents & tools69.350.5GPT-6 Astra
Maths77.553.9GPT-6 Astra
Knowledge55.853.0GPT-6 Astra
Instruction following–78.6not comparable
Human preference–66.0not comparable
Multimodal–58.3not comparable
Long context–59.1not comparable
Composite74.2–not 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 & toolsMathsKnowledgeGPT-6 Astra, Reasoning: 85.6GPT-6 Astra, Coding: 83.2GPT-6 Astra, Agents & tools: 69.3GPT-6 Astra, Maths: 77.5GPT-6 Astra, Knowledge: 55.8Grok 4.20, Reasoning: 61.0Grok 4.20, Coding: 53.5Grok 4.20, Agents & tools: 50.5Grok 4.20, Maths: 53.9Grok 4.20, Knowledge: 53.0

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

ModelIndex

Where sources disagree

Across the 1 benchmarks both have, GPT-6 Astra leads 1, Grok 4.20 leads 0. Every overall index that rates both puts GPT-6 Astra ahead.

Overall indexGPT-6 AstraGrok 4.20 (thinking)Leader
BenchLeader Index68.258.4GPT-6 Astra
Arena Text1476†1472GPT-6 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-6 AstraGrok 4.20 (thinking)
Blended price per 1M tokens$20.00$1.56
Input / output per 1M$10.00 / $50.00$1.25 / $2.50
Cached input per 1M$1.00$0.200
Price per task: chat reply$0.019$0.0013
Price per task: summarize a 30-page report$0.150$0.017
Price per task: code edit$0.135$0.011
Price per task: agentic coding session$0.800$0.085
Price per task: structured extraction$0.030$0.0030
Agentic coding session, input mostly cached$0.395$0.038
Output speed37 tok/s102 tok/s
Time to first answer4.1 s19.9 s
Time to a complete answer–50.3 s

Each model is compared at its best-ranked reasoning-effort setting. GPT-6 Astra scores 68.2 and takes 4 s to its first answer token; GPT-6 Astra (max) scores 67.8 in 239 s, GPT-6 Astra (high) scores 67.0 in 56 s. Grok 4.20 (thinking) scores 58.4 and takes 20 s to its first answer token. 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

164 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-6 AstraGrok 4.20 (thinking)
BenchLeader IndexBenchLeader68.258.4
Agents & tools scoreBenchLeader69.350.5
Coding scoreBenchLeader83.253.5
Composite scoreBenchLeader74.2–
Knowledge scoreBenchLeader55.853.0
Maths scoreBenchLeader77.553.9
Reasoning scoreBenchLeader85.661.0
Human preference scoreBenchLeader–66.0
Instruction following scoreBenchLeader–78.6
Long context scoreBenchLeader–59.1
Multimodal scoreBenchLeader–58.3
Blended price $/MBenchLeader$20.00$1.56
Output speedBenchLeader37 tok/s102 tok/s
Time to first answerBenchLeader4.1 s19.9 s
Context windowBenchLeader1,050,000 tokens2,000,000 tokens
Reasoning
GPQA DiamondEpoch AI Benchmarking Hub95.8%†89.3%
Humanity's Last ExamScale AI / CAIS54.8%–
Arena Hard PromptsArena (LMArena)1500†1489
LiveBench 2.0 ReasoningLiveBench92.7%–
LiveBench 1.0 ReasoningLiveBench92.7%†–
GPQA Diamond (AA)Artificial Analysis96.1%†91.1%
Humanity's Last Exam (AA)Artificial Analysis54.7%†34.5%
HLE-Diamond (AA)Artificial Analysis63.8%†–
GPQA Diamond (Vals)Vals AI–88.6%
Kagi LLM BenchmarkKagi LLM Benchmark–75.0%
ARC-AGI-1ARC Prize97.5%†89.5%
ARC-AGI-2ARC Prize95.0%†65.1%
ARC-AGI-3ARC Prize98.6%†0.1%
CritPtArtificial Analysis31.7%†6.6%
CritPt v2 hard subset (AA)Artificial Analysis70.5%†–
MysteryMechanismVals AI53.1%†–
Chess PuzzlesEpoch AI Benchmarking Hub74.0%†29.0%
Mystery Game PuzzlesEpoch AI Benchmarking Hub84.0%†–
CL-benchEpoch AI Benchmarking Hub–22.2%†
CL-bench LifeEpoch AI Benchmarking Hub–11.9%†
LMCAEpoch AI Benchmarking Hub64.1%38.7%†
DTBenchEpoch AI Benchmarking Hub97.1%90.1%†
ForecastBenchForecastBench–64.5%
BTF-3Epoch AI Benchmarking Hub13.5%†–
Coding
SciCodeSciCode56.5%†–
Show all 164 benchmarks
MetricPublisherGPT-6 AstraGrok 4.20 (thinking)
Coding
WeirdMLWeirdML93.3%†52.3%†
FrontierCodeCognition53.3%†–
GSO-BenchGSO-Bench79.4%–
Arena CodingArena (LMArena)1542†1511
Arena WebDevArena (LMArena)1773†1376
LiveBench 2.0 CodingLiveBench93.7%–
LiveBench 1.0 CodingLiveBench80.4%†–
LiveCodeBenchVals AI–84.3%
IOIVals AI100.0%†–
SWE-bench (Vals)Vals AI–72.2%
Code MigrationVals AI67.7%†0.3%
Code Migration v1.1Vals AI80.6%†–
ProgramBenchVals AI5.5%†–
SWE-bench MultimodalVals AI43.3%†–
Vibe Code Bench 1-100Vals AI27.6%†–
Vibe Code Bench v1.1Vals AI89.6%†4.1%
SWE Atlas: Codebase QnAScale AI SEAL59.1%†–
SWE Atlas: RefactoringScale AI SEAL59.0%†–
SWE Atlas: Test WritingScale AI SEAL50.7%†–
MirrorCodeEpoch AI Benchmarking Hub46.7%†–
DeepSWE v1.1Epoch AI Benchmarking Hub73.2%†–
ALE-BenchEpoch AI Benchmarking Hub2951.3†1150.3†
FrontierSWEEpoch AI Benchmarking Hub65.5%†–
Terminal-Bench 4.0 (AA)Artificial Analysis59.1%†–
Terminal-Bench 2.1 (AA)Artificial Analysis88.4%†–
Startrise FrontendStartrise79.9%–
GameBench 2 OverallDuelLab GameBench 288.7–
GameBench 2 by reasoning settingDuelLab GameBench 294.0†–
FrontierCode 1.1 (AA)Artificial Analysis47.0%†–
FrontierCode 2.0 (AA)Artificial Analysis42.4%†–
Agents & tools
Terminal-BenchTerminal-Bench58.2%†57.3%†
Remote Labor IndexScale AI / CAIS20.8%–
APEX-AgentsMercor64.7%–
Arena AgentArena (LMArena)14.7†–
LiveBench Long AgenticLiveBench77.5%–
LiveBench BrowserLiveBench91.3%–
LiveBench Agentic CodingLiveBench57.3%†–
Terminal-Bench HardArtificial Analysis–37.9%
τ²-Bench Telecom (AA)Artificial Analysis–93.0%
Terminal-Bench 2.1 (Vals)Vals AI87.3%†44.2%
HiL-BenchScale AI SEAL65.3%20.0%†
GDPval-AA v2.1, normalised scoreArtificial Analysis53.7%†–
GDPval-AA v2.1Artificial Analysis1584†–
τ³-Banking (AA)Artificial Analysis41.4%†–
ITBench SRE (AA)Artificial Analysis48.6%†–
Analyst Agent (AA)Artificial Analysis51.3%†–
APEX-Agents (AA)Artificial Analysis–14.2%
CUA-benchVals AI19.2%†–
CyberBenchVals AI41.1%†–
Finance Agent v2Vals AI53.5%†28.5%
Finance Agent v2.1Vals AI60.4%†–
Harvey's Legal Agent Benchmark v1.1Vals AI7.1%†–
Harvey's Legal Agent Benchmark v1Vals AI5.4%†0.0%
Legal Research BenchVals AI39.4%†13.9%
Robo-benchVals AI23.3%†–
SREBenchVals AI56.9%†–
Tax Agent BenchVals AI20.7%†7.0%
Terminal-Bench 2.0 (Vals)Vals AI–40.5%
Terminal-Bench 4.0 (Vals)Vals AI59.6%†–
Terminal-Bench ScienceVals AI62.9%†–
Time Horizon Index: KSPVals AI90.5%†–
Arena SearchArena (LMArena)1246†–
BALROGBALROG68.3%†–
Vending-Bench 2Andon Labs15514.74662.9†
AutomationBenchArtificial Analysis68.5%†–
GDP.pdf (AA)Artificial Analysis31.0%†–
Harvey LAB-AA v1.1Harvey8.6%†–
AA-Briefcase v1.1Artificial Analysis1569†–
GDPxGDPx by SemiAnalysis9.6%†–
Terminal-Bench-Science 0.1 (AA)Artificial Analysis63.3%†–
CWE-Bench (AA)Artificial Analysis63.3%†–
DeepSec-Bench (AA)Artificial Analysis36.9%†–
CyberGym end-to-end (AA)Artificial Analysis0.0%†–
FACTS SearchGoogle93.3%–
Maths
FrontierMath Tiers 1–3Epoch AI Benchmarking Hub93.7%†44.9%
FrontierMath Tier 4Epoch AI Benchmarking Hub97.6%†17.1%
OTIS Mock AIMEEpoch AI Benchmarking Hub100.0%†92.2%
ProofBenchVals AI99.0%14.0%
LiveBench MathematicsLiveBench96.8%†–
AIME (Vals)Vals AI–96.5%
Arena MathsArena (LMArena)1489†1465
Knowledge
SimpleQA VerifiedEpoch AI Benchmarking Hub75.6%†30.2%
LiveBench 2.0 Data AnalysisLiveBench83.0%–
LiveBench ResearchLiveBench85.2%–
LiveBench 1.0 Data AnalysisLiveBench83.0%†–
AA-Omniscience v1.1Artificial Analysis42.4†–
MMLU-ProVals AI–86.3%
LegalBenchVals AI–77.7%
CorpFinVals AI–63.7%
TaxEvalVals AI–74.1%
MedQAVals AI–94.5%
PRBench FinanceScale AI SEAL47.5%–
PRBench LegalScale AI SEAL48.4%–
BioMysteryBenchVals AI79.3%†–
CaseLaw v2Vals AI–54.5%
Excel Modeling BenchmarkVals AI71.7%†11.7%
Excel Modeling Benchmark v1.1Vals AI72.7%†–
MedCodeVals AI48.5%†32.2%
MedScribeVals AI87.9%†63.4%
DrugDiscoveryBenchScale AI SEAL68.7%†–
EBR-benchEpoch AI Benchmarking Hub76.2%†–
GDP.pdfEpoch AI Benchmarking Hub34.2%†–
AA-Omniscience v1.1: accuracyArtificial Analysis61.6%†–
AA-Omniscience v1.1: non-hallucinationArtificial Analysis50.2%†–
Global-MMLU-Lite (AA)Artificial Analysis–89.5%
HealthBench Professional (AA)OpenAI66.0%†–
Instruction following
LiveBench LanguageLiveBench89.4%†–
LiveBench Instruction FollowingLiveBench75.6%†–
IFBenchArtificial Analysis–81.2%
Arena Instruction FollowingArena (LMArena)1472†1447
Human preference
Arena TextArena (LMArena)1476†1472
Arena Creative WritingArena (LMArena)1450†1445
Arena Multi-turnArena (LMArena)1484†1479
Arena Longer QueriesArena (LMArena)1488†1466
Multimodal
Arena VisionArena (LMArena)1280†1262
LiveBench DesignLiveBench82.0%–
MMMU-ProArtificial Analysis86.9%†74.6%
MMMU-Pro (Vals)Vals AI–83.5%
MortgageTaxVals AI–45.4%
Blueprint-Bench 2Epoch AI Benchmarking Hub49.7%0.0%†
Code2Video-AAArtificial Analysis1427†–
AA-3DArtificial Analysis1164†–
Long context
AA-LCR v1.1Artificial Analysis80.7%†69.0%
Arena DocumentArena (LMArena)1473†1439
MLCRArtificial Analysis41.7%†–
Composite
Epoch Capabilities IndexEpoch AI Benchmarking Hub169.1152.3†
LiveBench 2.0LiveBench85.6%–
LiveBench 1.0LiveBench82.2%†–
AA Intelligence Index v5.0Artificial Analysis56.9†–
AA Intelligence Index v4.3.2Artificial Analysis52.7†25.7
AA Progress Index v1Artificial Analysis349.6†281.9
Vals IndexVals AI65.0†–
Vals Finance IndexVals AI55.6%†–
Vals RSI IndexVals AI27.1%†–
Vals Science IndexVals AI65.1%†–
Vals Multimodal IndexVals AI–39.1%
Vals Router IndexVals AI65.0%–
AA Cyber IndexArtificial Analysis33.4†–
Safety & honesty
FORTRESSScale AI SEAL12.1%–

Data as of 2026-11-06. Best configuration of each model; every score links to its source on the model pages.

DuelLab GameBench 2, release gb2-20261103-09b2218cbd85 (https://benchmarks.duellab.org/releases/gb2-20261103-09b2218cbd85/), 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 Astra: 1 score revision.
  • GPT-6 Astra: 1 new result (FORTRESS); 1 score revision.
  • GPT-6 Astra: 1 score revision.
  • Grok 4.20 (thinking): BenchLeader Index 58.9 → 58.4.
  • Grok 4.20 (thinking): 1 new result (AA Progress Index v1).
  • GPT-6 Astra: 3 score revisions.
  • Grok 4.20 (thinking): 1 new result (Global-MMLU-Lite (AA)).
  • GPT-6 Astra: 1 new result (Vals Router Index).
  • Grok 4.20 (thinking): 1 new result (Harvey's Legal Agent Benchmark v1).
  • GPT-6 Astra: 1 score revision.

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

GPT-6 Astra vs Grok 4.20: questions

Is GPT-6 Astra better than Grok 4.20?
GPT-6 Astra leads on quality: 68.2 vs 58.4. The BenchLeader Index combines every independent quality benchmark; GPT-6 Astra is ahead overall as of 6 Nov 2026, but check the category scores for your use.
GPT-6 Astra vs Grok 4.20: which is cheaper, and will that change?
Grok 4.20 is cheaper today: $1.56 against $20.00 per million tokens, blended at three input tokens per output token. No price change has been announced for either.
Is GPT-6 Astra or Grok 4.20 better for coding?
GPT-6 Astra scores higher in coding (83 vs 54 on the category index, where 50 is average).
Which is faster, GPT-6 Astra or Grok 4.20?
Grok 4.20 streams ~2× faster (102 vs 37 output tokens per second, Artificial Analysis 7-day median).
Is GPT-6 Astra or Grok 4.20 better for agents and tool use?
GPT-6 Astra scores higher in agentic tasks (69 vs 51 on the category index, where 50 is average).
More questions (4)
Which is better for long documents?
Grok 4.20 accepts more context: 2,000,000 against 1,050,000 tokens for GPT-6 Astra.
Which has lower latency?
GPT-6 Astra answers first: 4.1 s against 19.9 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-6 Astra vs Grok 4.20?
Across the 1 benchmarks both have, GPT-6 Astra leads 1, Grok 4.20 leads 0. Every overall index that rates both puts GPT-6 Astra ahead.
When was GPT-6 Astra released?
GPT-6 Astra was released on 2 Sep 2026.