BenchLeader

GPT-6.2 Astra vs Kimi K3.1 Pro Together

Full profiles: GPT-6.2 Astra (OpenAI) and Kimi K3.1 Pro Together (Moonshot AI): every benchmark, price and rank.

Updated · Data from 10 sources

As of 19 Nov 2026, GPT-6.2 Astra is ahead of Kimi K3.1 Pro Together on quality (68.9 vs 59.8 on the BenchLeader Index). GPT-6.2 Astra leads in every category we measure both in.

Key facts (as of 19 Nov 2026)

  • GPT-6.2 Astra scores 68.9 on the BenchLeader Index and Kimi K3.1 Pro Together scores 59.8 (BenchLeader, 19 Nov 2026).
  • GPT-6.2 Astra costs $16.00 per 1M tokens blended (list price, 19 Nov 2026).
  • Of the 11 benchmarks both models have, GPT-6.2 Astra leads 10 (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
MathsGPT-6.2 Astra73 vs 65 category score; FrontierMath Tiers 1–3 92.6% vs 78.3%High
ReasoningGPT-6.2 Astra73 vs 67 category score; GPQA Diamond 96.8% vs 94.2%Medium
KnowledgeGPT-6.2 Astra74 vs 65 category score; SimpleQA Verified 78.5% vs 62.2%High
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 59.8.
  • GPT-6.2 Astra (max) is stronger in knowledge, maths, reasoning.

At a glance

SpecGPT-6.2 AstraKimi K3.1 Pro Together
DeveloperOpenAIMoonshot AI
Released17 Nov 202623 Oct 2026
Context window (tokens)1,050,000–
Compared atGPT-6.2 Astra (max)Kimi K3.1 Pro Together (max)
Open weightsNoNo
Price in / out / cached input per 1M$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-6.2 Astra (max)Kimi K3.1 Pro Together (max)Leader
Reasoning73.166.9GPT-6.2 Astra
Coding76.1–not comparable
Agents & tools66.4–not comparable
Maths72.564.7GPT-6.2 Astra
Knowledge74.265.2GPT-6.2 Astra
Multimodal71.7–not comparable
Composite69.5–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 avg75ReasoningMathsKnowledgeGPT-6.2 Astra, Reasoning: 73.1GPT-6.2 Astra, Maths: 72.5GPT-6.2 Astra, Knowledge: 74.2Kimi K3.1 Pro Together, Reasoning: 66.9Kimi K3.1 Pro Together, Maths: 64.7Kimi K3.1 Pro Together, Knowledge: 65.2

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

Model, click to add or removeIndex

Where sources disagree

Across the 11 benchmarks both have, GPT-6.2 Astra leads 10, Kimi K3.1 Pro Together leads 0 and 1 is a tie. Every overall index that rates both puts GPT-6.2 Astra ahead.

Overall indexGPT-6.2 Astra (max)Kimi K3.1 Pro Together (max)Leader
BenchLeader Index68.959.8GPT-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-6.2 Astra (max)Kimi K3.1 Pro Together (max)
Blended price per 1M tokens$16.00–
Input / output per 1M$8.00 / $40.00– / –
Cached input per 1M$0.400–
Price per task: chat reply$0.015–
Price per task: summarize a 30-page report$0.120–
Price per task: code edit$0.108–
Price per task: agentic coding session$0.640–
Price per task: structured extraction$0.024–
Agentic coding session, input mostly cached$0.298–
Cost per benchmark task (AA Index)$4.00–
Output tokens per task (AA Index)41k–
Of which reasoning26k–
Time per task (AA Index)14 min–
Output speed49 tok/s–
Time to first answer202.7 s–

Each model is compared at its best-ranked reasoning-effort setting. 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. Kimi K3.1 Pro Together (max) scores 59.8. 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

81 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-6.2 Astra (max)Kimi K3.1 Pro Together (max)
BenchLeader IndexBenchLeader68.959.8
Agents & tools scoreBenchLeader66.4–
Coding scoreBenchLeader76.1–
Composite scoreBenchLeader69.5–
Knowledge scoreBenchLeader74.265.2
Maths scoreBenchLeader72.564.7
Multimodal scoreBenchLeader71.7–
Reasoning scoreBenchLeader73.166.9
Blended price $/MBenchLeader$16.00–
Output speedBenchLeader49 tok/s–
Time to first answerBenchLeader202.7 s–
Context windowBenchLeader1,050,000 tokens–
Reasoning
GPQA DiamondEpoch AI Benchmarking Hub96.8%94.2%
LiveBench 2.0 ReasoningLiveBench91.6%–
HLE-Diamond (AA)Artificial Analysis72.2%–
ARC-AGI-1ARC Prize98.5%–
ARC-AGI-2ARC Prize95.8%–
ARC-AGI-3ARC Prize99.8%–
CritPt v2 hard subset (AA)Artificial Analysis71.5%–
MysteryMechanismVals AI53.1%–
HLE-Diamond (Scale)Scale AI SEAL66.9%†–
Chess PuzzlesEpoch AI Benchmarking Hub78.0%45.0%
Mystery Game PuzzlesEpoch AI Benchmarking Hub95.0%42.0%
Coding
Arena WebDevArena (LMArena)1781–
LiveBench 2.0 CodingLiveBench91.9%–
Vibe Code Bench v1.1Vals AI89.9%–
Vibe Code Bench 1-100 v1.1Vals AI55.2%–
SWE Atlas v1.1Scale AI SEAL60.5%†–
SWE Atlas v1.1: Codebase QnAScale AI SEAL67.7%†–
SWE Atlas v1.1: Test WritingScale AI SEAL62.2%†–
SWE Atlas v1.1: RefactoringScale AI SEAL51.4%†–
MirrorCodeEpoch AI Benchmarking Hub67.8%43.3%
Terminal-Bench 4.0 (AA)Artificial Analysis64.1%–
Design Arena WebsiteDesign Arena1339†–
FrontierCode 2.0 (AA)Artificial Analysis45.4%–
Agents & tools
Terminal-BenchTerminal-Bench62.1%–
Arena AgentArena (LMArena)15.2–
Show all 81 benchmarks
MetricPublisherGPT-6.2 Astra (max)Kimi K3.1 Pro Together (max)
Agents & tools
LiveBench Long AgenticLiveBench70.8%–
LiveBench BrowserLiveBench93.0%–
Analyst Agent (AA)Artificial Analysis62.5%–
Finance Agent v2.1Vals AI60.8%–
Harvey's Legal Agent Benchmark v1.1Vals AI16.3%–
Legal Research BenchVals AI42.8%–
Robo-benchVals AI32.7%–
Tax Agent BenchVals AI21.8%–
Terminal-Bench 4.0 (Vals)Vals AI63.1%–
Terminal-Bench ScienceVals AI71.4%–
Arena Agent Task CompletionArena (LMArena)12.4–
SpireClimbEpoch AI Benchmarking Hub41.3%0.0%
AutomationBenchArtificial Analysis77.3%–
GDP.pdf (AA)Artificial Analysis29.8%–
Harvey LAB-AA v1.1Harvey15.0%–
EnterpriseOps-GymArtificial Analysis29.9%–
AA-Briefcase v1.1Artificial Analysis1643–
Design Arena AgenticDesign Arena1304–
Terminal-Bench-Science 0.1 (AA)Artificial Analysis78.1%–
CWE-Bench (AA)Artificial Analysis65.0%–
DeepSec-Bench (AA)Artificial Analysis46.8%–
CyberGym end-to-end (AA)Artificial Analysis0.0%–
MKA (AA)Artificial Analysis52.7%–
Maths
FrontierMath Tiers 1–3Epoch AI Benchmarking Hub92.6%78.3%
FrontierMath Tier 4Epoch AI Benchmarking Hub100.0%65.8%
OTIS Mock AIMEEpoch AI Benchmarking Hub100.0%100.0%
Autoform-HardVals AI87.5%–
FrontierMath ErdősEpoch AI Benchmarking Hub10.3%–
Knowledge
SimpleQA VerifiedEpoch AI Benchmarking Hub78.5%62.2%
LiveBench 2.0 Data AnalysisLiveBench84.1%–
LiveBench ResearchLiveBench91.5%–
AA-Omniscience v1.1Artificial Analysis50–
PRBench LegalScale AI SEAL43.1%†–
BioMysteryBenchVals AI80.7%–
Excel Modeling Benchmark v1.1Vals AI76.0%–
DrugDiscoveryBenchScale AI SEAL65.5%†–
EBR-benchEpoch AI Benchmarking Hub92.4%37.1%
AA-Omniscience v1.1: accuracyArtificial Analysis68.6%–
AA-Omniscience v1.1: non-hallucinationArtificial Analysis40.9%–
Instruction following
MultiChallengeScale AI SEAL73.2%†–
Multimodal
LiveBench DesignLiveBench82.8%–
MMMU-ProArtificial Analysis89.5%–
Humanity's Sixth SenseScale AI SEAL63.2%†–
Furniture AssemblyEpoch AI Benchmarking Hub89.2%55.8%
VIALS-AAHandshake AI65.8%–
AA-3DArtificial Analysis1318–
Long context
MLCRArtificial Analysis39.4%–
Composite
Epoch Capabilities IndexEpoch AI Benchmarking Hub174.2†–
LiveBench 2.0LiveBench84.9%–
AA Intelligence Index v5.0.1Artificial Analysis62.1–
Vals IndexVals AI67.7–
Vals Finance IndexVals AI57.3%–
Vals RSI IndexVals AI29.0%–
Vals Science IndexVals AI68.4%–
Vals Legal IndexVals AI69.7%–
AA Cyber IndexArtificial Analysis37.3–

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

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…).
  • Kimi K3.1 Pro Together (max): 3 new results (EBR-bench, MirrorCode, SpireClimb).
  • GPT-6.2 Astra (max): BenchLeader Index 67.2 → 67.8.
  • GPT-6.2 Astra (max): 2 new results (MKA (AA), Terminal-Bench).
  • Kimi K3.1 Pro Together (max): BenchLeader Index 60.5 → 60.0.
  • Kimi K3.1 Pro Together (max): first results.

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

GPT-6.2 Astra vs Kimi K3.1 Pro Together: questions

Is GPT-6.2 Astra better than Kimi K3.1 Pro Together?
GPT-6.2 Astra (max) leads on quality: 68.9 vs 59.8. 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.
Do independent benchmarks agree on GPT-6.2 Astra vs Kimi K3.1 Pro Together?
Across the 11 benchmarks both have, GPT-6.2 Astra leads 10, Kimi K3.1 Pro Together leads 0 and 1 is a tie. 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.