BenchLeader

Gemini 4 Hydrogen vs GPT-6.2 Astra

Full profiles: Gemini 4 Hydrogen (Google) and GPT-6.2 Astra (OpenAI): every benchmark, price and rank.

Updated · Data from 11 sources

As of 19 Nov 2026, GPT-6.2 Astra is ahead of Gemini 4 Hydrogen on quality (68.9 vs 63.6 on the BenchLeader Index), but Gemini 4 Hydrogen is 5.3× cheaper at $3.00 vs $16.00 per 1M tokens blended. GPT-6.2 Astra leads in every category we measure both in. Gemini 4 Hydrogen streams 3×+ faster.

Key facts (as of 19 Nov 2026)

  • Gemini 4 Hydrogen scores 63.6 on the BenchLeader Index and GPT-6.2 Astra scores 68.9 (BenchLeader, 19 Nov 2026).
  • Gemini 4 Hydrogen costs $3.00 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).
  • Gemini 4 Hydrogen streams 3×+ faster than GPT-6.2 Astra (Artificial Analysis, 7-day median, 19 Nov 2026).
  • Of the 40 benchmarks both models have, GPT-6.2 Astra leads 33 (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 68 category score; FrontierCode 2.0 (AA) 45.4% vs 39.0%High
MathsGPT-6.2 Astra73 vs 70 category score; FrontierMath Tiers 1–3 92.6% vs 88.1%Medium
ReasoningGPT-6.2 Astra73 vs 70 category score; GPQA Diamond 96.8% vs 96.8%Medium
KnowledgeGPT-6.2 Astra74 vs 70 category score; SimpleQA Verified 78.5% vs 70.9%Medium
Lowest costGemini 4 Hydrogen$3.00 vs $16.00 per 1M blended; agentic coding session $0.120 vs $0.640–
Fastest first answerGemini 4 Hydrogen47.3 s (Gemini 4 Hydrogen (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 63.6.
  • GPT-6.2 Astra (max) is stronger in coding, composite, knowledge, maths, reasoning.
  • Gemini 4 Hydrogen (xhigh) is 5.3× cheaper ($3.00 vs $16.00 per 1M blended).
  • Gemini 4 Hydrogen (xhigh) streams 3.5× faster (168 vs 49 tokens per second).

At a glance

SpecGemini 4 HydrogenGPT-6.2 Astra
DeveloperGoogleOpenAI
Released12 Nov 202617 Nov 2026
Context window (tokens)1,048,5761,050,000
Compared atGemini 4 Hydrogen (xhigh)GPT-6.2 Astra (max)
Open weightsNoNo
Price in / out / cached input per 1M$1.50 / $7.50 / $0.075$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.

CategoryGemini 4 Hydrogen (xhigh)GPT-6.2 Astra (max)Leader
Reasoning70.073.1GPT-6.2 Astra
Coding67.776.1GPT-6.2 Astra
Agents & tools–66.4not comparable
Maths69.872.5GPT-6.2 Astra
Knowledge70.274.2GPT-6.2 Astra
Multimodal–71.7not comparable
Composite60.969.5GPT-6.2 Astra

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 avg75ReasoningCodingMathsKnowledgeGemini 4 Hydrogen, Reasoning: 70.0Gemini 4 Hydrogen, Coding: 67.7Gemini 4 Hydrogen, Maths: 69.8Gemini 4 Hydrogen, Knowledge: 70.2GPT-6.2 Astra, Reasoning: 73.1GPT-6.2 Astra, Coding: 76.1GPT-6.2 Astra, Maths: 72.5GPT-6.2 Astra, Knowledge: 74.2

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

Model, click to add or removeIndex

Where sources disagree

Across the 40 benchmarks both have, Gemini 4 Hydrogen leads 6, GPT-6.2 Astra leads 33 and 1 is a tie. Independent sources split: Gemini 4 Hydrogen leads on Vals Index; GPT-6.2 Astra leads on BenchLeader Index and AA Intelligence Index v5.0.1.

Overall indexGemini 4 Hydrogen (xhigh)GPT-6.2 Astra (max)Leader
BenchLeader Index63.668.9GPT-6.2 Astra
AA Intelligence Index v5.0.153.962.1GPT-6.2 Astra
Vals Index68.567.7Gemini 4 Hydrogen

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.

MeasureGemini 4 Hydrogen (xhigh)GPT-6.2 Astra (max)
Blended price per 1M tokens$3.00$16.00
Input / output per 1M$1.50 / $7.50$8.00 / $40.00
Cached input per 1M$0.075$0.400
Price per task: chat reply$0.0029$0.015
Price per task: summarize a 30-page report$0.022$0.120
Price per task: code edit$0.020$0.108
Price per task: agentic coding session$0.120$0.640
Price per task: structured extraction$0.0045$0.024
Agentic coding session, input mostly cached$0.056$0.298
Cost per benchmark task (AA Index)$3.16$4.00
Output tokens per task (AA Index)154k41k
Of which reasoning97k26k
Time per task (AA Index)15 min14 min
Output speed168 tok/s49 tok/s
Time to first answer47.3 s202.7 s
Time to a complete answer, last measured 13 Nov 2026103.9 s–

Per token is not per task: an average task of Artificial Analysis' Intelligence Index cost $3.16 on Gemini 4 Hydrogen (xhigh) and $4.00 on GPT-6.2 Astra (max), and Gemini 4 Hydrogen (xhigh) wrote 3.8× as many output tokens for the same work, which a per-token price hides.

Each model is compared at its best-ranked reasoning-effort setting. Gemini 4 Hydrogen (xhigh) scores 63.6 and takes 47 s to its first answer token; Gemini 4 Hydrogen (high) scores 63.5 in 21 s, Gemini 4 Hydrogen scores 61.1 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

102 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.

MetricPublisherGemini 4 Hydrogen (xhigh)GPT-6.2 Astra (max)
BenchLeader IndexBenchLeader63.668.9
Coding scoreBenchLeader67.776.1
Composite scoreBenchLeader60.969.5
Knowledge scoreBenchLeader70.274.2
Maths scoreBenchLeader69.872.5
Reasoning scoreBenchLeader70.073.1
Agents & tools scoreBenchLeader–66.4
Multimodal scoreBenchLeader–71.7
Blended price $/MBenchLeader$3.00$16.00
Output speedBenchLeader168 tok/s49 tok/s
Time to first answerBenchLeader47.3 s202.7 s
Context windowBenchLeader1,048,576 tokens1,050,000 tokens
Reasoning
GPQA DiamondEpoch AI Benchmarking Hub96.8%96.8%
Arena Hard PromptsArena (LMArena)1511†–
LiveBench 2.0 ReasoningLiveBench88.6%91.6%
HLE-Diamond (AA)Artificial Analysis58.8%72.2%
ARC-AGI-1ARC Prize–98.5%
ARC-AGI-2ARC Prize–95.8%
ARC-AGI-3ARC Prize–99.8%
CritPt v2 hard subset (AA)Artificial Analysis66.0%71.5%
MysteryMechanismVals AI–53.1%
HLE-Diamond (Scale)Scale AI SEAL–66.9%†
Arena ExpertArena (LMArena)1538†–
Chess PuzzlesEpoch AI Benchmarking Hub60.0%78.0%
Mystery Game PuzzlesEpoch AI Benchmarking Hub65.0%95.0%
Coding
Arena CodingArena (LMArena)1515†–
Arena WebDevArena (LMArena)1703†1781
LiveBench 2.0 CodingLiveBench78.5%91.9%
Vibe Code Bench v1.1Vals AI87.8%89.9%
Vibe Code Bench 1-100 v1.1Vals AI–55.2%
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 Hub–67.8%
CursorBenchCursor57.8%–
ALE-BenchEpoch AI Benchmarking Hub1825.7–
Show all 102 benchmarks
MetricPublisherGemini 4 Hydrogen (xhigh)GPT-6.2 Astra (max)
Coding
FrontierSWEEpoch AI Benchmarking Hub53.1%†–
Terminal-Bench 4.0 (AA)Artificial Analysis53.5%64.1%
Design Arena WebsiteDesign Arena1326†1339†
FrontierCode 2.0 (AA)Artificial Analysis39.0%45.4%
Agents & tools
Terminal-BenchTerminal-Bench56.4%†62.1%
APEX-AgentsMercor78.0%†–
Arena AgentArena (LMArena)13.5†15.2
LiveBench Long AgenticLiveBench61.8%70.8%
LiveBench BrowserLiveBench77.1%93.0%
HiL-BenchScale AI SEAL55.3%†–
Analyst Agent (AA)Artificial Analysis–62.5%
Finance Agent v2.1Vals AI65.1%60.8%
Harvey's Legal Agent Benchmark v1.1Vals AI11.3%16.3%
Legal Research BenchVals AI53.9%42.8%
Robo-benchVals AI–32.7%
Tax Agent BenchVals AI52.3%21.8%
Terminal-Bench 4.0 (Vals)Vals AI54.0%63.1%
Terminal-Bench ScienceVals AI–71.4%
Arena Agent Task CompletionArena (LMArena)15.7†12.4
Vending-Bench 2Andon Labs14549.5†–
SpireClimbEpoch AI Benchmarking Hub7.7%41.3%
AutomationBenchArtificial Analysis76.1%77.3%
GDP.pdf (AA)Artificial Analysis27.6%29.8%
Harvey LAB-AA v1.1Harvey–15.0%
EnterpriseOps-GymArtificial Analysis–29.9%
AA-Briefcase v1.1Artificial Analysis15681643
Design Arena AgenticDesign Arena1266†1304
Terminal-Bench-Science 0.1 (AA)Artificial Analysis49.5%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%
FACTS SearchGoogle93.6%†–
Maths
FrontierMath Tiers 1–3Epoch AI Benchmarking Hub88.1%92.6%
FrontierMath Tier 4Epoch AI Benchmarking Hub87.8%100.0%
OTIS Mock AIMEEpoch AI Benchmarking Hub100.0%100.0%
Autoform-HardVals AI64.3%87.5%
Arena MathsArena (LMArena)1504†–
FrontierMath ErdősEpoch AI Benchmarking Hub–10.3%
Knowledge
SimpleQA VerifiedEpoch AI Benchmarking Hub70.9%78.5%
LiveBench 2.0 Data AnalysisLiveBench83.2%84.1%
LiveBench ResearchLiveBench86.8%91.5%
AA-Omniscience v1.1Artificial Analysis46.350
PRBench FinanceScale AI SEAL52.3%†–
PRBench LegalScale AI SEAL53.1%†43.1%†
BioMysteryBenchVals AI–80.7%
Excel Modeling Benchmark v1.1Vals AI75.8%76.0%
DrugDiscoveryBenchScale AI SEAL69.1%†65.5%†
EBR-benchEpoch AI Benchmarking Hub70.5%92.4%
AA-Omniscience v1.1: accuracyArtificial Analysis50.5%68.6%
AA-Omniscience v1.1: non-hallucinationArtificial Analysis91.5%40.9%
FACTS ParametricGoogle76.7%†–
Instruction following
MultiChallengeScale AI SEAL80.6%†73.2%†
Arena Instruction FollowingArena (LMArena)1499†–
Human preference
Arena TextArena (LMArena)1490†–
Arena Creative WritingArena (LMArena)1491†–
Arena Multi-turnArena (LMArena)1480†–
Arena Longer QueriesArena (LMArena)1511†–
Multimodal
LiveBench DesignLiveBench77.3%82.8%
MMMU-ProArtificial Analysis–89.5%
Humanity's Sixth SenseScale AI SEAL63.4%63.2%†
Furniture AssemblyEpoch AI Benchmarking Hub57.5%89.2%
VIALS-AAHandshake AI–65.8%
AA-3DArtificial Analysis–1318
FACTS MultimodalGoogle63.1%†–
Long context
MLCRArtificial Analysis21.1%39.4%
FACTS GroundingGoogle78.7%†–
Composite
Epoch Capabilities IndexEpoch AI Benchmarking Hub–174.2†
LiveBench 2.0LiveBench77.0%84.9%
AA Intelligence Index v5.0.1Artificial Analysis53.962.1
Vals IndexVals AI68.567.7
Vals Finance IndexVals AI66.1%57.3%
Vals RSI IndexVals AI–29.0%
Vals Science IndexVals AI–68.4%
Vals Legal IndexVals AI73.6%69.7%
AA Cyber IndexArtificial Analysis–37.3
Safety & honesty
FORTRESSScale AI SEAL17.5%†–

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.
  • Gemini 4 Hydrogen (xhigh): 1 new result (Humanity's Sixth Sense).
  • GPT-6.2 Astra (max): 6 new results (VIALS-AA, Arena WebDev, Arena Agent…).
  • Gemini 4 Hydrogen (xhigh): BenchLeader Index 60.8 → 63.7.
  • GPT-6.2 Astra (max): BenchLeader Index 67.2 → 67.8.
  • Gemini 4 Hydrogen (xhigh): 10 new results (GPQA Diamond, FrontierMath Tiers 1–3, FrontierMath Tier 4…).
  • GPT-6.2 Astra (max): 2 new results (MKA (AA), Terminal-Bench).
  • Gemini 4 Hydrogen (xhigh): BenchLeader Index 61.8 → 60.8.
  • Gemini 4 Hydrogen (xhigh): 2 new results (CursorBench, ALE-Bench).
  • Gemini 4 Hydrogen (xhigh): 8 new results (LiveBench 2.0, LiveBench Long Agentic, LiveBench 2.0 Reasoning…).

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

Gemini 4 Hydrogen vs GPT-6.2 Astra: questions

Is GPT-6.2 Astra better than Gemini 4 Hydrogen?
GPT-6.2 Astra (max) leads on quality: 68.9 vs 63.6. 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.
Gemini 4 Hydrogen vs GPT-6.2 Astra: which is cheaper, and will that change?
Gemini 4 Hydrogen is cheaper today: $3.00 against $16.00 per million tokens, blended at three input tokens per output token. No price change has been announced for either.
Is Gemini 4 Hydrogen or GPT-6.2 Astra better for coding?
GPT-6.2 Astra scores higher in coding (76 vs 68 on the category index, where 50 is average). On FrontierCode 2.0 (AA), the top shared coding benchmark, it is 39.0% for Gemini 4 Hydrogen against 45.4% for GPT-6.2 Astra.
Which is faster, Gemini 4 Hydrogen or GPT-6.2 Astra?
Gemini 4 Hydrogen streams 3×+ faster (168 vs 49 output tokens per second, Artificial Analysis 7-day median).
Which is better for long documents?
Both take about 1.1M tokens: Gemini 4 Hydrogen 1,048,576, GPT-6.2 Astra 1,050,000.
More questions (3)
Which has lower latency?
Gemini 4 Hydrogen answers first: 47.3 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 Gemini 4 Hydrogen vs GPT-6.2 Astra?
Across the 40 benchmarks both have, Gemini 4 Hydrogen leads 6, GPT-6.2 Astra leads 33 and 1 is a tie. Independent sources split: Gemini 4 Hydrogen leads on Vals Index; GPT-6.2 Astra leads on BenchLeader Index and AA Intelligence Index v5.0.1.
When was GPT-6.2 Astra released?
GPT-6.2 Astra was released on 17 Nov 2026.