Skip to main content
BenchLM
Advertise

Model comparison

Claude Sonnet 4.6 vs GPT-5.2

Data verified

Head-to-head evidence from 21 shared benchmark results across 7 categories. Overall scores shown here use the public BenchAlign v5 ranking lane.

65.07/100
Margin
6.6pts
← winning
GPT-5.2
OpenAI
58.43/100
1 category wins4 category wins

Public leaderboard positions: Claude Sonnet 4.6 #32 (Supported); GPT-5.2 #64 (Estimated). Intervals and evidence labels describe ranking uncertainty, not a guarantee for a specific workload.

Evidence parity. Claude Sonnet 4.6 and GPT-5.2 share 21 comparable benchmark results. 5 of 8 categories are comparable. 12 results are unique to Claude Sonnet 4.6; 7 to GPT-5.2.

Updated July 20, 2026
Shared results
21
Claude Sonnet 4.6 only
12
GPT-5.2 only
7
Comparable categories
5 / 8

Pick Claude Sonnet 4.6 if you want the stronger benchmark profile. GPT-5.2 only becomes the better choice if knowledge is the priority or you want the cheaper token bill.

Confidence note. This is a partial-evidence comparison with 21 shared benchmark results across 7 evidence categories; 5 of 8 categories currently have scoreable aggregates for both models. Treat the verdict as directional until coverage is more balanced.

Why this result

Claude Sonnet 4.6 is clearly ahead on the BenchAlign aggregate, 65.07 to 58.43. The gap is large enough that you do not need to squint at the spreadsheet to see the difference.

Claude Sonnet 4.6's sharpest advantage is in agentic, where it averages 65.2 against 55.7. The single biggest benchmark swing on the page is OSWorld-Verified, 72.1% to 47.3%. GPT-5.2 does hit back in knowledge, so the answer changes if that is the part of the workload you care about most.

Claude Sonnet 4.6 is also the more expensive model on tokens at $3.00 input / $15.00 output per 1M tokens, versus $1.75 input / $14.00 output per 1M tokens for GPT-5.2. GPT-5.2 is the reasoning model in the pair, while Claude Sonnet 4.6 is not. That usually helps on harder chain-of-thought-heavy tests, but it can also mean more latency and more token spend in real use. GPT-5.2 gives you the larger context window at 400K, compared with 200K for Claude Sonnet 4.6.

Category breakdown

Exact category averages are shown below. Not measured means BenchLM does not have enough sourced public coverage for that model and category.

Category scores and score margins for Claude Sonnet 4.6 and GPT-5.2
CategoryClaude Sonnet 4.6ΔGPT-5.2
KnowledgeClaude Sonnet 4.666.0Margin→ 26.4GPT-5.292.4
AgenticClaude Sonnet 4.665.2Margin← 9.5GPT-5.255.7
MathClaude Sonnet 4.626.4Margin→ 8.8GPT-5.235.2
MultimodalClaude Sonnet 4.677.4Margin→ 3.0GPT-5.280.4
CodingClaude Sonnet 4.669.1Margin→ 1.5GPT-5.270.6
ReasoningClaude Sonnet 4.6Not measuredMarginNo overlapGPT-5.252.9

Decisive benchmark drivers

The largest measured benchmark gaps in this matchup, with exact reported values.

More
A · Claude Sonnet 4.6B · GPT-5.2

  1. OSWorld-Verified

    Agentic
    Source ↗
    A 72.1%B 47.3%
    Winner: Claude Sonnet 4.6Δ 24.8
    OSWorld-Verified: Claude Sonnet 4.6 scored 72.1%; GPT-5.2 scored 47.3%. Claude Sonnet 4.6 wins this benchmark.

  2. FrontierMath v2 (Tier 4)

    Math
    Source ↗
    A 8.300%B 18.800%
    Winner: GPT-5.2Δ 10.5
    FrontierMath v2 (Tier 4): Claude Sonnet 4.6 scored 8.300%; GPT-5.2 scored 18.800%. GPT-5.2 wins this benchmark.

  3. FrontierMath v2 (Tiers 1-3)

    Math
    Source ↗
    A 32.400%B 40.700%
    Winner: GPT-5.2Δ 8.3
    FrontierMath v2 (Tiers 1-3): Claude Sonnet 4.6 scored 32.400%; GPT-5.2 scored 40.700%. GPT-5.2 wins this benchmark.

  4. CharXiv

    Multimodal
    Source ↗
    A 77.4%B 82.1%
    Winner: GPT-5.2Δ 4.7
    CharXiv: Claude Sonnet 4.6 scored 77.4%; GPT-5.2 scored 82.1%. GPT-5.2 wins this benchmark.

  5. GPQA

    Knowledge
    Source ↗
    A 89.9%B 92.4%
    Winner: GPT-5.2Δ 2.5
    GPQA: Claude Sonnet 4.6 scored 89.9%; GPT-5.2 scored 92.4%. GPT-5.2 wins this benchmark.

Operational comparison

Runtime and commercial metrics are compared only when both models have a complete sourced value.

MetricClaude Sonnet 4.6GPT-5.2Comparison
Input / output priceUSD per 1M tokensClaude Sonnet 4.6$3 input / $15 outputGPT-5.2$1.75 input / $14 outputGPT-5.2 has the lower combined listed price.
Generation speedtokens per secondClaude Sonnet 4.644 tok/sGPT-5.273 tok/sGPT-5.2 has the higher measured throughput.
First-answer latencyseconds to first tokenClaude Sonnet 4.61.48 sGPT-5.2130.34 sClaude Sonnet 4.6 reaches the first token sooner.
Context windowmaximum listed tokensClaude Sonnet 4.6200KGPT-5.2400KGPT-5.2 lists the larger context window.

Benchmark Deep Dive

AgenticClaude Sonnet 4.6 wins
BenchmarkClaude Sonnet 4.6GPT-5.2Result
Terminal-Bench 2.0Source 59.1%—Not comparable
OSWorld-VerifiedSource 72.1%47.3%Claude Sonnet 4.6 leads
Claw-EvalSource 67.8%—Not comparable
CyberGymSource 65.2%—Not comparable
τ²-bench resultsSource 79.5%84.8%GPT-5.2 leads
Gert LabsSource 62.92%46.54%Claude Sonnet 4.6 leads
OSWorld 2.0Source 8.3%—Not comparable
JobBenchSource 36.9%34.3%Claude Sonnet 4.6 leads
BrowseCompSource —65.8%Not comparable
CodingGPT-5.2 wins
BenchmarkClaude Sonnet 4.6GPT-5.2Result
SWE-bench VerifiedSource 79.6%80%GPT-5.2 leads
SWE-RebenchSource 60.7%—Not comparable
React Native EvalsSource 80.6%—Not comparable
Vibe Code BenchSource 51.48%53.50%GPT-5.2 leads
cursorBench31Source 48.8%—Not comparable
AA-SciCodeSource 46.9%52.1%GPT-5.2 leads
FrontierCode 1.1 MainSource 24.3%—Not comparable
SWE-bench ProSource —55.6%Not comparable
Reasoning
BenchmarkClaude Sonnet 4.6GPT-5.2Result
AA-LCRSource 57.7%72.7%GPT-5.2 leads
CritPtSource 0.9%11.6%GPT-5.2 leads
ARC-AGI-2Source —52.9%Not comparable
KnowledgeGPT-5.2 wins
BenchmarkClaude Sonnet 4.6GPT-5.2Result
GPQASource 89.9%92.4%GPT-5.2 leads
SuperGPQASource 95%—Not comparable
MMLU-ProSource 79.2%—Not comparable
HLESource 49%—Not comparable
Artificial Analysis Intelligence IndexSource 35.9%42.2%GPT-5.2 leads
AA-GPQA DiamondSource 79.9%90.3%GPT-5.2 leads
AA-HLESource 13.2%35.4%GPT-5.2 leads
AA-Omniscience IndexSource -2.9%-1.0%GPT-5.2 leads
AA-Omniscience AccuracySource 38.0%43.8%GPT-5.2 leads
AA-Omniscience Hallucination RateSource 65.9%79.7%Claude Sonnet 4.6 leads
MathGPT-5.2 wins
BenchmarkClaude Sonnet 4.6GPT-5.2Result
FrontierMath v2 (Tiers 1-3)Source 32.400%40.700%GPT-5.2 leads
FrontierMath v2 (Tier 4)Source 8.300%18.800%GPT-5.2 leads
AA AIME 2025Source —99.0%Not comparable
MultimodalGPT-5.2 wins
BenchmarkClaude Sonnet 4.6GPT-5.2Result
CharXivSource 77.4%82.1%GPT-5.2 leads
AA-MMMU-ProSource 70.6%—Not comparable
Design Arena WebsiteSource 13171227Claude Sonnet 4.6 leads
MMMU-ProSource —79.5%Not comparable
MathVisionSource —83.0%Not comparable
V*Source —75.9%Not comparable
Inst. Following
BenchmarkClaude Sonnet 4.6GPT-5.2Result
AA-IFBenchSource 41.2%75.4%GPT-5.2 leads
Frequently Asked Questions (6)

Which is better, Claude Sonnet 4.6 or GPT-5.2?

Claude Sonnet 4.6 is ahead on BenchLM's BenchAlign leaderboard, 65.07 to 58.43. The biggest single separator in this matchup is OSWorld-Verified, where the scores are 72.1% and 47.3%.

Which is better for knowledge tasks, Claude Sonnet 4.6 or GPT-5.2?

GPT-5.2 has the edge for knowledge tasks in this comparison, averaging 92.4 versus 66. Inside this category, AA-HLE is the benchmark that creates the most daylight between them.

Which is better for coding, Claude Sonnet 4.6 or GPT-5.2?

GPT-5.2 has the edge for coding in this comparison, averaging 70.6 versus 69.1. Inside this category, AA-SciCode is the benchmark that creates the most daylight between them.

Which is better for math, Claude Sonnet 4.6 or GPT-5.2?

GPT-5.2 has the edge for math in this comparison, averaging 35.2 versus 26.4. Inside this category, FrontierMath v2 (Tier 4) is the benchmark that creates the most daylight between them.

Which is better for agentic tasks, Claude Sonnet 4.6 or GPT-5.2?

Claude Sonnet 4.6 has the edge for agentic tasks in this comparison, averaging 65.2 versus 55.7. Inside this category, OSWorld-Verified is the benchmark that creates the most daylight between them.

Which is better for multimodal and grounded tasks, Claude Sonnet 4.6 or GPT-5.2?

GPT-5.2 has the edge for multimodal and grounded tasks in this comparison, averaging 80.4 versus 77.4. Inside this category, Design Arena Website is the benchmark that creates the most daylight between them.

Related Comparisons

Explore More

Anthropic Compare Pricing Methodology Find Your Best LLM Overall Rankings
Last updated: July 20, 2026

Choose a model with this week’s evidence

Join 2,000+ readers for ranking moves, pricing changes, and the claims that still need proof.

One email each week. Unsubscribe anytime.