Agentic work
Tool use, computer use, and multi-step task completion
Claude Sonnet 5.5
Claude Sonnet 5.5 leads on the public agentic lane, 66.1 to 42.6, with Supported evidence for both models and non-overlapping 90% intervals.
Confidence: stronger
Claude Sonnet 5.5 vs GPT-5.2
Updated September 28, 2026. Rank says Claude Sonnet 5.5 is ahead. Price, access, and your workload can each overturn that. Public scores include evidence status and uncertainty.
Both of these models will change. Get the price, version and retirement notices for the pair, each with its source. Follow model changes
Share or export
Decision reading
Claude Sonnet 5.5 has the higher public score, 80.49 versus 61.32, and the 90% score intervals do not overlap. 1 results are shared. Category rows resting on Estimated evidence or different benchmark sets are marked directional and do not name a winner.
- Shared results
- 1
- Claude Sonnet 5.5 only
- 46
- GPT-5.2 only
- 14
- Like-for-like categories
- 2 / 8
Estimated: Claude Sonnet 5.5 · Supported: GPT-5.2How the comparison works
Which one for your work
Recommendations appear only when a shared evidence basis or an explicit operating constraint supports the call. Secondary and unsupported use cases stay disclosed below the initial list.
Long documents
Prompts that approach the documented context limit
Claude Sonnet 5.5
Claude Sonnet 5.5 has the larger documented context window.
Confidence: documentedChat turn cost
1K fresh input + 500 output tokens
Claude Sonnet 5.5
Claude Sonnet 5.5 has the lower estimated token cost for this stated workload. Costs use the listed standard API rates.
Confidence: listed-rates
Show secondary and unsupported calls
Cache-heavy agent loop cost
200K cached + 20K fresh input + 10K output tokens
Claude Sonnet 5.5
Claude Sonnet 5.5 has the lower estimated token cost for this stated workload. GPT-5.2 has no published cached-input rate, so cached tokens use its listed input rate.
Confidence: rate-fallbackRepository review cost
50K fresh input + 3K output tokens
GPT-5.2
GPT-5.2 has the lower estimated token cost for this stated workload. Costs use the listed standard API rates.
Confidence: listed-ratesCoding work
Code generation, repair, and software-engineering tasks
Not enough matched evidence
Claude Sonnet 5.5 is scored on Estimated evidence for coding, so the reading is directional rather than like-for-like.
Confidence: limited
Which one for a specific job
Choose a job from the LLM Selector's task catalog to see the category row it rests on, under the same basis rules as the table below. A directional row stays directional; choosing a job never creates a winner.
The same task catalog as the LLM Selector. Each job names the evidence surface it rests on; nothing here adds a new score.
79.6Claude Sonnet 5.539.5GPT-5.2
Directional only · BenchAlign v5.7
Claude Sonnet 5.5 scores higher, but at least one score rests on Estimated evidence or a different benchmark set. Directional only, no winner.
Coding scores combine specific tasks and setups. Match the editor, harness, and effort to your workflow.
Same basis rules as the category table below
What is actually comparable
Shared results can support a head-to-head reading. Results present for only one model describe coverage, not superiority.
46 Claude Sonnet 5.5 only
1 Shared
14 GPT-5.2 only
1 category rests on Estimated evidence or different evidence sets. Those rows remain visible for coverage context but do not name a winner.
Shape of the matched evidence
Only shared public evidence is shown. Sparse evidence stays a ruled list rather than being closed into a radar shape.
Too few matched category axes support a radar. The ruled list below shows only shared benchmark results; positions use each benchmark’s normalized display scale when available.
SWE-bench ProCoding
Normalized gap 25.7Claude Sonnet 5.5:81.3%GPT-5.2:55.6%
Bars run 0–100 on each benchmark’s normalized display scale
Category results, on a stated basis
Each row shows the public-lane category score for both models: the BenchAlign v5.7 lane for agentic, coding, and knowledge, and the weighted public rows elsewhere, with the same rank each model page reports. A row is like-for-like only when both scores rest on Supported evidence or the same weighted set. Directional and not-comparable rows remain visible, but they never receive a winner.
Agentic
Like-for-like- Claude Sonnet 5.5
- 66.1
- Supported · #10/111
- GPT-5.2
- 42.6
- Supported · #48/111
- Basis
- BenchAlign v5.7 lane · 11 vs 4 public rows
- Reading
- Claude Sonnet 5.5 leads
Knowledge
Like-for-like- Claude Sonnet 5.5
- 80.9
- Supported · #6/160
- GPT-5.2
- 57.9
- Supported · #40/160
- Basis
- BenchAlign v5.7 lane · 16 vs 1 public rows
- Reading
- Claude Sonnet 5.5 leads
Coding
Directional only- Claude Sonnet 5.5
- 79.6
- Estimated · #3/136
- GPT-5.2
- 39.5
- Supported · #56/136
- Basis
- BenchAlign v5.7 lane · 9 vs 3 public rows
- Reading
- Directional only
Reasoning
Not comparable- Claude Sonnet 5.5
- 79.2
- #7/27
- GPT-5.2
- 60.7
- Unranked · 3 rankable rows
- Basis
- Provisional lane · 0 vs 1 weighted rows
- Reading
- Not comparable
Multimodal
Not comparable- Claude Sonnet 5.5
- 83.3
- Unranked · 7 rankable rows
- GPT-5.2
- 67.3
- #23/50
- Basis
- Provisional lane · 1 vs 2 weighted rows
- Reading
- Not comparable
Multilingual
Not comparable- Claude Sonnet 5.5
- Not ranked
- GPT-5.2
- Not ranked
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Not comparable
Instruction following
Not comparable- Claude Sonnet 5.5
- Not ranked
- GPT-5.2
- 91.2
- #15/124
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Not comparable
Math
Not comparable- Claude Sonnet 5.5
- Not ranked
- GPT-5.2
- 57.4
- Unranked · 2 rankable rows
- Basis
- Provisional lane · 0 vs 2 weighted rows
- Reading
- Not comparable
| Category | Claude Sonnet 5.5 | GPT-5.2 | Basis | Reading |
|---|---|---|---|---|
| Agentic | 66.1Supported · #10/111 | 42.6Supported · #48/111 | Like-for-likeBenchAlign v5.7 lane · 11 vs 4 public rows | Claude Sonnet 5.5 leads |
| Knowledge | 80.9Supported · #6/160 | 57.9Supported · #40/160 | Like-for-likeBenchAlign v5.7 lane · 16 vs 1 public rows | Claude Sonnet 5.5 leads |
| Coding | 79.6Estimated · #3/136 | 39.5Supported · #56/136 | Directional onlyBenchAlign v5.7 lane · 9 vs 3 public rows | Directional only |
| Reasoning | 79.2#7/27 | 60.7Unranked · 3 rankable rows | Not comparableProvisional lane · 0 vs 1 weighted rows | Not comparable |
| Multimodal | 83.3Unranked · 7 rankable rows | 67.3#23/50 | Not comparableProvisional lane · 1 vs 2 weighted rows | Not comparable |
| Multilingual | Not ranked | Not ranked | Not comparableProvisional lane · 0 vs 0 weighted rows | Not comparable |
| Instruction following | Not ranked | 91.2#15/124 | Not comparableProvisional lane · 0 vs 0 weighted rows | Not comparable |
| Math | Not ranked | 57.4Unranked · 2 rankable rows | Not comparableProvisional lane · 0 vs 2 weighted rows | Not comparable |
Ranks count the models scored in each category’s lane, so the agentic, coding, and knowledge denominators (BenchAlign v5.7) differ from the provisional-lane categories. Unranked scores sit on the provisional lane’s scale but never name a winner. Like-for-like rows within 0.5 points read as a practical tie.
Supported evidence per lane · bars run 0–100Methodology
What each workload costs
Three fixed token mixes turn per-token rates into comparable decisions. Each scenario states context fit and whether cached input had to fall back to the published list-input rate.
Chat turn
1K fresh input + 500 output tokens
- Claude Sonnet 5.5
- $0.007
- Fits in one request
- GPT-5.2
- $0.00875
- Fits in one request
Claude Sonnet 5.5 has the lower modeled cost
Costs use the listed standard API rates.
Repository review
50K fresh input + 3K output tokens
- Claude Sonnet 5.5
- $0.13
- Fits in one request
- GPT-5.2
- $0.1295
- Fits in one request
GPT-5.2 has the lower modeled cost
Costs use the listed standard API rates.
Cache-heavy agent loop
200K cached + 20K fresh input + 10K output tokens
- Claude Sonnet 5.5
- $0.18
- Fits in one request
- GPT-5.2
- $0.525
- Fits in one request
- Cached input priced at the published list-input rate
Claude Sonnet 5.5 has the lower modeled cost
GPT-5.2 has no published cached-input rate, so cached tokens use its listed input rate.
Cached input falls back to the list input rate only where a cached rate is unpublished
Specification differences
Sourced differences are shown directly. Missing facts stay explicit instead of being inferred from a model name or family.
SpecificationClaude Sonnet 5.5GPT-5.2
Context window
Maximum documented context; output-token limits may be lower.
Claude Sonnet 5.5
GPT-5.2
400K
API model ID
Claude Sonnet 5.5
claude-sonnet-5-5
Anthropic Claude Sonnet 5.5 model documentationGPT-5.2
Not sourced
Cached-input rate
A missing cached-input rate falls back to the listed input rate only in the stated workload estimate.
Claude Sonnet 5.5
$0.2 per 1M cached input tokens
Claude API pricingGPT-5.2
Not published
Documented inputs
Claude Sonnet 5.5
GPT-5.2
Not sourced
Documented outputs
Claude Sonnet 5.5
GPT-5.2
Not sourced
Provider availability
Claude Sonnet 5.5
Generally Available · Claude API, Amazon Bedrock, Google Cloud, Microsoft Foundry, Claude Platform on AWS
Anthropic Claude Sonnet 5.5 model documentationGPT-5.2
Not sourced
Reasoning profile
Claude Sonnet 5.5
Reasoning
GPT-5.2
Reasoning
Weight access
Claude Sonnet 5.5
Proprietary
GPT-5.2
Proprietary
License
Claude Sonnet 5.5
Proprietary
GPT-5.2
Proprietary
Release date
Claude Sonnet 5.5
2026-09-28
GPT-5.2
2025-12-11
If you already use one of these models
- Deployment change
- The models list different providers, so authentication, endpoint behavior, limits, and feature support may change.
- Quality signal
- Claude Sonnet 5.5 has the higher public score, 80.49 versus 61.32, and the 90% score intervals do not overlap.
- Workload cost
- Repository review: $0.13 vs $0.1295. Cache-heavy agent loop: $0.18 vs $0.525.
- Context tradeoff
- Claude Sonnet 5.5 has the larger documented window (1M).
Run the same representative tasks against both endpoints before changing production traffic.
Questions
Which is better, Claude Sonnet 5.5 or GPT-5.2?
Claude Sonnet 5.5 has the higher public score, 80.49 versus 61.32, and the 90% score intervals do not overlap. That is the clearest overall quality signal in the current public data.
Which is better for coding, Claude Sonnet 5.5 or GPT-5.2?
Claude Sonnet 5.5 scores higher for coding on the public lane, 79.6 to 39.5. Claude Sonnet 5.5 is scored on Estimated evidence for coding, so the reading is directional rather than like-for-like. BenchLM does not name a winner for coding from a directional reading; read the shared benchmark rows directly and test the models on the same task set.
Which is better for agentic tasks, Claude Sonnet 5.5 or GPT-5.2?
Claude Sonnet 5.5 leads the public agentic tasks lane, 66.1 to 42.6, with Supported evidence for both models and non-overlapping 90% intervals.
Which costs less, Claude Sonnet 5.5 or GPT-5.2?
For the stated presets, chat costs $0.007 on Claude Sonnet 5.5 and $0.00875 on GPT-5.2; repository review costs $0.13 and $0.1295; the cache-heavy agent loop costs $0.18 and $0.525. GPT-5.2 has no published cached-input rate, so cached tokens use its listed input rate.
Which has the larger context window, Claude Sonnet 5.5 or GPT-5.2?
Claude Sonnet 5.5 has the larger documented context window: 1M, compared with 400K.
Benchmark evidence
The full public result ledger is available for audit without forcing a wide desktop table onto a phone.
Browse raw public benchmark evidence61 rows
Agentic
Terminal-Bench 4.0
Claude Sonnet 5.570.60%SourceGPT-5.2—Not directly comparable
HLE w/ tools
Claude Sonnet 5.564.5%SourceGPT-5.2—Not directly comparable
DRACO
Claude Sonnet 5.587.0%SourceGPT-5.2—Not directly comparable
Terminal-Bench-Science 0.1
Claude Sonnet 5.559.9%SourceGPT-5.2—Not directly comparable
LAB all-pass (Harvey held-out)
Claude Sonnet 5.510.0%SourceGPT-5.2—Not directly comparable
LAB criterion-pass (Harvey held-out)
Claude Sonnet 5.593.1%SourceGPT-5.2—Not directly comparable
Toolathlon Verified Pass@3
Claude Sonnet 5.585.2%SourceGPT-5.2—Not directly comparable
Toolathlon Verified Pass³
Claude Sonnet 5.568.5%SourceGPT-5.2—Not directly comparable
Toolathlon Verified avg. turns
Claude Sonnet 5.531.6 turnsSourceGPT-5.2—Not directly comparable
AutomationBench (Zapier 1.0.6)
Claude Sonnet 5.544.7%SourceGPT-5.2—Not directly comparable
Toolathlon-Verified
Claude Sonnet 5.577.8%SourceGPT-5.2—Not directly comparable
BrowseComp
Claude Sonnet 5.5—GPT-5.265.8%SourceNot directly comparable
OSWorld-Verified
Claude Sonnet 5.5—GPT-5.247.3%SourceNot directly comparable
Gert Labs
Claude Sonnet 5.5—GPT-5.246.54%SourceNot directly comparable
JobBench
Claude Sonnet 5.5—GPT-5.234.3%SourceNot directly comparable
Coding
FrontierCode 1.1 Main
Claude Sonnet 5.546.2%SourceGPT-5.2—Not directly comparable
cursorBench40
Claude Sonnet 5.555.5%SourceGPT-5.2—Not directly comparable
SWE-bench Pro
Claude Sonnet 5.581.3%SourceGPT-5.255.6%SourceClaude Sonnet 5.5 leads this result
SWE Multilingual
Claude Sonnet 5.590.3%SourceGPT-5.2—Not directly comparable
SWE Multimodal
Claude Sonnet 5.554.3%SourceGPT-5.2—Not directly comparable
DeepSWE
Claude Sonnet 5.571.0%SourceGPT-5.2—Not directly comparable
FrontierCode 1.1 Extended
Claude Sonnet 5.559.1%SourceGPT-5.2—Not directly comparable
ProgramBench
Claude Sonnet 5.579.7%SourceGPT-5.2—Not directly comparable
FrontierSWE v2
Claude Sonnet 5.561.9%SourceGPT-5.2—Not directly comparable
SWE-bench Verified
Claude Sonnet 5.5—GPT-5.280%SourceNot directly comparable
Vibe Code Bench
Claude Sonnet 5.5—GPT-5.253.50%SourceNot directly comparable
Reasoning
ARC-AGI-2
Claude Sonnet 5.5—GPT-5.252.9%SourceNot directly comparable
Multimodal
Chartography (no tools)
Claude Sonnet 5.561.6%SourceGPT-5.2—Not directly comparable
Chartography (tools)
Claude Sonnet 5.590.2%SourceGPT-5.2—Not directly comparable
BenchCAD Vision2Code (no tools)
Claude Sonnet 5.50.747SourceGPT-5.2—Not directly comparable
BenchCAD Vision2Code (tools)
Claude Sonnet 5.50.963SourceGPT-5.2—Not directly comparable
Biomedical image analysis
Claude Sonnet 5.572.2%SourceGPT-5.2—Not directly comparable
OfficeQA
Claude Sonnet 5.576.9%SourceGPT-5.2—Not directly comparable
OfficeQA Pro
Claude Sonnet 5.565.6%SourceGPT-5.2—Not directly comparable
MMMU-Pro
Claude Sonnet 5.5—GPT-5.279.5%SourceNot directly comparable
MathVision
Claude Sonnet 5.5—GPT-5.283.0%SourceNot directly comparable
CharXiv
Claude Sonnet 5.5—GPT-5.282.1%SourceNot directly comparable
V*
Claude Sonnet 5.5—GPT-5.275.9%SourceNot directly comparable
Knowledge
HealthBench (raw)
Claude Sonnet 5.569.4%SourceGPT-5.2—Not directly comparable
HealthBench (length-adjusted)
Claude Sonnet 5.565.4%SourceGPT-5.2—Not directly comparable
HealthBench Professional (raw)
Claude Sonnet 5.577.1%SourceGPT-5.2—Not directly comparable
HealthBench Professional
Claude Sonnet 5.569.2%SourceGPT-5.2—Not directly comparable
BioMysteryBench (human-solvable)
Claude Sonnet 5.589.2%SourceGPT-5.2—Not directly comparable
BioMysteryBench (human-difficult)
Claude Sonnet 5.544.7%SourceGPT-5.2—Not directly comparable
SpatialBench Verified
Claude Sonnet 5.572.5%SourceGPT-5.2—Not directly comparable
SingleCellBench
Claude Sonnet 5.559.1%SourceGPT-5.2—Not directly comparable
Protein Design
Claude Sonnet 5.551.0%SourceGPT-5.2—Not directly comparable
Morphology-to-molecule matching
Claude Sonnet 5.525.0%SourceGPT-5.2—Not directly comparable
Medicinal chemistry
Claude Sonnet 5.565.3%SourceGPT-5.2—Not directly comparable
Protein Design library ranking
Claude Sonnet 5.554.8%SourceGPT-5.2—Not directly comparable
De novo protein-binder design
Claude Sonnet 5.582.3%SourceGPT-5.2—Not directly comparable
Protocols (troubleshooting)
Claude Sonnet 5.567.3%SourceGPT-5.2—Not directly comparable
Protocols (understanding)
Claude Sonnet 5.566.6%SourceGPT-5.2—Not directly comparable
HLE w/o tools
Claude Sonnet 5.556.9%SourceGPT-5.2—Not directly comparable
GPQA
Claude Sonnet 5.5—GPT-5.292.4%SourceNot directly comparable
Multilingual
Math
ArXivMath Aug. 2026 (no tools)
Claude Sonnet 5.586.8%SourceGPT-5.2—Not directly comparable
ArXivMath Aug. 2026 (tools)
Claude Sonnet 5.595.2%SourceGPT-5.2—Not directly comparable
FrontierMath v2 (Tiers 1-3)
Claude Sonnet 5.5—GPT-5.240.700%SourceNot directly comparable
FrontierMath v2 (Tier 4)
Claude Sonnet 5.5—GPT-5.218.800%SourceNot directly comparable
61 public results · 1 shared
Watch Claude Sonnet 5.5 vs GPT-5.2
One weekly email when material rank, price, or benchmark evidence changes make this matchup worth revisiting.
Read a sample issueJoin 2,000+ readers.
Last updated September 28, 2026