Agentic
Like-for-like- Claude Opus 5
- 78.8
- Supported · #3/156
- Claude Opus 5.5
- 82.7
- Supported · #1/156
- Basis
- BenchAlign lane · 20 vs 11 public rows
- Reading
- Claude Opus 5.5 leads · intervals overlap
Claude Opus 5 vs Claude Opus 5.5
Decision reading
Claude Opus 5.5 has the higher public score estimate, 81.03 versus 80.93, but the 90% score intervals overlap. Treat that as a lead, not a settled winner.
39 results are shared. Category rows resting on Estimated evidence or different benchmark sets are marked directional and do not name a winner.
Updated September 22, 2026. Rank cannot separate these two. Price, access, and your workload decide. 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
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.
Agentic work
Tool use, computer use, and multi-step task completion
Claude Opus 5.5
Claude Opus 5.5 leads on the public agentic lane, 82.7 to 78.8, with Supported evidence for both models, although the 90% intervals overlap.
Confidence: limited
Chat turn cost
1K fresh input + 500 output tokens
Claude Opus 5.5
Claude Opus 5.5 has the lower estimated token cost for this stated workload. Costs use the listed standard API rates.
Confidence: listed-rates
Cache-heavy agent loop cost
200K cached + 20K fresh input + 10K output tokens
Claude Opus 5.5
Claude Opus 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
Repository review cost
50K fresh input + 3K output tokens
Claude Opus 5.5
Claude Opus 5.5 has the lower estimated token cost for this stated workload. Costs use the listed standard API rates.
Confidence: listed-rates
Coding work
Code generation, repair, and software-engineering tasks
Not enough matched evidence
Claude Opus 5.5 is scored on Estimated evidence for coding, so the reading is directional rather than like-for-like.
Confidence: limited
Long documents
Prompts that approach the documented context limit
No clear pick
The documented context windows are equal.
Confidence: documented
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.
75.9Claude Opus 576.0Claude Opus 5.5
Directional only · BenchAlign
Claude Opus 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.
What is actually comparable
Shared results can support a head-to-head reading. Results present for only one model describe coverage, not superiority.
36 Claude Opus 5 only
39 Shared
9 Claude Opus 5.5 only
- Shared results
- 39
- Claude Opus 5 only
- 36
- Claude Opus 5.5 only
- 9
- Like-for-like categories
- 2 / 8
3 categories rest on Estimated evidence or different evidence sets. Those rows remain visible for coverage context but do not name a winner.
Category results, on a stated basis
Each row shows the public-lane category score for both models: the BenchAlign 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.
Multimodal
Like-for-like- Claude Opus 5
- 88.8
- #2/50
- Claude Opus 5.5
- 88.8
- #3/50
- Basis
- Provisional lane · 1 vs 1 weighted rows
- Reading
- Tie
Coding
Directional only- Claude Opus 5
- 75.9
- Supported · #5/159
- Claude Opus 5.5
- 76.0
- Estimated · #4/159
- Basis
- BenchAlign lane · 17 vs 9 public rows
- Reading
- Directional only
Reasoning
Directional only- Claude Opus 5
- 75.7
- #10/19
- Claude Opus 5.5
- 78.5
- #4/19
- Basis
- Provisional lane · 2 vs 0 weighted rows
- Reading
- Directional only
Knowledge
Directional only- Claude Opus 5
- 81.4
- Supported · #5/187
- Claude Opus 5.5
- 81.5
- Estimated · #3/187
- Basis
- BenchAlign lane · 19 vs 17 public rows
- Reading
- Directional only
Multilingual
Not comparable- Claude Opus 5
- Not ranked
- Claude Opus 5.5
- Not ranked
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Not comparable
Instruction following
Not comparable- Claude Opus 5
- Not ranked
- Claude Opus 5.5
- Not ranked
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Not comparable
Math
Not comparable- Claude Opus 5
- Not ranked
- Claude Opus 5.5
- Not ranked
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Not comparable
| Category | Claude Opus 5 | Claude Opus 5.5 | Basis | Reading |
|---|---|---|---|---|
| Agentic | 78.8Supported · #3/156 | 82.7Supported · #1/156 | Like-for-likeBenchAlign lane · 20 vs 11 public rows | Claude Opus 5.5 leads · intervals overlap |
| Multimodal | 88.8#2/50 | 88.8#3/50 | Like-for-likeProvisional lane · 1 vs 1 weighted rows | Tie |
| Coding | 75.9Supported · #5/159 | 76.0Estimated · #4/159 | Directional onlyBenchAlign lane · 17 vs 9 public rows | Directional only |
| Reasoning | 75.7#10/19 | 78.5#4/19 | Directional onlyProvisional lane · 2 vs 0 weighted rows | Directional only |
| Knowledge | 81.4Supported · #5/187 | 81.5Estimated · #3/187 | Directional onlyBenchAlign lane · 19 vs 17 public rows | Directional only |
| Multilingual | Not ranked | Not ranked | Not comparableProvisional lane · 0 vs 0 weighted rows | Not comparable |
| Instruction following | Not ranked | Not ranked | Not comparableProvisional lane · 0 vs 0 weighted rows | Not comparable |
| Math | Not ranked | Not ranked | Not comparableProvisional lane · 0 vs 0 weighted rows | Not comparable |
Ranks count the models scored in each category’s lane, so the agentic, coding, and knowledge denominators (BenchAlign) differ from the provisional-lane categories. Unranked scores sit on the lane’s scale but never name a winner. Like-for-like rows within 0.5 points read as a practical tie.
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.
OSWorld 2.0
Agentic
Claude Opus 5: 70.6%Claude Opus 5.5: 48.7%Normalized gap 21.9Claude Opus 5 source Claude Opus 5.5 sourceAutomationBench
Agentic
Claude Opus 5: 26.0%Claude Opus 5.5: 40.0%Normalized gap 14.0Claude Opus 5 source Claude Opus 5.5 sourceSWE-bench Pro
Coding
Claude Opus 5: 79.2%Claude Opus 5.5: 89.9%Normalized gap 10.7Claude Opus 5 source Claude Opus 5.5 sourceHLE w/o tools
Knowledge
Claude Opus 5: 56.3%Claude Opus 5.5: 64.4%Normalized gap 8.1Claude Opus 5 source Claude Opus 5.5 sourceSWE Multilingual
Coding
Claude Opus 5: 89.5%Claude Opus 5.5: 93.9%Normalized gap 4.4Claude Opus 5 source Claude Opus 5.5 source
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 Opus 5
- $0.0175
- Fits in one request
- Claude Opus 5.5
- $0.014
- Fits in one request
Claude Opus 5.5 has the lower modeled cost
Costs use the listed standard API rates.
Repository review
50K fresh input + 3K output tokens
- Claude Opus 5
- $0.325
- Fits in one request
- Claude Opus 5.5
- $0.26
- Fits in one request
Claude Opus 5.5 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 Opus 5
- $0.45
- Fits in one request
- Claude Opus 5.5
- $0.32
- Fits in one request
Claude Opus 5.5 has the lower modeled cost
Costs use the listed standard API rates.
Specification differences
Sourced differences are shown directly. Missing facts stay explicit instead of being inferred from a model name or family.
SpecificationClaude Opus 5Claude Opus 5.5
Context window
Maximum documented context; output-token limits may be lower.
Claude Opus 5
Claude Opus 5.5
API model ID
Claude Opus 5
claude-opus-5
Anthropic model overviewClaude Opus 5.5
claude-opus-5-5
Anthropic Claude Opus 5.5 model documentationCached-input rate
A missing cached-input rate falls back to the listed input rate only in the stated workload estimate.
Claude Opus 5
$0.5 per 1M cached input tokens
Claude API pricingClaude Opus 5.5
$0.2 per 1M cached input tokens
Claude Opus 5.5 model documentationDocumented inputs
Claude Opus 5
text, image
Anthropic model overviewClaude Opus 5.5
Documented outputs
Claude Opus 5
Claude Opus 5.5
Provider availability
Claude Opus 5
Generally Available · Claude API
Anthropic model overviewClaude Opus 5.5
Generally Available · Claude API, Amazon Bedrock, Google Cloud, Microsoft Foundry, Claude Platform on AWS
Anthropic Claude Opus 5.5 model documentationReasoning profile
Claude Opus 5
Reasoning
Claude Opus 5.5
Reasoning
Weight access
Claude Opus 5
Proprietary
Claude Opus 5.5
Proprietary
License
Claude Opus 5
Proprietary
Claude Opus 5.5
Proprietary
Release date
Claude Opus 5
2026-07-24
Claude Opus 5.5
2026-09-22
If you are considering the documented upgrade path
- Deployment change
- Both entries list Anthropic as the provider. Confirm endpoint, model ID, limits, and feature support before switching.
- Quality signal
- Claude Opus 5.5 has the higher public score estimate, 81.03 versus 80.93, but the 90% score intervals overlap.
- Workload cost
- Repository review: $0.325 vs $0.26. Cache-heavy agent loop: $0.45 vs $0.32.
- Context tradeoff
- Both models list 1M.
Run the same representative tasks against both endpoints before changing production traffic.
Benchmark evidence
The full public result ledger is available for audit without forcing a wide desktop table onto a phone.
Browse raw public benchmark evidence84 rows
Agentic
Terminal-Bench 3.0
Claude Opus 542.7%SourceClaude Opus 5.5—Not directly comparable
BrowseComp
Claude Opus 590.8%SourceClaude Opus 5.5—Not directly comparable
HLE w/ tools
Claude Opus 564.7%SourceClaude Opus 5.567.7%SourceClaude Opus 5.5 leads this result
DeepSearchQA
Claude Opus 595.0%SourceClaude Opus 5.5—Not directly comparable
DRACO
Claude Opus 588.6%SourceClaude Opus 5.5—Not directly comparable
BrowseComp (10-agent, prerelease)
Claude Opus 593.6%SourceClaude Opus 5.5—Not directly comparable
OSWorld 2.0
Claude Opus 570.6%SourceClaude Opus 5.548.7%SourceClaude Opus 5 leads this result
MCP Atlas
Claude Opus 585.8%SourceClaude Opus 5.5—Not directly comparable
MCP-Atlas claim coverage
Claude Opus 589.1%SourceClaude Opus 5.5—Not directly comparable
LAB all-pass (Anthropic harness)
Claude Opus 523.58%SourceClaude Opus 5.5—Not directly comparable
LAB criterion-pass (Anthropic harness)
Claude Opus 593.74%SourceClaude Opus 5.5—Not directly comparable
LAB all-pass (Harvey held-out)
Claude Opus 511.7%SourceClaude Opus 5.58.3%SourceClaude Opus 5 leads this result
LAB criterion-pass (Harvey held-out)
Claude Opus 594.1%SourceClaude Opus 5.591.2%SourceClaude Opus 5 leads this result
Toolathlon-Verified
Claude Opus 580.6%SourceClaude Opus 5.577.8%SourceClaude Opus 5 leads this result
Toolathlon Verified Pass@3
Claude Opus 587.0%SourceClaude Opus 5.582.4%SourceClaude Opus 5 leads this result
Toolathlon Verified Pass³
Claude Opus 573.1%SourceClaude Opus 5.572.2%SourceClaude Opus 5 leads this result
Toolathlon Verified avg. turns
Claude Opus 523.5 turnsSourceClaude Opus 5.526.9 turnsSourceClaude Opus 5.5 leads this result
AutomationBench
Claude Opus 526.0%SourceClaude Opus 5.540.0%SourceClaude Opus 5.5 leads this result
Terminal-Bench 2.1 (Vals)
Claude Opus 584.6%SourceClaude Opus 5.5—Not directly comparable
ApprenticeBench
Claude Opus 536%SourceClaude Opus 5.5—Not directly comparable
Terminal-Bench 4.0
Claude Opus 5—Claude Opus 5.566.40%SourceNot directly comparable
Terminal-Bench-Science 0.1
Claude Opus 5—Claude Opus 5.558.7%SourceNot directly comparable
Coding
Bug Hunt Bench
Claude Opus 527 fixesSourceClaude Opus 5.5—Not directly comparable
SWE-bench Verified
Claude Opus 596%SourceClaude Opus 5.5—Not directly comparable
SWE-bench Pro
Claude Opus 579.2%SourceClaude Opus 5.589.9%SourceClaude Opus 5.5 leads this result
SWE Multilingual
Claude Opus 589.5%SourceClaude Opus 5.593.9%SourceClaude Opus 5.5 leads this result
SWE Multimodal
Claude Opus 559.4%SourceClaude Opus 5.561.4%SourceClaude Opus 5.5 leads this result
DeepSWE
Claude Opus 568.8%SourceClaude Opus 5.574.2%SourceClaude Opus 5.5 leads this result
FrontierCode 1.1 Main
Claude Opus 553.4%SourceClaude Opus 5.554.4%SourceClaude Opus 5.5 leads this result
FrontierCode 1.1 Extended
Claude Opus 563.6%SourceClaude Opus 5.563.6%SourceTie
FrontierSWE v2
Claude Opus 552.0%SourceClaude Opus 5.562.3%SourceClaude Opus 5.5 leads this result
ProgramBench (episode 1)
Claude Opus 583.0%SourceClaude Opus 5.5—Not directly comparable
ProgramBench
Claude Opus 593.0%SourceClaude Opus 5.591.2%SourceClaude Opus 5 leads this result
cursorBench32
Claude Opus 570.0%SourceClaude Opus 5.5—Not directly comparable
VulcanBench v3
Claude Opus 587.0%SourceClaude Opus 5.5—Not directly comparable
VulcanBench CII v1
Claude Opus 596.4%SourceClaude Opus 5.5—Not directly comparable
LiveCodeBench (Vals)
Claude Opus 589.0%SourceClaude Opus 5.5—Not directly comparable
SWE-bench (Vals)
Claude Opus 597.0%SourceClaude Opus 5.5—Not directly comparable
cursorBench40
Shared sourceClaude Opus 546.6%Claude Opus 5.557.8%Claude Opus 5.5 leads this result
Reasoning
Multimodal
Chartography (no tools)
Claude Opus 529.6%SourceClaude Opus 5.564.4%SourceClaude Opus 5.5 leads this result
Chartography (tools)
Claude Opus 583.0%SourceClaude Opus 5.589.0%SourceClaude Opus 5.5 leads this result
BenchCAD Vision2Code (no tools)
Claude Opus 50.366SourceClaude Opus 5.50.730SourceClaude Opus 5.5 leads this result
BenchCAD Vision2Code (tools)
Claude Opus 50.821SourceClaude Opus 5.50.962SourceClaude Opus 5.5 leads this result
GDP.pdf (no tools)
Claude Opus 583.4%SourceClaude Opus 5.5—Not directly comparable
GDP.pdf (tools)
Claude Opus 585.5%SourceClaude Opus 5.5—Not directly comparable
OfficeQA
Claude Opus 578.1%SourceClaude Opus 5.578.9%SourceClaude Opus 5.5 leads this result
OfficeQA Pro
Claude Opus 566.9%SourceClaude Opus 5.567.7%SourceClaude Opus 5.5 leads this result
Biomedical image analysis
Claude Opus 5—Claude Opus 5.571.4%SourceNot directly comparable
Knowledge
HLE
Claude Opus 564.7%SourceClaude Opus 5.567.7%SourceClaude Opus 5.5 leads this result
HLE w/o tools
Claude Opus 556.3%SourceClaude Opus 5.564.4%SourceClaude Opus 5.5 leads this result
HLE-Verified
Claude Opus 554.4%SourceClaude Opus 5.5—Not directly comparable
LABBench2
Claude Opus 584.2%SourceClaude Opus 5.5—Not directly comparable
HealthBench (raw)
Claude Opus 567.1%SourceClaude Opus 5.568.1%SourceClaude Opus 5.5 leads this result
HealthBench (length-adjusted)
Claude Opus 557.8%SourceClaude Opus 5.560.6%SourceClaude Opus 5.5 leads this result
HealthBench Professional
Claude Opus 559.8%SourceClaude Opus 5.565.6%SourceClaude Opus 5.5 leads this result
HealthBench Professional (raw)
Claude Opus 573.4%SourceClaude Opus 5.577.1%SourceClaude Opus 5.5 leads this result
BioMysteryBench (human-solvable)
Claude Opus 590.1%SourceClaude Opus 5.589.3%SourceClaude Opus 5 leads this result
BioMysteryBench (human-difficult)
Claude Opus 549.4%SourceClaude Opus 5.550.0%SourceClaude Opus 5.5 leads this result
SpatialBench Verified
Claude Opus 572.5%SourceClaude Opus 5.572.0%SourceClaude Opus 5 leads this result
SingleCellBench
Claude Opus 560.6%SourceClaude Opus 5.561.2%SourceClaude Opus 5.5 leads this result
ProteinGym Hard
Claude Opus 547.7%SourceClaude Opus 5.5—Not directly comparable
Protein Design
Claude Opus 542.5%SourceClaude Opus 5.560.2%SourceClaude Opus 5.5 leads this result
Organic chemistry V2
Claude Opus 561.6%SourceClaude Opus 5.5—Not directly comparable
Protocols (troubleshooting)
Claude Opus 561.1%SourceClaude Opus 5.573.7%SourceClaude Opus 5.5 leads this result
Protocols (understanding)
Claude Opus 578.4%SourceClaude Opus 5.569.0%SourceClaude Opus 5 leads this result
GPQA Diamond (Vals)
Claude Opus 593.4%SourceClaude Opus 5.5—Not directly comparable
MMLU-Pro (Vals)
Claude Opus 591.6%SourceClaude Opus 5.5—Not directly comparable
Morphology-to-molecule matching
Claude Opus 5—Claude Opus 5.534.0%SourceNot directly comparable
Medicinal chemistry
Claude Opus 5—Claude Opus 5.563.5%SourceNot directly comparable
Protein Design library ranking
Claude Opus 5—Claude Opus 5.556.0%SourceNot directly comparable
De novo protein-binder design
Claude Opus 5—Claude Opus 5.582.6%SourceNot directly comparable
Multilingual
Math
IMO 2026
Claude Opus 542/42SourceClaude Opus 5.5—Not directly comparable
RiemannBench (no tools)
Claude Opus 560.0%SourceClaude Opus 5.5—Not directly comparable
RiemannBench (tools)
Claude Opus 579.0%SourceClaude Opus 5.5—Not directly comparable
ArXivMath Jun. 2026 (no tools)
Claude Opus 590.8%SourceClaude Opus 5.5—Not directly comparable
ArXivMath Jun. 2026 (tools)
Claude Opus 591.3%SourceClaude Opus 5.5—Not directly comparable
ArXivMath Aug. 2026 (no tools)
Claude Opus 5—Claude Opus 5.591.2%SourceNot directly comparable
ArXivMath Aug. 2026 (tools)
Claude Opus 5—Claude Opus 5.596.9%SourceNot directly comparable
Questions
Which is better, Claude Opus 5 or Claude Opus 5.5?
Claude Opus 5.5 has the higher public score estimate, 81.03 versus 80.93, but the 90% score intervals overlap. The higher estimate is not a decisive winner because the uncertainty ranges overlap.
Which is better for coding, Claude Opus 5 or Claude Opus 5.5?
Claude Opus 5.5 scores higher for coding on the public lane, 76 to 75.9. Claude Opus 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 Opus 5 or Claude Opus 5.5?
Claude Opus 5.5 leads the public agentic tasks lane, 82.7 to 78.8, with Supported evidence for both models, although the 90% intervals overlap.
Which costs less, Claude Opus 5 or Claude Opus 5.5?
For the stated presets, chat costs $0.0175 on Claude Opus 5 and $0.014 on Claude Opus 5.5; repository review costs $0.325 and $0.26; the cache-heavy agent loop costs $0.45 and $0.32. Costs use the listed standard API rates.
Which has the larger context window, Claude Opus 5 or Claude Opus 5.5?
Both models list the same context window, 1M.
Related comparisons
Last updated September 22, 2026
Watch Claude Opus 5 vs Claude Opus 5.5
One weekly email when material rank, price, or benchmark evidence changes make this matchup worth revisiting.
Read a sample issueJoin 2,000+ readers.