Head-to-head comparison

Claude Opus 5.5 vs. Claude Sonnet 5

Compare published benchmark results from matching versions and cohorts, API pricing, and technical specifications. Any documented test setup differences remain visible.

shared result series: 195
distinct benchmarks: 25
latest retrieval: 2026-10-15

First modelSecond model

Specifications

General specifications

Release date, availability, context window, architecture, and published parameter counts at a glance. Undisclosed details remain marked as unavailable.

Specifications for Claude Opus 5.5 and Claude Sonnet 5
Attribute	Claude Opus 5.5	Claude Sonnet 5
Provider	Anthropic	Anthropic
Released	Sep 22, 2026	Jun 30, 2026
Availability	Active	Active
Model type	Proprietary	Proprietary
Parameters	Not published	Not published
Architecture	Not published	Not published
Context window	1,000,000 tokens	1,000,000 tokens
Knowledge cutoff	June 2026	January 2026
Technical sources	Model	Model, Context

Pricing

API pricing compared

The table compares official standard API rates per 1M tokens, including published cache pricing. Batch, fast, priority, regional surcharges, tool calls, and cloud platform pricing are excluded.

API pricing for Claude Opus 5.5 and Claude Sonnet 5
Price type	Claude Opus 5.5	Claude Sonnet 5
API input	$4	$2
API output	$20	$10
Cache read	$0.2	$0.2
Cache write (5 min.)	$5	$2.5
Cache write (1 hr.)	$8	$4
Price verified	09/22/2026 Source	09/22/2026 Source

Claude Opus 5.5Claude Sonnet 5

API inputUSD per 1M tokens, base tier

Claude Opus 5.5$4

Claude Sonnet 5$2

API outputUSD per 1M tokens, base tier

Claude Opus 5.5$20

Claude Sonnet 5$10

Cache readUSD per 1M tokens, base tier

Claude Opus 5.5$0.2

Claude Sonnet 5$0.2

Key to the marks:better valuebetter, but not conclusivetie

Pricing by provider

Direct developer pricing and provider endpoint tariffs routed through OpenRouter are listed separately. Prices are in USD per 1M tokens. Context-dependent direct tiers plus regional, Flex, Priority, and other OpenRouter tariffs remain individually identifiable.

Provider pricing for Claude Opus 5.5 and Claude Sonnet 5
Model	Provider	Purchase route and tariff	Input per 1M tokens	Output per 1M tokens	Source
Claude Opus 5.5	Anthropic	Direct from developerStandard API	Lowest listed price: $4	Lowest listed price: $20	Source09/22/2026
	Amazon Bedrock	Via OpenRouteramazon-bedrock	Lowest listed price: $4	Lowest listed price: $20	Source10/17/2026
	Amazon Bedrock	Via OpenRouteramazon-bedrock/eu-west-1	$4.4	$22	Source10/17/2026
	Amazon Bedrock	Via OpenRouteramazon-bedrock/us-east-1	$4.4	$22	Source10/17/2026
	Anthropic	Via OpenRouteranthropic	Lowest listed price: $4	Lowest listed price: $20	Source10/17/2026
	Anthropic	Via OpenRouteranthropic/fast	$8	$40	Source10/17/2026
	Azure	Via OpenRouterazure/global	Lowest listed price: $4	Lowest listed price: $20	Source10/17/2026
	Azure	Via OpenRouterazure/us	$4.4	$22	Source10/17/2026
	Claude Platform on AWS	Via OpenRouterclaude-on-aws	Lowest listed price: $4	Lowest listed price: $20	Source10/17/2026
	Google	Via OpenRoutergoogle-vertex/europe	$4.4	$22	Source10/17/2026
	Google	Via OpenRoutergoogle-vertex/global	Lowest listed price: $4	Lowest listed price: $20	Source10/17/2026
	Google	Via OpenRoutergoogle-vertex/us	$4.4	$22	Source10/17/2026
Claude Sonnet 5	Anthropic	Direct from developerStandard API	Lowest listed price: $2	Lowest listed price: $10	Source09/22/2026
	Amazon Bedrock	Via OpenRouteramazon-bedrock/eu-west-1	$2.2	$11	Source10/17/2026
	Amazon Bedrock	Via OpenRouteramazon-bedrock/global	Lowest listed price: $2	Lowest listed price: $10	Source10/17/2026
	Amazon Bedrock	Via OpenRouteramazon-bedrock/us-east-1	$2.2	$11	Source10/17/2026
	Anthropic	Via OpenRouteranthropic	Lowest listed price: $2	Lowest listed price: $10	Source10/17/2026
	Azure	Via OpenRouterazure/global	Lowest listed price: $2	Lowest listed price: $10	Source10/17/2026
	Azure	Via OpenRouterazure/us	$2.2	$11	Source10/17/2026
	Claude Platform on AWS	Via OpenRouterclaude-on-aws	Lowest listed price: $2	Lowest listed price: $10	Source10/17/2026
	Google	Via OpenRoutergoogle-vertex/europe	$2.2	$11	Source10/17/2026
	Google	Via OpenRoutergoogle-vertex/global	Lowest listed price: $2	Lowest listed price: $10	Source10/17/2026
	Google	Via OpenRoutergoogle-vertex/us	$2.2	$11	Source10/17/2026

Performance

Shared benchmarks

Only published results with the same benchmark version, task, metric, and comparison cohort are paired. Different reasoning levels, agents, harnesses, or output limits appear directly in the table.

Capability profile from matched benchmarks

Each axis averages directly matched benchmark families. A value of 100 means the stronger value within this pair, not a universal quality score.

Claude Opus 5.5

Claude Sonnet 5

Coding: 99.7 / 59.5 (6 benchmark families)

Reasoning: 100 / 97.7 (1 benchmark family)

Mathematics: 100 / 87.7 (2 benchmark families)

Knowledge: 100 / 97.3 (1 benchmark family)

Finance: 99.8 / 88 (4 benchmark families)

Legal: 95.3 / 96.3 (3 benchmark families)

Agentic tasks: 100 / 19.2 (1 benchmark family)

Multimodal: 100 / 97.4 (1 benchmark family)

183 matched rows from 20 benchmark families were considered. The radar shows 8 of 14 comparable categories.

Claude Opus 5.5Claude Sonnet 5

Code MigrationOverall. Higher is better. No documented setup difference

Claude Opus 5.566.646 %

Claude Sonnet 544.392 %

CyberBench v1.1Overall. Higher is better. No documented setup difference

Claude Opus 5.555.357 %

Claude Sonnet 561.905 %

EMBOverall. Higher is better. No documented setup difference

Claude Opus 5.575.938 %

Claude Sonnet 566.316 %

Finance Agent (v2)Overall. Higher is better. No documented setup difference

Claude Opus 5.558.587 %

Claude Sonnet 553.909 %

Harvey's Legal Agent BenchmarkOverall · Task fully resolved. Higher is better. No documented setup difference

Claude Opus 5.53.75 %

Claude Sonnet 55 %

IOIOverall. Higher is better. No documented setup difference

Claude Opus 5.595.056 %

Claude Sonnet 545 %

Legal Research BenchOverall · All-pass. Higher is better. No documented setup difference

Claude Opus 5.550.481 %

Claude Sonnet 541.827 %

LMArena Code Arena WebDevOverall. Higher is better. Test setup not documented as identical

Claude Opus 5.51,811.149

Claude Sonnet 51,539.919

LMArena Text, Style ControlOverall. Higher is better. No documented setup difference

Claude Opus 5.51,508.083

Claude Sonnet 51,461.284

LMArena Vision, Style ControlOverall. Higher is better. No documented setup difference

Claude Opus 5.51,292.688

Claude Sonnet 51,263.255

MedCodeOverall. Higher is better. Test setup not documented as identical

Claude Opus 5.549.797 %

Claude Sonnet 547.54 %

MedScribeOverall. Higher is better. Test setup not documented as identical

Claude Opus 5.591.43 %

Claude Sonnet 576.054 %

Key to the marks:better valuebetter, but not conclusivetie

Benchmark scores for Claude Opus 5.5 and Claude Sonnet 5
Benchmark and task	Claude Opus 5.5	Claude Sonnet 5	Source
Code MigrationOverall Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	Winner66.646 %	44.392 %	Vals AI Source details Data as of: 10/12/2026
CyberBench v1.1Overall Test details No documented difference in the test setupVersion: 1.1Metric: accuracyScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	55.357 %	Winner61.905 %	Vals AI Source details Data as of: 10/12/2026
EMBOverall Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	Winner75.938 %	66.316 %	Vals AI Source details Data as of: 10/12/2026
Finance Agent (v2)Overall Test details No documented difference in the test setupVersion: 2Metric: accuracyScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	Winner58.587 %	53.909 %	Vals AI Source details Data as of: 10/12/2026
Harvey's Legal Agent BenchmarkOverall · Task fully resolved Test details No documented difference in the test setupVersion: 1Metric: task resolution rateScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	3.75 %	Winner5 %	Vals AI Source details Data as of: 10/12/2026
IOIOverall Test details No documented difference in the test setupVersion: 2Metric: accuracyScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	Winner95.056 %	45 %	Vals AI Source details Data as of: 10/14/2026
Legal Research BenchOverall · All-pass Test details No documented difference in the test setupVersion: 1Metric: all-pass rateScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	Winner50.481 %	41.827 %	Vals AI Source details Data as of: 10/12/2026
LMArena Code Arena WebDevOverall Test details Test setup not documented as identicalNo direct winnerVersion: webdevMetric: WebDev Arena Bradley-Terry ratingScoring: Higher is betterClaude Opus 5.5: Snapshot: claude-opus-5.5-max, Reasoning: max, Harness: Arena Code Arena WebDev Claude Sonnet 5: Snapshot: claude-sonnet-5-high, Reasoning: high, Harness: Arena Code Arena WebDevClaude Opus 5.5: Arena rating from 2,597 published pairwise votes.Claude Sonnet 5: Arena rating from 15,175 published pairwise votes.	Higher measured value1,811.149 95% confidence interval: ±14.586The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	1,539.919 95% confidence interval: ±6.061The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	LMArena Source details Data as of: 10/13/2026
LMArena Text, Style ControlOverall Test details No documented difference in the test setupVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterClaude Opus 5.5: Snapshot: claude-opus-5.5-high, Reasoning: high, Harness: Arena Text, style control Claude Sonnet 5: Snapshot: claude-sonnet-5-high, Reasoning: high, Harness: Arena Text, style controlClaude Opus 5.5: Arena rating from 7,687 published pairwise votes.Claude Sonnet 5: Arena rating from 48,998 published pairwise votes.	Winner1,508.083 95% confidence interval: ±7.535The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	1,461.284 95% confidence interval: ±4.141The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	LMArena Source details Data as of: 10/13/2026
LMArena Vision, Style ControlOverall Test details No documented difference in the test setupVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterClaude Opus 5.5: Snapshot: claude-opus-5.5-high, Reasoning: high, Harness: Arena Vision, style control Claude Sonnet 5: Snapshot: claude-sonnet-5-high, Reasoning: high, Harness: Arena Vision, style controlClaude Opus 5.5: Arena rating from 2,508 published pairwise votes.Claude Sonnet 5: Arena rating from 13,503 published pairwise votes.	Winner1,292.688 95% confidence interval: ±12.73The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	1,263.255 95% confidence interval: ±6.921The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	LMArena Source details Data as of: 10/09/2026
MedCodeOverall Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterClaude Opus 5.5: Compute: max, Temperature: 1, Output limit: 128,000 tokens Claude Sonnet 5: Compute: max, Temperature: 1, Output limit: 30,000 tokens	Higher measured value49.797 %	47.54 %	Vals AI Source details Data as of: 10/13/2026
MedScribeOverall Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterClaude Opus 5.5: Compute: max, Temperature: 1, Output limit: 128,000 tokens Claude Sonnet 5: Compute: max, Temperature: 1, Output limit: 30,000 tokens	Higher measured value91.43 %	76.054 %	Vals AI Source details Data as of: 10/13/2026
ProgramBenchOverall Test details No documented difference in the test setupVersion: 1Metric: fully resolved rateScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	Winner18.5 %	0 %	Vals AI Source details Data as of: 10/12/2026
ProofBenchOverall Test details No documented difference in the test setupVersion: 1.1Metric: accuracyScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	Winner100 %	77 %	Vals AI Source details Data as of: 10/12/2026
Public Benefits Bench v1.1Overall Test details No documented difference in the test setupVersion: 1.1Metric: accuracyScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	Winner70.636 %	66.035 %	Vals AI Source details Data as of: 10/12/2026
SAGEOverall Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 30,000 tokens	45.831 %	Winner48.925 %	Vals AI Source details Data as of: 10/08/2026
SRE BenchOverall Test details No documented difference in the test setupVersion: 1Metric: fully solved rateScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	Winner33.588 %	0.382 %	Vals AI Source details Data as of: 10/14/2026
Tax Agent BenchOverall Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	Winner45.573 %	29.374 %	Vals AI Source details Data as of: 10/12/2026
Terminal-Bench 2.1Overall Test details Test setup not documented as identicalNo direct winnerVersion: 2.1Metric: accuracyScoring: Higher is betterClaude Opus 5.5: Compute: high, Temperature: 1, Output limit: 128,000 tokens Claude Sonnet 5: Compute: max, Temperature: 1, Output limit: 128,000 tokens	Higher measured value87.64 %	74.532 %	Vals AI Source details Data as of: 09/28/2026
Terminal-Bench 4.0Overall Test details No documented difference in the test setupVersion: 4.0Metric: accuracyScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	Winner65.152 %	10.606 %	Vals AI Source details Data as of: 10/13/2026
Terminal-Bench 4.0Overall Test details No documented difference in the test setupVersion: 4.0Metric: trial pass rateScoring: Higher is betterSettings: Reasoning: max, Agent: Claude Code, Attempts: 330 graded trials330 graded trials, Claude Code agent, max reasoning. Dataset version 1922072f-a433-429a-8929-350d5e1bcf02.	Winner64.85 % 95% confidence interval: ±3.11Selection of exactly identified models from the official 35-row leaderboard. Agents and reasoning efforts differ. These values do not belong to the separate Vals Mini-SWE-agent cohort.	12.42 % Selection of exactly identified models from the official 35-row leaderboard. Agents and reasoning efforts differ. These values do not belong to the separate Vals Mini-SWE-agent cohort.	Terminal-Bench Source details Data as of: 10/06/2026
Terminal-Bench ScienceOverall Test details No documented difference in the test setupVersion: 0.1Metric: accuracyScoring: Higher is betterSettings: Compute: max, Harness: Mini-SWE-agent, Temperature: 1, Output limit: 128,000 tokens	Winner47.143 %	5.714 %	Vals AI Source details Data as of: 10/14/2026
Vals IndexOverall Test details No documented difference in the test setupVersion: 2.1Metric: weighted index scoreScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	Winner66.972 %	51.898 %	Vals AI Source details Data as of: 10/13/2026
Vibe Code Bench 1-100Overall Test details Test setup not documented as identicalNo direct winnerVersion: 1.0Metric: accuracyScoring: Higher is betterClaude Opus 5.5: Compute: xhigh, Harness: OpenHands, Temperature: 1, Output limit: 128,000 tokens Claude Sonnet 5: Compute: max, Harness: OpenHands, Temperature: 1, Output limit: 128,000 tokens	Higher measured value30.361 %	13.822 %	Vals AI Source details Data as of: 10/14/2026
Vibe Code Bench v1.1Overall Test details No documented difference in the test setupVersion: 1.1Metric: accuracyScoring: Higher is betterSettings: Compute: max, Harness: OpenHands, Temperature: 1, Output limit: 128,000 tokens	Winner90.294 %	81.325 %	Vals AI Source details Data as of: 10/12/2026
Code MigrationCLI Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	Winner65.528 %	36.332 %	Vals AI Source details Data as of: 10/12/2026
CyberBench v1.1Patch Test details No documented difference in the test setupVersion: 1.1Metric: accuracyScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	Winner85.714 %	82.143 %	Vals AI Source details Data as of: 10/12/2026
EMBComps Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	Winner76.405 %	69.384 %	Vals AI Source details Data as of: 10/12/2026
Finance Agent (v2)Adjustments Test details No documented difference in the test setupVersion: 2Metric: accuracyScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	Winner54.528 %	48.715 %	Vals AI Source details Data as of: 10/12/2026
Harvey's Legal Agent BenchmarkAntitrust Competition · Criteria passed Test details No documented difference in the test setupVersion: 1Metric: criteria pass rateScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	82.246 %	Winner88.768 %	Vals AI Source details Data as of: 10/12/2026
IOIIOI 2024 Test details No documented difference in the test setupVersion: 2Metric: accuracyScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	Winner100 %	50.5 %	Vals AI Source details Data as of: 10/14/2026
Legal Research BenchAdministrative / Regulatory · All-pass Test details No documented difference in the test setupVersion: 1Metric: all-pass rateScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	Winner65.152 %	54.545 %	Vals AI Source details Data as of: 10/12/2026
LMArena Code Arena WebDevImage to webdev Test details Test setup not documented as identicalNo direct winnerVersion: webdevMetric: WebDev Arena Bradley-Terry ratingScoring: Higher is betterClaude Opus 5.5: Snapshot: claude-opus-5.5-max, Reasoning: max, Harness: Arena Code Arena WebDev Claude Sonnet 5: Snapshot: claude-sonnet-5-high, Reasoning: high, Harness: Arena Code Arena WebDevClaude Opus 5.5: Arena rating from 493 published pairwise votes.Claude Sonnet 5: Arena rating from 7,633 published pairwise votes.	Higher measured value1,748.511 95% confidence interval: ±30.162The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	1,539.169 95% confidence interval: ±7.844The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	LMArena Source details Data as of: 10/09/2026
LMArena Text, Style ControlBusiness, management and finance Test details No documented difference in the test setupVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterClaude Opus 5.5: Snapshot: claude-opus-5.5-high, Reasoning: high, Harness: Arena Text, style control Claude Sonnet 5: Snapshot: claude-sonnet-5-high, Reasoning: high, Harness: Arena Text, style controlClaude Opus 5.5: Arena rating from 1,370 published pairwise votes.Claude Sonnet 5: Arena rating from 9,443 published pairwise votes.	Winner1,492.054 95% confidence interval: ±16.446The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	1,464.402 95% confidence interval: ±6.971The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	LMArena Source details Data as of: 10/13/2026
LMArena Vision, Style ControlCreative writing vision Test details No documented difference in the test setupVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterClaude Opus 5.5: Snapshot: claude-opus-5.5-high, Reasoning: high, Harness: Arena Vision, style control Claude Sonnet 5: Snapshot: claude-sonnet-5-high, Reasoning: high, Harness: Arena Vision, style controlClaude Opus 5.5: Arena rating from 219 published pairwise votes.Claude Sonnet 5: Arena rating from 1,067 published pairwise votes.	Winner1,301.035 95% confidence interval: ±42.823The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	1,236.247 95% confidence interval: ±19.371The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	LMArena Source details Data as of: 10/09/2026
ProgramBenchAlmost Resolved Test details No documented difference in the test setupVersion: 1Metric: almost-resolved rateScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	Winner65 %	13.5 %	Vals AI Source details Data as of: 10/12/2026
SAGEAutomata Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 30,000 tokens	32.54 %	Winner34.92 %	Vals AI Source details Data as of: 10/08/2026
SRE BenchCapability Score Test details No documented difference in the test setupVersion: 1Metric: capability scoreScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	Winner53.562 %	7.697 %	Vals AI Source details Data as of: 10/14/2026
Tax Agent BenchControversy & Precedence Analysis Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	Winner30 %	10 %	Vals AI Source details Data as of: 10/12/2026
Terminal-Bench 2.1Easy Test details Test setup not documented as identicalNo direct winnerVersion: 2.1Metric: accuracyScoring: Higher is betterClaude Opus 5.5: Compute: high, Temperature: 1, Output limit: 128,000 tokens Claude Sonnet 5: Compute: max, Temperature: 1, Output limit: 128,000 tokens	No direct winner100 %	No direct winner100 %	Vals AI Source details Data as of: 09/28/2026
Terminal-Bench 4.0Hardware Test details No documented difference in the test setupVersion: 4.0Metric: accuracyScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	Winner60 %	0 %	Vals AI Source details Data as of: 10/13/2026
Terminal-Bench ScienceEarth Sciences Test details No documented difference in the test setupVersion: 0.1Metric: accuracyScoring: Higher is betterSettings: Compute: max, Harness: Mini-SWE-agent, Temperature: 1, Output limit: 128,000 tokens	Winner37.5 %	0 %	Vals AI Source details Data as of: 10/14/2026
Vals IndexCode Migration Test details No documented difference in the test setupVersion: 2.1Metric: benchmark scoreScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	Winner68.901 %	44.089 %	Vals AI Source details Data as of: 10/13/2026
Code MigrationCOBOL Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	Winner70 %	68.571 %	Vals AI Source details Data as of: 10/12/2026
CyberBench v1.1PoC Test details No documented difference in the test setupVersion: 1.1Metric: accuracyScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	25 %	Winner41.667 %	Vals AI Source details Data as of: 10/12/2026
EMBDataroom Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	84.873 %	Winner84.909 %	Vals AI Source details Data as of: 10/12/2026
Finance Agent (v2)All-Pass Test details No documented difference in the test setupVersion: 2Metric: accuracyScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	Winner48.111 %	42.084 %	Vals AI Source details Data as of: 10/12/2026
Harvey's Legal Agent BenchmarkAntitrust Competition · Task fully resolved Test details No documented difference in the test setupVersion: 1Metric: task resolution rateScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	Tie0 %	Tie0 %	Vals AI Source details Data as of: 10/12/2026
IOIIOI 2025 Test details No documented difference in the test setupVersion: 2Metric: accuracyScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	Winner100 %	42 %	Vals AI Source details Data as of: 10/14/2026
Legal Research BenchAdministrative / Regulatory · Weighted partial credit Test details No documented difference in the test setupVersion: 1Metric: weighted partial-credit scoreScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	85.427 %	Winner87.81 %	Vals AI Source details Data as of: 10/12/2026
LMArena Code Arena WebDevWebdev Test details Test setup not documented as identicalNo direct winnerVersion: webdevMetric: WebDev Arena Bradley-Terry ratingScoring: Higher is betterClaude Opus 5.5: Snapshot: claude-opus-5.5-max, Reasoning: max, Harness: Arena Code Arena WebDev Claude Sonnet 5: Snapshot: claude-sonnet-5-high, Reasoning: high, Harness: Arena Code Arena WebDevClaude Opus 5.5: Arena rating from 2,597 published pairwise votes.Claude Sonnet 5: Arena rating from 15,175 published pairwise votes.	Higher measured value1,811.149 95% confidence interval: ±14.586The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	1,539.919 95% confidence interval: ±6.061The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	LMArena Source details Data as of: 10/13/2026
LMArena Text, Style ControlChinese Test details No documented difference in the test setupVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterClaude Opus 5.5: Snapshot: claude-opus-5.5-high, Reasoning: high, Harness: Arena Text, style control Claude Sonnet 5: Snapshot: claude-sonnet-5-high, Reasoning: high, Harness: Arena Text, style controlClaude Opus 5.5: Arena rating from 704 published pairwise votes.Claude Sonnet 5: Arena rating from 3,748 published pairwise votes.	Winner1,574.432 95% confidence interval: ±22.751The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	1,497.538 95% confidence interval: ±10.206The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	LMArena Source details Data as of: 10/13/2026
LMArena Vision, Style ControlDiagram Test details No documented difference in the test setupNo clear leadVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterClaude Opus 5.5: Snapshot: claude-opus-5.5-high, Reasoning: high, Harness: Arena Vision, style control Claude Sonnet 5: Snapshot: claude-sonnet-5-high, Reasoning: high, Harness: Arena Vision, style controlClaude Opus 5.5: Arena rating from 589 published pairwise votes.Claude Sonnet 5: Arena rating from 3,582 published pairwise votes.	No clear lead1,324.12 95% confidence interval: ±25.101The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	1,295.106 95% confidence interval: ±10.613The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	LMArena Source details Data as of: 10/09/2026
ProgramBenchFully Resolved Test details No documented difference in the test setupVersion: 1Metric: fully resolved rateScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	Winner18.5 %	0 %	Vals AI Source details Data as of: 10/12/2026
SAGELinear Algebra Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 30,000 tokens	66.859 %	Winner68.999 %	Vals AI Source details Data as of: 10/08/2026
SRE BenchFully Solved Test details No documented difference in the test setupVersion: 1Metric: fully solved rateScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	Winner33.588 %	0.382 %	Vals AI Source details Data as of: 10/14/2026
Tax Agent BenchControversy & Precedence Analysis Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	Winner66.871 %	57.41 %	Vals AI Source details Data as of: 10/12/2026
Terminal-Bench 2.1Hard Test details Test setup not documented as identicalNo direct winnerVersion: 2.1Metric: accuracyScoring: Higher is betterClaude Opus 5.5: Compute: high, Temperature: 1, Output limit: 128,000 tokens Claude Sonnet 5: Compute: max, Temperature: 1, Output limit: 128,000 tokens	Higher measured value81.111 %	57.778 %	Vals AI Source details Data as of: 09/28/2026
Terminal-Bench 4.0Media Test details No documented difference in the test setupVersion: 4.0Metric: accuracyScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	Winner75 %	8.333 %	Vals AI Source details Data as of: 10/13/2026
Terminal-Bench ScienceEngineering Sciences Test details No documented difference in the test setupVersion: 0.1Metric: accuracyScoring: Higher is betterSettings: Compute: max, Harness: Mini-SWE-agent, Temperature: 1, Output limit: 128,000 tokens	Winner22.222 %	0 %	Vals AI Source details Data as of: 10/14/2026
Vals IndexEMB Test details No documented difference in the test setupVersion: 2.1Metric: benchmark scoreScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	Winner75.938 %	66.316 %	Vals AI Source details Data as of: 10/13/2026
Code MigrationCode Quality Test details No documented difference in the test setupVersion: 1Metric: mean reviewer code-quality ratingScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	4.541 / 5	Winner4.898 / 5	Vals AI Source details Data as of: 10/12/2026
EMBDCF Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	Winner70.139 %	56.404 %	Vals AI Source details Data as of: 10/12/2026
Finance Agent (v2)Comparables Test details No documented difference in the test setupVersion: 2Metric: accuracyScoring: Higher is betterSettings: Compute: max, Temperature: 1, Output limit: 128,000 tokens	45.756 %	Winner45.817 %	Vals AI Source details Data as of: 10/12/2026
Key to the marks:better valuebetter, but not conclusivetie

Our model tests

Our tests

SVG short test

One SVG prompt and fixed output settings. Captures are shown only with 0 reported reasoning tokens or documented lack of reasoning controls. Identical test conditions do not make the models' internal compute identical.

Shared test round: 4ae5d8eeb9a9

Claude Opus 5.5

Unavailable in this round

Claude Sonnet 5

Unavailable in this round

Earlier tests

Pelican on a bicycle

A community classic for free-form SVG drawing.

Claude Opus 5.5

Reasoning (requested): high

Current attempt

Run date: Oct 1, 2026
API provider: openrouter
Requested API model: anthropic/claude-opus-5.5
API model reported in response: anthropic/claude-opus-5.5
Reasoning setting: high
Pinned endpoint: Anthropic | anthropic/claude-opus-5.5-20260921
Temperature: Provider default
Token limit including reasoning: 8,192 tokens
Test protocol: community-visual-tests-v1
Technical checks: Passed, not a quality rating
Response time: 33.2 s

Visually reviewed. This individual test does not establish overall model quality.

Claude Sonnet 5

Reasoning (requested): high

Current attempt

Run date: Sep 7, 2026
API provider: openrouter
Requested API model: anthropic/claude-sonnet-5
API model reported in response: anthropic/claude-sonnet-5
Reasoning setting: high
Pinned endpoint: Anthropic | anthropic/claude-sonnet-5-20260630
Temperature: Provider default
Token limit including reasoning: 8,192 tokens
Test protocol: community-visual-tests-v1
Technical checks: Passed, not a quality rating
Response time: 47.9 s

Visually reviewed. This individual test does not establish overall model quality.

Prompt and test conditions

Generate an SVG of a pelican riding a bicycle

Token limit including reasoning: 8,192 tokens

Task origin (Simon Willison)

How to read this comparison

Winning one benchmark is not an overall verdict

Your workload, budget, and required context length matter most. A coding benchmark says little about visual understanding or agent performance.

The standard price table uses direct developer rates. The provider comparison labels OpenRouter endpoints, regions, and special tariffs separately.

Every result links to its measurement source and retrieval date. Labels distinguish vendor reports, official benchmark leaderboards, and independent evaluations. Disputed or archived results are excluded.

More matchups

Compare Claude Opus 5.5 and Claude Sonnet 5 with other leading models using the same data and benchmark logic.

View the complete LLM comparison

Claude Opus 5.5 vs. Claude Sonnet 5

Change model selection

General specifications

API pricing compared

Shared benchmarks

Capability profile from matched benchmarks

Our tests

SVG short test

Claude Opus 5.5

Claude Sonnet 5

Pelican on a bicycle

Winning one benchmark is not an overall verdict

Related LLM comparisons