Noometry

Model comparison

Claude Opus 4.8 vs Claude Sonnet 4.5

Claude Opus 4.8 is the stronger model overall, scoring 60.7 to 44.1 on the Noometry Index. Claude Sonnet 4.5 costs 1.7× less per token, which makes it the better buy when Claude Opus 4.8's lead doesn't matter for your workload.

Last verified . 51 shared benchmarks.

Claude Opus 4.8 Anthropic

60.7

Rank #13 Confirmed

Claude Sonnet 4.5 Anthropic

44.1

Rank #81 Confirmed

Summary

  • They share 51 benchmarks with published results for both. Claude Opus 4.8 scores higher in 10 categories and Claude Sonnet 4.5 in 0 categories; 9 gaps are clear of the uncertainty.
  • The widest gap is in math, where Claude Opus 4.8 leads 78.4 to 32.3.
  • The biggest single-benchmark swing is ARC-AGI-2: 72.1% for Claude Opus 4.8 and 13.6% for Claude Sonnet 4.5.
  • Claude Sonnet 4.5 is cheaper at $3 / $15 per million input/output tokens, against $5 / $25 for Claude Opus 4.8.
  • Claude Opus 4.8 accepts more context: 1M tokens versus 200K.

Side by side

Claude Opus 4.8Claude Sonnet 4.5
ProviderAnthropicAnthropic
Noometry Index60.744.1
Released2026-05-282025-09-29
WeightsProprietaryProprietary
Context window1M200K
Max output128K64K
Input $ / M tokens$5$3
Output $ / M tokens$25$15
Results tracked6573

Sponsored placements are available on pages like this one. Advertise on Noometry

Category by category

Coding Claude Opus 4.8 leads

Claude Opus 4.8: 59.9 (#12), Claude Sonnet 4.5: 47.3 (#61)

BenchmarkClaude Opus 4.8Claude Sonnet 4.5
LMArena WebDev15561393
SciCode53.5%44.7%
GSO47.1%14.7%
WeirdML82.9%47.7%
LMArena Coding14901489
ALE-Bench1,564796.15
SWE-bench Verified—71.3%
DeepSWE59%—
FrontierCode46.5%—
SWE-bench Verified (bash only)—71.4%
SWE-bench Multilingual—67%
AlgoTune—1.52

Agentic & Tool Use Claude Opus 4.8 leads

Claude Opus 4.8: 47.6 (#11), Claude Sonnet 4.5: 38.3 (#32)

BenchmarkClaude Opus 4.8Claude Sonnet 4.5
Remote Labor Index8.3%2.1%
τ²-bench Banking39.7%25.3%
DeepResearch Bench50.2%52.6%
LMArena Search12041159
Vending-Bench 25,7873,839
Terminal-Bench—46.5%
APEX-Agents48.9%—
Berkeley Function Calling Leaderboard—73.2%
OSWorld 2.020.6%—
GDPval—42.5%
τ²-bench Airline—72%
τ²-bench Retail—72.4%
τ²-bench Telecom—84.9%
Cybench—60%
OSWorld—62.9%
PostTrainBench33.8%—
GBAEval70.9%—
GDP.pdf24%—
METR Time Horizons—67.4%

Reasoning Claude Opus 4.8 leads

Claude Opus 4.8: 64.7 (#16), Claude Sonnet 4.5: 26.9 (#125)

BenchmarkClaude Opus 4.8Claude Sonnet 4.5
ARC-AGI-272.1%13.6%
SimpleBench64.8%54.3%
Kagi LLM Benchmark88.8%57.9%
NYT Connections (extended)91.1%37.3%
ARC-AGI-192.5%63.7%
CritPt20.9%1.1%
Chess Puzzles34%12%
EnigmaEval23.5%6%
EBR-Bench28.6%2.4%
LMArena Hard Prompts14821462
Mystery Game Puzzles36%17%
DTBench94.9%83.2%
LMCA57.5%38.8%
Epoch Capabilities Index158.21146.84
ForecastBench59.961.9
Surface Evolver Bench87.5%—
Bench to the Future 30.14—

Math Claude Opus 4.8 leads

Claude Opus 4.8: 78.4 (#13), Claude Sonnet 4.5: 32.3 (#216)

BenchmarkClaude Opus 4.8Claude Sonnet 4.5
FrontierMath (Tiers 1-3)80%23.9%
FrontierMath Tier 456.1%2.4%
OTIS Mock AIME 2024-202598.3%77.8%
ProofBench69%19%
LMArena Math14871449
FrontierMath (Feb 2025 set)47.2%15.2%
FrontierMath Tier 4 (v1)31.3%4.2%
MathArena Final-Answer Competitions91.8%—
Omni-MATH—55.3%
MATH Level 5—97.7%

Knowledge Claude Opus 4.8 leads

Claude Opus 4.8: 61.3 (#29), Claude Sonnet 4.5: 48.4 (#76)

BenchmarkClaude Opus 4.8Claude Sonnet 4.5
GPQA Diamond91%82.3%
SimpleQA Verified53%30.7%
LMArena Expert15021482
Humanity's Last Exam—13.7%
MMLU-Pro—86.9%
Vectara Hallucination Rate—12%
GPQA (HELM)—68.6%

Multimodal Claude Opus 4.8 leads

Claude Opus 4.8: 42.9 (#26), Claude Sonnet 4.5: 34.8 (#89)

BenchmarkClaude Opus 4.8Claude Sonnet 4.5
LMArena Document14751450
LMArena Vision1294—
VPCT—39.8%
Blueprint-Bench 214.5%—
Furniture Assembly42.5%—

Multilingual Claude Opus 4.8 leads

Claude Opus 4.8: 55.2 (#33), Claude Sonnet 4.5: 53.4 (#69)

BenchmarkClaude Opus 4.8Claude Sonnet 4.5
LMArena Non-English14501425
LMArena Chinese15071459
LMArena French14811458
LMArena German14721427
LMArena Japanese14401390
LMArena Korean14321403
LMArena Russian14741437
LMArena Spanish14661457

Instruction Following Claude Opus 4.8 leads

Claude Opus 4.8: 77.4 (#24), Claude Sonnet 4.5: 75.0 (#78)

BenchmarkClaude Opus 4.8Claude Sonnet 4.5
LMArena Instruction Following14761459
IFEval—85%

Long Context Too close to call

Claude Opus 4.8: 45.4 (#35), Claude Sonnet 4.5: 45.2 (#46)

BenchmarkClaude Opus 4.8Claude Sonnet 4.5
LMArena Longer Query14831476

Writing & Preference Claude Opus 4.8 leads

Claude Opus 4.8: 72.0 (#16), Claude Sonnet 4.5: 66.5 (#34)

BenchmarkClaude Opus 4.8Claude Sonnet 4.5
LMArena Text14611439
LMArena Creative Writing14541442
EQ-Bench Creative Writing18401678
LMArena Multi-Turn14761465
WildBench—85.4%
EQ-Bench 41281—

Frequently asked questions

Is Claude Opus 4.8 better than Claude Sonnet 4.5?

Claude Opus 4.8 is the stronger model overall, scoring 60.7 to 44.1 on the Noometry Index. Claude Sonnet 4.5 costs 1.7× less per token, which makes it the better buy when Claude Opus 4.8's lead doesn't matter for your workload.

Which is cheaper, Claude Opus 4.8 or Claude Sonnet 4.5?

Claude Sonnet 4.5 is cheaper. It lists at $3 per million input tokens and $15 per million output tokens; Claude Opus 4.8 lists at $5 and $25.

Is Claude Opus 4.8 or Claude Sonnet 4.5 better for coding?

Claude Opus 4.8 scores higher on coding benchmarks: 59.9 versus 47.3 in the Noometry coding category.

Which has the bigger context window?

Claude Opus 4.8 does, with 1M tokens against 200K.

How many benchmarks do Claude Opus 4.8 and Claude Sonnet 4.5 share?

51 benchmarks have published results for both models. Claude Opus 4.8 has 65 scored results on Noometry and Claude Sonnet 4.5 has 73.

Related comparisons

Go deeper