llmboard.ai
Benchmarks
CompareRankings
llmboard.ai
Benchmarks
CompareRankings
HomeModelsClaude Opus 4.8

Anthropic model product

Claude Opus 4.8

Claude Opus 4.8 is Anthropic's upgrade to Opus 4.7 and its most capable general-access model at release, with improvements across software engineering, agentic tool use, reasoning, computer use, and knowledge-work benchmarks while shipping at the same price ($5/$25 per million input/output tokens). Performance gains include SWE-Bench Verified (88.6%), SWE-Bench Pro (69.2%), Terminal-Bench 2.1 (74.6%), GPQA Diamond (93.6%), USAMO 2026 (96.7%), Humanity's Last Exam with tools (57.9%), OSWorld-Verified (83.4%), BrowseComp (84.3% single-agent, 88.5% multi-agent), MCP-Atlas (82.2%), and GDPval-AA (1890 Elo). The alignment assessment reports honesty improvements with around a four-fold drop in letting flaws in self-written code pass unremarked, a 17-fold drop relative to Sonnet 4.6 on dishonest agentic code summaries, and broadly improved adherence to Claude's constitution. The model defaults to high effort and exposes new 'extra' (xhigh) and 'max' levels for harder problems. Launches alongside Claude Code dynamic workflows (parallel subagents that plan, execute, and verify codebase-scale migrations), effort control in claude.ai and Cowork, and a Messages API extension that accepts system entries inside the messages array so harnesses can update instructions mid-task without breaking the prompt cache. Fast mode runs at 2.5× speed at $10/$50 per million input/output tokens, three times cheaper than fast mode on previous models. Available across Claude products, the Claude API as `claude-opus-4-8`, Amazon Bedrock, Google Cloud Vertex AI, and Microsoft Foundry.

Updated Aug 10, 2026. Default version: Claude Opus 4.8

Compare
LLMBoard score82.9Claude Opus 4.8
Coverage80%6 benchmark families
Context window1MTokens
Official input price$5Anthropic API

On this page

  • Specification
  • Capability
  • Benchmarks
  • Arena
  • Pricing
  • Versions
  • About
  • Compare
  • Similar models
  • FAQ

Model specification

Structured fields from the published default version.

Version
Claude Opus 4.8
Released
May 28, 2026
Knowledge cutoff
Jan 1, 2026
Parameters
N/A
Context window
1M
Max output
128K
Inputs
image, pdf, text
Outputs
text
Open weights
No
License
Proprietary

Capability profile

This profile uses the latest version under this unique model that has a calculated LLMBoard score. Arena and price are excluded.

Claude Opus 4.8 category scores

Benchmark results

Published benchmark records for the scored version Claude Opus 4.8.

26 rows
Columns

Show columns

Include0.9131100.0%CAug 7, 2026
ScreenSpot Pro0.9124100.0%CAug 7, 2026
DeepSearchQA0.92885.7%CAug 7, 2026
SWE-bench Multilingual0.823497.0%CAug 7, 2026
SWE-Bench Multimodal0.42350.0%CAug 7, 2026
FrontierSWE0.831585.7%BAug 7, 2026
OfficeQA Pro0.73766.7%CAug 7, 2026
OSWorld-Verified0.832290.5%CAug 7, 2026
SWE-Bench Pro0.734495.3%CAug 7, 2026
SWE-Bench Verified0.9310498.1%CAug 7, 2026
CharXiv-R0.944793.5%CAug 7, 2026
CyberGym0.841170.0%CAug 7, 2026
Finance Agent v20.542688.0%BAug 7, 2026
FrontierCode 1.10.541578.6%BAug 7, 2026
Graphwalks parents >128k0.841882.3%CAug 7, 2026
Toolathlon0.643190.0%CAug 7, 2026
GPQA0.9523398.3%CAug 7, 2026
MCP Atlas0.853086.2%CAug 7, 2026
HealthBench Professional0.66937.5%CAug 7, 2026
Humanity's Last Exam0.669294.5%CAug 7, 2026
LiveBench0.863886.5%BAug 7, 2026
Finance Agent0.57814.3%CAug 7, 2026
Terminal-Bench 2.00.774987.5%CAug 7, 2026
DeepSWE 1.10.681961.1%BAug 7, 2026
Graphwalks BFS >128k0.7112252.4%CAug 7, 2026
BrowseComp0.8135879.0%CAug 7, 2026

Arena results

Preference and agent-evaluation signals from published Arena datasets.

100 rows
Columns

Show columns

text style controllonger query91500.417,313N/AAug 6, 2026
text style controlmulti turn91499.86,950N/AAug 6, 2026
document style controloverall101481.88,193N/AJul 30, 2026
search factualityoverall101199.348,610N/AJul 21, 2026
search style controloverall101195.648,610N/AJul 21, 2026
text style controlhard prompts english101507.812,069N/AAug 6, 2026
searchoverall111204.948,610N/AJul 21, 2026
text style controlindustry business and management and financial operations111490.07,592N/AAug 6, 2026
text style controlindustry legal and government121495.93,096N/AAug 6, 2026
text style controlexpert131515.43,976N/AAug 6, 2026
visionhomework131327.71,425N/AAug 6, 2026
vision style controlenglish131281.44,791N/AAug 6, 2026
vision style controlhomework131321.71,425N/AAug 6, 2026
text factualityjapanese141446.6342N/AAug 6, 2026
text style controlcoding141526.810,721N/AAug 6, 2026
webdevwebdev-react141541.66,622N/AAug 6, 2026
textlonger query151481.117,313N/AAug 6, 2026
text factualityfrench151494.61,179N/AAug 6, 2026
text factualityindustry mathematical151478.11,572N/AAug 6, 2026
text style controlfrench151505.41,455N/AAug 6, 2026
text style controlhard prompts151504.125,254N/AAug 6, 2026
documentoverall161468.98,193N/AJul 30, 2026
text factualitymulti turn161485.66,254N/AAug 6, 2026
text style controlindustry life and physical and social science161499.86,198N/AAug 6, 2026
text style controlindustry medicine and healthcare161496.02,768N/AAug 6, 2026
text style controlindustry software and it services161513.715,148N/AAug 6, 2026
text style controlinstruction following161476.113,038N/AAug 6, 2026
text style controlkorean161444.4615N/AAug 6, 2026
visionenglish161292.84,791N/AAug 6, 2026
vision style controlchinese161332.8566N/AAug 6, 2026
vision style controlhumor161270.9434N/AAug 6, 2026
webdevoverall161539.18,565N/AAug 6, 2026
webdevwebdev161539.18,565N/AAug 6, 2026
text factualitychinese171517.61,719N/AAug 6, 2026
text style controlchinese171527.62,021N/AAug 6, 2026
text factualitymath181471.11,477N/AAug 6, 2026
text factualityspanish181456.8909N/AAug 6, 2026
text style controlcreative writing181462.46,816N/AAug 6, 2026
text style controlindustry entertainment and sports and media181453.58,764N/AAug 6, 2026
text factualityexpert191506.13,411N/AAug 6, 2026
text factualitylonger query191488.217,268N/AAug 6, 2026
text style controlindustry writing and literature and language191467.79,714N/AAug 6, 2026
visioncreative writing vision191284.9641N/AAug 6, 2026
visionhumor191284.4434N/AAug 6, 2026
vision style controlcreative writing vision191272.3641N/AAug 6, 2026
text style controlrussian201486.13,801N/AAug 6, 2026
textexpert211491.63,976N/AAug 6, 2026
text style controlspanish211471.41,156N/AAug 6, 2026
visionchinese211343.3566N/AAug 6, 2026
vision style controloverall211280.411,571N/AAug 6, 2026
vision style controlocr211292.58,366N/AAug 6, 2026
textmulti turn221477.26,950N/AAug 6, 2026
text factualityindustry legal and government221491.92,619N/AAug 6, 2026
visionoverall221289.711,571N/AAug 6, 2026
visionocr221298.18,366N/AAug 6, 2026
text factualityindustry life and physical and social science231490.65,719N/AAug 6, 2026
text factualityhard prompts english241500.212,027N/AAug 6, 2026
visiondiagram241301.13,125N/AAug 6, 2026
text factualitycreative writing251449.16,227N/AAug 6, 2026
text factualityhard prompts251492.025,207N/AAug 6, 2026
text factualityindustry entertainment and sports and media251442.48,556N/AAug 6, 2026
text factualityindustry writing and literature and language251456.59,572N/AAug 6, 2026
text style controlenglish251480.917,552N/AAug 6, 2026
text style controlpolish251477.8694N/AAug 6, 2026
vision style controldiagram251303.93,125N/AAug 6, 2026
webdevwebdev-html251527.51,116N/AAug 6, 2026
textindustry business and management and financial operations261456.47,592N/AAug 6, 2026
textindustry legal and government261473.43,096N/AAug 6, 2026
text factualityindustry business and management and financial operations261470.27,107N/AAug 6, 2026
text factualityinstruction following261469.112,992N/AAug 6, 2026
text factualityrussian261473.13,308N/AAug 6, 2026
textcreative writing271445.46,816N/AAug 6, 2026
textfrench271476.91,455N/AAug 6, 2026
textinstruction following271460.013,038N/AAug 6, 2026
textrussian271465.53,801N/AAug 6, 2026
text style controlindustry mathematical271486.31,953N/AAug 6, 2026
textindustry mathematical281477.61,953N/AAug 6, 2026
textindustry writing and literature and language281453.09,714N/AAug 6, 2026
text factualitycoding281519.210,676N/AAug 6, 2026
text style controloverall281472.738,131N/AAug 6, 2026
text style controlexclude ties281482.628,780N/AAug 6, 2026
text style controlgerman281469.5661N/AAug 6, 2026
texthard prompts english291475.612,069N/AAug 6, 2026
text factualityindustry software and it services291502.115,107N/AAug 6, 2026
text style controlnon english291461.720,579N/AAug 6, 2026
texthard prompts301472.925,254N/AAug 6, 2026
textkorean301423.3615N/AAug 6, 2026
text factualityoverall301462.838,059N/AAug 6, 2026
text factualitynon english301452.520,539N/AAug 6, 2026
textindustry life and physical and social science311469.56,198N/AAug 6, 2026
text factualityexclude ties311465.628,727N/AAug 6, 2026
text style controlmath311475.21,850N/AAug 6, 2026
textchinese321507.82,021N/AAug 6, 2026
textcoding321484.310,721N/AAug 6, 2026
textenglish321458.317,552N/AAug 6, 2026
textindustry entertainment and sports and media321432.58,764N/AAug 6, 2026
textmath321467.91,850N/AAug 6, 2026
text style controljapanese321441.7422N/AAug 6, 2026
textspanish331456.01,156N/AAug 6, 2026
text factualityindustry medicine and healthcare331486.52,349N/AAug 6, 2026

Pricing

Official vendor API PAYG pricing is summarized first. The table then lists individual provider offerings without treating their minimum as the official price.

Official API
$5 input, $25 output per 1M
Official provider
Anthropic
Lowest third-party
From $0.425 input, $2.13 output per 1M via UnoRouter
Tracked offerings
19
18 rows
Columns

Show columns

Kenariclaude-opus-4-8globalN/AN/A1MAug 7, 2026
UnoRouterclaude-opus-4-8global$0.425$2.131MAug 7, 2026
Xpersonaclaude-opus-4-8global$1.5$9.25200KAug 7, 2026
DaoXEclaude-opus-4-8global$5$251MAug 7, 2026
LLM Gatewayclaude-opus-4-8global$5$251MAug 7, 2026
OpenCode Zenclaude-opus-4-8global$5$251MAug 7, 2026
Neonclaude-opus-4-8global$5$251MAug 7, 2026
Pioneerclaude-opus-4-8global$5$251MAug 7, 2026
Abacusclaude-opus-4-8global$5$251MAug 7, 2026
Azure Cognitive Servicesclaude-opus-4-8global$5$251MAug 7, 2026
Azureclaude-opus-4-8global$5$251MAug 7, 2026
AIHubMixclaude-opus-4-8global$5$25200KAug 7, 2026
routing.runclaude-opus-4-8global$5$251MAug 7, 2026
FreeModelclaude-opus-4-8global$5$251MAug 7, 2026
Requestyclaude-opus-4-8global$5$251MAug 7, 2026
Modelisclaude-opus-4-8global$5$251MAug 7, 2026
Anthropicclaude-opus-4-8global$5$251MAug 7, 2026
Venice AIclaude-opus-4-8global$6$301MAug 7, 2026

Official prices use only the vendor's configured official Provider and positive standard USD PAYG rates. Third-party offers remain explicitly labeled.

Claude Opus 4.8 versions

All published versions linked to this unique model. The score columns identify the version used by the current overall ranking.

1 rows
Columns

Show columns

Claude Opus 4.8May 28, 202682.9N/A1M128KNoProprietary

What is Claude Opus 4.8?

A concise description based on the published model registry.

Claude Opus 4.8 is Anthropic's upgrade to Opus 4.7 and its most capable general-access model at release, with improvements across software engineering, agentic tool use, reasoning, computer use, and knowledge-work benchmarks while shipping at the same price ($5/$25 per million input/output tokens). Performance gains include SWE-Bench Verified (88.6%), SWE-Bench Pro (69.2%), Terminal-Bench 2.1 (74.6%), GPQA Diamond (93.6%), USAMO 2026 (96.7%), Humanity's Last Exam with tools (57.9%), OSWorld-Verified (83.4%), BrowseComp (84.3% single-agent, 88.5% multi-agent), MCP-Atlas (82.2%), and GDPval-AA (1890 Elo). The alignment assessment reports honesty improvements with around a four-fold drop in letting flaws in self-written code pass unremarked, a 17-fold drop relative to Sonnet 4.6 on dishonest agentic code summaries, and broadly improved adherence to Claude's constitution. The model defaults to high effort and exposes new 'extra' (xhigh) and 'max' levels for harder problems. Launches alongside Claude Code dynamic workflows (parallel subagents that plan, execute, and verify codebase-scale migrations), effort control in claude.ai and Cowork, and a Messages API extension that accepts system entries inside the messages array so harnesses can update instructions mid-task without breaking the prompt cache. Fast mode runs at 2.5× speed at $10/$50 per million input/output tokens, three times cheaper than fast mode on previous models. Available across Claude products, the Claude API as `claude-opus-4-8`, Amazon Bedrock, Google Cloud Vertex AI, and Microsoft Foundry.

Use the benchmark, Arena and pricing sections above as separate evidence. A missing field means the current data snapshot does not support that claim.

Data snapshot: 2026-08-07. Editorial model content is not available in the backend.

Claude Opus 4.8 vs nearby models

Open a comparison with the three ranked models immediately above and below this model.

Claude Opus 4.8vsSeed 2.1 ProClaude Opus 4.8vsGPT-5.6-TerraClaude Opus 4.8vsGemini 3.1 ProClaude Opus 4.8vsKimi K2.6Claude Opus 4.8vsGemini 3 ProClaude Opus 4.8vsSeed 2.1 Turbo

Models similar to Claude Opus 4.8

Recommendations prioritize the same model type and family, then the closest published LLMBoard score.

#13-4.4
AN

Claude Opus 4.7

Anthropic

78.4 LLMBoard

DetailsCompare
#22-9.4
AN

Claude Sonnet 3.5

Anthropic

73.4 LLMBoard

DetailsCompare
#33-13.6
AN

Claude Opus 4.6

Anthropic

69.2 LLMBoard

DetailsCompare
#41-19.3
AN

Claude Sonnet 3.7

Anthropic

63.6 LLMBoard

DetailsCompare
#55-24.0
AN

Claude Opus 3

Anthropic

58.8 LLMBoard

DetailsCompare
#110-41.7
AN

Claude Haiku 3.5

Anthropic

41.2 LLMBoard

DetailsCompare

FAQ

Common questions about Claude Opus 4.8.

When was Claude Opus 4.8 released?

Claude Opus 4.8's default version was released on May 28, 2026.

How much does Claude Opus 4.8 cost?

Claude Opus 4.8's official API price is $5 per million input tokens and $25 per million output tokens via Anthropic. The lowest tracked third-party offer starts at $0.425 input and $2.13 output via UnoRouter.

Who created Claude Opus 4.8?

Claude Opus 4.8 is published under Anthropic in the model registry.

What is the context window for Claude Opus 4.8?

The default version has a 1M token context window.

Is Claude Opus 4.8 open weight?

No. The default version is not marked as having publicly available weights.

How many API providers offer Claude Opus 4.8?

19 published provider offerings are linked to the default version.

What models should I compare Claude Opus 4.8 with?

Nearby ranked alternatives include Seed 2.1 Pro, GPT-5.6-Terra, Gemini 3.1 Pro.

Rankings

OverallCodingText ArenaPricing

Modalities

Image GenerationVideo GenerationSpeech-to-TextEmbeddings

Benchmarks

All BenchmarksReasoningMathCoding

Vendors

All VendorsOpenAIAnthropicGoogle
llmboard.aiCopyright 2026 llmboard.ai