Enter opens the highlighted result. Esc closes.

Claude Opus 5 vs GPT-6 Astra

GPT-6 Astra leads Claude Opus 5 on coding, 85.1 to 83.0 on the Vibe Coding Index, and Claude Opus 5 costs 2.0x less per output token ($25.00 vs $50.00 per million). Updated October 22, 2026; 111 benchmarks have a result for both.

Replace Claude Opus 5 Replace GPT-6 Astra
Claude Opus 5Anthropic
Coding
83.0
Overall
81.6
Output $/M
$25.00
57
GPT-6 AstraOpenAI
Coding
85.1
Overall
83.0
Output $/M
$50.00
51
vs boards won of 111 shared, 3 tied

Claude Opus 5 or GPT-6 Astra: which to pick

Pick GPT-6 Astra for coding: it scores 85.1 on the Vibe Coding Index against 83.0 for Claude Opus 5, though on 5 benchmarks against 6.

Pick Claude Opus 5 for output tokens 2.0x cheaper ($25.00 vs $50.00 per million) and work measured by AA Coding Index or Vibe Code Bench 1-100 (Vals AI), where it leads.

Model AClaude Opus 5Anthropic83.0/1006 independent benchmarks$1.33 per solved taskModel BGPT-6 AstraOpenAI85.1/1005 independent benchmarksNo cost per solved task
Shared results
111
Claude Opus 5 only
20
GPT-6 Astra only
10
Like-for-like categories
8 of 9

Only benchmarks that tested both models count toward a pick; a result only one model has is shown but not compared.

Which one for your work

Each card is decided by one number both models carry. Where only one has it, the card says so and makes no call.

At a glance

Coding index
Claude Opus 5 83.0 from 6 benchmarks, GPT-6 Astra 85.1 from 5 benchmarks.
Widest category
Agentic work, 14.8 points apart on average over 22 shared benchmarks.
Widest single gap
Time Horizon Index (Vals AI): Claude Opus 5 18.8% against GPT-6 Astra 90.5%, 71.7 points apart on a 0 to 100 scale.
Output price
Claude Opus 5 $25.00 against GPT-6 Astra $50.00 per million output tokens, 2.0x apart.
Context window
Claude Opus 5 takes 1M tokens, GPT-6 Astra 1.05M.

Benchmark by benchmark

First the score: boards each model leads, per category, where both have a result. Then every board, a category to a fold; bars are each result on a 0 to 100 scale, so compare the two in a row, not one row with another.

Boards won, where both have a result
CategoryClaude Opus 5BarsGPT-6 AstraTies
Coding715
Intelligence02
Agentic work9112 ties
Reasoning69
Math231 tie
Knowledge51
Web design18
General272
All shared boards57513 ties
CodingLike-for-likeSix of these boards feed the Vibe Coding Index715
Coding benchmarks for Claude Opus 5 and GPT-6 Astra
BenchmarkClaude Opus 5GPT-6 AstraLeader
LiveBench coding90.8%93.7%GPT-6 Astra
LMArena WebDev16891774GPT-6 Astra
AA Coding Index78.076.9Claude Opus 5
LiveCodeBench (Vals AI)89.0%Not measuredOnly Claude Opus 5 measured
SWE-bench Verified (Vals AI)97.0%Not measuredOnly Claude Opus 5 measured
Terminal-Bench 4 (Vals AI)53.5%59.6%GPT-6 Astra
Vibe Code Bench (Vals AI)88.4%89.6%GPT-6 Astra
Code migration (Vals AI)57.5%67.7%GPT-6 Astra
ProgramBench (Vals AI)3.0%5.5%GPT-6 Astra
IOI (Vals AI)84.3%100.0%GPT-6 Astra
Terminal-Bench 2.1 (Vals AI)84.6%87.3%GPT-6 Astra
SWE-bench Multimodal (Vals AI)Not measured43.3%Only GPT-6 Astra measured
Vibe Code Bench 1-100 (Vals AI)28.5%27.6%Claude Opus 5
SkillsBench (Vals AI)60.4%Not measuredOnly Claude Opus 5 measured
CyberBench (Vals AI)65.4%41.1%Claude Opus 5
SciCode56.4%56.5%GPT-6 Astra
FrontierCode53.4%53.3%Claude Opus 5
FrontierSWE52.0%65.5%GPT-6 Astra
GSO-BenchNot measured79.4%Only GPT-6 Astra measured
WeirdML91.8%93.3%GPT-6 Astra
MirrorCodeNot measured46.7%Only GPT-6 Astra measured
ALE-Bench2164.62951.3GPT-6 Astra
LMArena Text: coding15331489Claude Opus 5
LMArena Text: software and IT15271474Claude Opus 5
LMArena WebDev: React17111798GPT-6 Astra
LMArena WebDev: HTML16801775GPT-6 Astra
LMArena WebDev: from an image16621726GPT-6 Astra
SciCode13.8%12.3%Claude Opus 5
Neither model has a result on the other 11 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index02
Intelligence benchmarks for Claude Opus 5 and GPT-6 Astra
BenchmarkClaude Opus 5GPT-6 AstraLeader
AA Intelligence Index50.852.7GPT-6 Astra
Epoch Capabilities Index163.9168.8GPT-6 Astra
Agentic workLike-for-likePublic boards, shown as published, kept out of the index911
Agentic work benchmarks for Claude Opus 5 and GPT-6 Astra
BenchmarkClaude Opus 5GPT-6 AstraLeader
AA Agentic Index56.551.0Claude Opus 5
Design Arena: agents building web apps12661373GPT-6 Astra
Design Arena: agents building full-stack apps13061357GPT-6 Astra
Design Arena: agents building mobile apps13481312Claude Opus 5
Design Arena: agents building Android apps1257Not measuredOnly Claude Opus 5 measured
Design Arena: agents building Godot games12791348GPT-6 Astra
Design Arena: agents building games1261Not measuredOnly Claude Opus 5 measured
Design Arena: PowerPoint slides12401219Claude Opus 5
Design Arena: spreadsheetsNot measured1365Only GPT-6 Astra measured
LiveBench research88.9%85.2%Claude Opus 5
LiveBench browser84.0%84.0%Tie
LiveBench long agentic62.1%82.4%GPT-6 Astra
SREBench (Vals AI)12.2%56.9%GPT-6 Astra
Finance Agent (Vals AI)58.6%53.5%Claude Opus 5
Tax Agent Bench (Vals AI)45.5%20.7%Claude Opus 5
Legal Research Bench (Vals AI)55.3%39.4%Claude Opus 5
Harvey's Legal Agent Benchmark (Vals AI)Not measured7.1%Only GPT-6 Astra measured
Excel Modeling Benchmark (Vals AI)73.6%71.7%Claude Opus 5
BioMysteryBench (Vals AI)79.3%79.3%Tie
Terminal-Bench-Science (Vals AI)27.1%62.9%GPT-6 Astra
Vals Science Index47.9%65.1%GPT-6 Astra
Time Horizon Index (Vals AI)18.8%90.5%GPT-6 Astra
CUA-bench (Vals AI)9.0%19.2%GPT-6 Astra
Robo-bench (Vals AI)Not measured23.3%Only GPT-6 Astra measured
APEX-Agents65.8%64.7%Claude Opus 5
Remote Labor IndexNot measured20.8%Only GPT-6 Astra measured
OSWorld 2.031.4%Not measuredOnly Claude Opus 5 measured
BALROG63.4%68.3%GPT-6 Astra
PostTrainBench35.0%Not measuredOnly Claude Opus 5 measured
SpireClimbNot measured25.7%Only GPT-6 Astra measured
Vending-Bench 2$11,182$15,515GPT-6 Astra
LMArena Search1242Not measuredOnly Claude Opus 5 measured
Neither model has a result on the other 9 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index69
Reasoning benchmarks for Claude Opus 5 and GPT-6 Astra
BenchmarkClaude Opus 5GPT-6 AstraLeader
GPQA Diamond93.4%Not measuredOnly Claude Opus 5 measured
LiveBench reasoning91.2%92.7%GPT-6 Astra
MysteryMechanism (Vals AI)37.4%53.1%GPT-6 Astra
MMMU Pro (Vals AI)89.9%Not measuredOnly Claude Opus 5 measured
GPQA Diamond (Artificial Analysis run)93.7%96.3%GPT-6 Astra
Humanity's Last Exam (Artificial Analysis run)54.9%54.7%Claude Opus 5
GPQA Diamond (Epoch AI run)93.9%95.8%GPT-6 Astra
ARC-AGI97.5%98.5%GPT-6 Astra
ARC-AGI-290.4%95.0%GPT-6 Astra
SimpleBench80.6%Not measuredOnly Claude Opus 5 measured
Chess puzzles43.0%74.0%GPT-6 Astra
Mystery game puzzles59.0%84.0%GPT-6 Astra
EBR-bench45.7%76.2%GPT-6 Astra
LMArena Text: hard prompts15271463Claude Opus 5
LMArena Text: expert15541489Claude Opus 5
LMArena Text: hard prompts in English15231464Claude Opus 5
LMArena Vision: diagrams13481297Claude Opus 5
LMArena Vision: homework13381337Claude Opus 5
Neither model has a result on the other 4 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index23
Math benchmarks for Claude Opus 5 and GPT-6 Astra
BenchmarkClaude Opus 5GPT-6 AstraLeader
OTIS Mock AIME98.9%100.0%GPT-6 Astra
FrontierMath, tiers 1 to 385.6%93.7%GPT-6 Astra
FrontierMath, tier 473.2%97.6%GPT-6 Astra
ProofBench99.0%99.0%Tie
LMArena Text: math15331465Claude Opus 5
LMArena Text: mathematical work15441484Claude Opus 5
Neither model has a result on the other 7 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index51
Knowledge benchmarks for Claude Opus 5 and GPT-6 Astra
BenchmarkClaude Opus 5GPT-6 AstraLeader
MMLU Pro91.6%Not measuredOnly Claude Opus 5 measured
LegalBench (Vals AI)87.0%Not measuredOnly Claude Opus 5 measured
CorpFin (Vals AI)73.2%Not measuredOnly Claude Opus 5 measured
TaxEval (Vals AI)75.1%Not measuredOnly Claude Opus 5 measured
MortgageTax (Vals AI)72.1%Not measuredOnly Claude Opus 5 measured
MedCode (Vals AI)63.6%48.5%Claude Opus 5
SimpleQA Verified59.9%75.6%GPT-6 Astra
LMArena Text: science15311451Claude Opus 5
LMArena Text: medicine and healthcare15251425Claude Opus 5
LMArena Text: legal and government15301462Claude Opus 5
LMArena Text: business and finance15011439Claude Opus 5
FACTS Benchmark SuiteNot measured75.7%Only GPT-6 Astra measured
Neither model has a result on the other 6 boards here
Web designLike-for-likePublic boards, shown as published, kept out of the index18
Web design benchmarks for Claude Opus 5 and GPT-6 Astra
BenchmarkClaude Opus 5GPT-6 AstraLeader
Design Arena: websites13091336GPT-6 Astra
Design Arena: UI components13461393GPT-6 Astra
Design Arena: game dev13371478GPT-6 Astra
Design Arena: data viz13291335GPT-6 Astra
Design Arena: 3D13341500GPT-6 Astra
Design Arena: all code categories13231379GPT-6 Astra
Design Arena: SVG13111431GPT-6 Astra
Design Arena: ASCII art13521353GPT-6 Astra
LiveBench design83.0%82.0%Claude Opus 5
GeneralLike-for-likePublic boards, shown as published, kept out of the index272
General benchmarks for Claude Opus 5 and GPT-6 Astra
BenchmarkClaude Opus 5GPT-6 AstraLeader
Vals IndexNot measured65.0%Only GPT-6 Astra measured
LiveBench data analysis74.6%83.0%GPT-6 Astra
Public Benefits Bench (Vals AI)76.9%Not measuredOnly Claude Opus 5 measured
SAGE (Vals AI)49.4%46.4%Claude Opus 5
MedScribe (Vals AI)91.0%87.9%Claude Opus 5
Vals Multimodal Index73.9%Not measuredOnly Claude Opus 5 measured
Long context reasoning (AA-LCR)82.0%80.7%Claude Opus 5
DTBench97.9%97.3%Claude Opus 5
LMCA64.5%64.4%Claude Opus 5
Furniture assembly60.8%80.0%GPT-6 Astra
LMArena Text15091441Claude Opus 5
LMArena Text: instruction following15181455Claude Opus 5
LMArena Text: creative writing14931420Claude Opus 5
LMArena Vision13201280Claude Opus 5
LMArena Text: English15081447Claude Opus 5
LMArena Text: non-English15021432Claude Opus 5
LMArena Text: Chinese15731489Claude Opus 5
LMArena Text: German15271459Claude Opus 5
LMArena Text: Japanese15201405Claude Opus 5
LMArena Text: Korean15171428Claude Opus 5
LMArena Text: French15191432Claude Opus 5
LMArena Text: Spanish15231422Claude Opus 5
LMArena Text: Russian15071438Claude Opus 5
LMArena Text: Polish15221422Claude Opus 5
LMArena Text: multi-turn15001447Claude Opus 5
LMArena Text: longer queries15171460Claude Opus 5
LMArena Text: writing and language14991437Claude Opus 5
LMArena Text: entertainment and media14811427Claude Opus 5
LMArena Vision: English13151280Claude Opus 5
LMArena Vision: Chinese14001361Claude Opus 5
LMArena Vision: reading text13291292Claude Opus 5
LMArena Vision: humor1293Not measuredOnly Claude Opus 5 measured
LMArena Vision: creative writing13131286Claude Opus 5
LMArena Document1516Not measuredOnly Claude Opus 5 measured
Neither model has a result on the other 12 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for Claude Opus 5 and GPT-6 Astra
BenchmarkClaude Opus 5GPT-6 AstraLeader
Neither model has a result on 5 boards here

What each workload costs

Three fixed token mixes at each model's listed price, so per-token rates become a bill.

Chat turn

1K input + 500 output tokens

Claude Opus 5$0.018Fits in one request
GPT-6 Astra$0.035Fits in one request

Claude Opus 5 costs less

Repository review

50K input + 3K output tokens

Claude Opus 5$0.325Fits in one request
GPT-6 Astra$0.650Fits in one request

Claude Opus 5 costs less

Agent loop

200K input + 10K output tokens

Claude Opus 5$1.25Fits in one request
GPT-6 Astra$2.50Fits in one request

Claude Opus 5 costs less

Listed OpenRouter rates, without cache or batch discounts. A workload that does not fit a model's context window would need splitting.

Specification differences

The leading value in a row is marked; a row with a missing value names no leader. Our two rows are marked in red.

Specifications of Claude Opus 5 and GPT-6 Astra
SpecificationAnthropic Claude Opus 5OpenAI GPT-6 Astra
MakerAnthropicOpenAI
WeightsNo public weights listedNo public weights listed
ReleasedJuly 24, 2026September 4, 2026
Context window1M1.05M
Input $/Mtok$5.00$10.00
Output $/Mtok$25.00$50.00
Coding index83.085.1
Independent measurements behind it65
Cost per solved task$1.33Not computable
SourceOpenRouter listingOpenRouter listing

Questions

Is Claude Opus 5 better than GPT-6 Astra for coding?

Pick GPT-6 Astra for coding: it scores 85.1 on the Vibe Coding Index against 83.0 for Claude Opus 5, though on 5 benchmarks against 6. Pick Claude Opus 5 for output tokens 2.0x cheaper ($25.00 vs $50.00 per million) and work measured by AA Coding Index or Vibe Code Bench 1-100 (Vals AI), where it leads.

Which is cheaper, Claude Opus 5 or GPT-6 Astra?

Claude Opus 5 costs $5.00 per million input tokens and $25.00 per million output tokens; GPT-6 Astra costs $10.00 and $50.00. Cost per solved task is known for Claude Opus 5 only.

Which has the bigger context window, Claude Opus 5 or GPT-6 Astra?

GPT-6 Astra, with 1.05M tokens against 1M.

How many benchmarks is the Claude Opus 5 vs GPT-6 Astra comparison based on?

111 benchmarks have a result for both models as of October 22, 2026. Claude Opus 5's coding index rests on 6 independent benchmarks and GPT-6 Astra's on 5. Where only one model has a result, the page says so instead of naming a winner.

Compare next

Embed this comparison on your site

Free under CC BY 4.0, while the source link stays visible. Preview the frame · Terms and other embeds

<iframe src="https://llmbenchmark.ai/embed/compare/claude-opus-5-vs-gpt-6-astra" title="Claude Opus 5 vs GPT-6 Astra, LLMBenchmark.ai" width="100%" height="360" style="border:0;max-width:100%" loading="lazy"></iframe>
<p style="font:13px/1.4 sans-serif;margin:6px 0 0">Source: <a href="https://llmbenchmark.ai/compare/claude-opus-5-vs-gpt-6-astra">LLMBenchmark.ai, Claude Opus 5 vs GPT-6 Astra</a></p>