CompareLLM.ai
Live
Leaderboard
Models
Compare
Best of & Stacks
Research & News
…
CompareLLM.ai
Precision Benchmarks

Programmatic, dated AI model benchmarks, head-to-head comparisons, and Stack Engine presets.

Daily ingest · 06:00 UTC

Analytics & Benchmarks

  • AI Model Leaderboard
  • Head-to-Head Compare Hub
  • Models Directory
  • Stack Engine Presets
  • Frontier Models
  • Open Weights Catalog

Guides & Intent Lists

  • Best LLM Lists (2026)
  • Best Coding LLM
  • Best Cheap LLM
  • Fastest Low-Latency LLM
  • Claude vs GPT Benchmark
  • What is Elo?
  • Methodology Guides
  • News & Dispatches

Transparency & API

  • Evaluation Methodology
  • Benchmark Changelog
  • Public JSON API
  • llms.txt Specification
  • Privacy Policy
  • Sign In / Account

© 2026 CompareLLM. Public benchmark data aggregated from Arena Elo, LiveBench, SWE-bench & OpenRouter.

Every score has a dated snapshot.

Theme:
Currency:
  1. Home
  2. News
  3. Sonnet vs Opus in 2026: same family, different invoice
Sonnet vs Opus in 2026: same family, different invoice
compareVerified Dispatch
CompareLLM Intelligence Desk·Aug 10, 2026

Sonnet vs Opus in 2026: same family, different invoice

Usually Sonnet 5 vs Opus 5. Coding delta versus output $/1M. The workhorse question.

Search intent: sonnet vs opus

Evaluated Models (11):
Claude Sonnet 5Anthropic
Claude Opus 5Anthropic
Claude Opus 4.5Anthropic
Claude Opus 4.6Anthropic
Claude Sonnet 4.5Anthropic
Claude 3.7 SonnetAnthropic
Claude Opus 4.8Anthropic
Claude 3 OpusAnthropic
Claude Opus 4Anthropic
Claude 3.5 SonnetAnthropic
Claude Sonnet 4Anthropic
Share Analysis:
WhatsAppTelegramXLinkedInReddit
Verified Benchmark Scorecard11 Models Evaluated

Claude Sonnet 5 & Frontier Peer Benchmark Comparison

Live benchmark scores, throughput speeds, and token pricing with dynamic peer comparison.

LMSYS Arena Elo1,556Claude Sonnet 5
SWE-bench Verified73.8%Coding resolve %
Throughput Speed118 tok/sGeneration rate
Output Price / 1M$10.00List rate
Interactive Model Comparison:11 models selected
Claude Sonnet 5ArticleClaude Opus 5ArticleClaude Opus 4.5ArticleClaude Opus 4.6ArticleClaude Sonnet 4.5ArticleClaude 3.7 SonnetArticleClaude Opus 4.8ArticleClaude 3 OpusArticleClaude Opus 4ArticleClaude 3.5 SonnetArticleClaude Sonnet 4Article
LMSYS Arena EloOverall human preference
quality
Claude Sonnet 5
1,556
Claude Opus 5Best
1,624
Claude Opus 4.5
1,568
Claude Opus 4.6
1,574
Claude Sonnet 4.5
1,524
Claude 3.7 Sonnet
1,362
Claude Opus 4.8
1,598
Claude 3 Opus
1,248
Claude Opus 4
1,490
Claude 3.5 Sonnet
1,280
Claude Sonnet 4
1,455
Coding EloProgramming preference
quality
Claude Sonnet 5
1,562
Claude Opus 5Best
1,618
Claude Opus 4.5
1,574
Claude Opus 4.6
1,570
Claude Sonnet 4.5
1,531
Claude 3.7 Sonnet
—
Claude Opus 4.8
1,604
Claude 3 Opus
—
Claude Opus 4
1,502
Claude 3.5 Sonnet
—
Claude Sonnet 4
1,468
SWE-bench VerifiedGitHub issue resolve %
quality
Claude Sonnet 5
73.8%
Claude Opus 5Best
79.2%
Claude Opus 4.5
76.8%
Claude Opus 4.6
75.9%
Claude Sonnet 4.5
70.1%
Claude 3.7 Sonnet
70.3%
Claude Opus 4.8
78.4%
Claude 3 Opus
38.4%
Claude Opus 4
72.5%
Claude 3.5 Sonnet
49%
Claude Sonnet 4
62.8%
GPQA DiamondPhD-level science reasoning %
quality
Claude Sonnet 5
82.8%
Claude Opus 5Best
87.1%
Claude Opus 4.5
84.1%
Claude Opus 4.6
84.8%
Claude Sonnet 4.5
80.2%
Claude 3.7 Sonnet
78.5%
Claude Opus 4.8
86.2%
Claude 3 Opus
50.4%
Claude Opus 4
80.4%
Claude 3.5 Sonnet
65%
Claude Sonnet 4
76.4%
LiveBenchContamination-free reasoning
quality
Claude Sonnet 5
70.4%
Claude Opus 5Best
74.8%
Claude Opus 4.5
71.2%
Claude Opus 4.6
72%
Claude Sonnet 4.5
67.4%
Claude 3.7 Sonnet
58.4%
Claude Opus 4.8
73.6%
Claude 3 Opus
46.2%
Claude Opus 4
66.8%
Claude 3.5 Sonnet
52.1%
Claude Sonnet 4
63.2%
Throughput (Speed)Output tokens / second
speed
Claude Sonnet 5Best
118 tok/s
Claude Opus 5
72 tok/s
Claude Opus 4.5
62 tok/s
Claude Opus 4.6
64 tok/s
Claude Sonnet 4.5
88 tok/s
Claude 3.7 Sonnet
85 tok/s
Claude Opus 4.8
66 tok/s
Claude 3 Opus
48 tok/s
Claude Opus 4
58 tok/s
Claude 3.5 Sonnet
80 tok/s
Claude Sonnet 4
92 tok/s
Time-to-First-TokenInitial latency (ms)
speed
Claude Sonnet 5Best
175 ms
Claude Opus 5
310 ms
Claude Opus 4.5
380 ms
Claude Opus 4.6
360 ms
Claude Sonnet 4.5
240 ms
Claude 3.7 Sonnet
240 ms
Claude Opus 4.8
330 ms
Claude 3 Opus
420 ms
Claude Opus 4
390 ms
Claude 3.5 Sonnet
280 ms
Claude Sonnet 4
220 ms
Output Token Price$ per 1M reply tokens
price
Claude Sonnet 5Best
$10/1M
Claude Opus 5
$25/1M
Claude Opus 4.5
$75/1M
Claude Opus 4.6
$75/1M
Claude Sonnet 4.5
$15/1M
Claude 3.7 Sonnet
$15/1M
Claude Opus 4.8
$25/1M
Claude 3 Opus
$75/1M
Claude Opus 4
$75/1M
Claude 3.5 Sonnet
$15/1M
Claude Sonnet 4
$15/1M
Input Token Price$ per 1M prompt tokens
price
Claude Sonnet 5Best
$2/1M
Claude Opus 5
$5/1M
Claude Opus 4.5
$15/1M
Claude Opus 4.6
$15/1M
Claude Sonnet 4.5
$3/1M
Claude 3.7 Sonnet
$3/1M
Claude Opus 4.8
$5/1M
Claude 3 Opus
$15/1M
Claude Opus 4
$15/1M
Claude 3.5 Sonnet
$3/1M
Claude Sonnet 4
$3/1M
Context WindowMax token capacity
capacity
Claude Sonnet 5Best
1M
Claude Opus 5
1M
Claude Opus 4.5
200k
Claude Opus 4.6
200k
Claude Sonnet 4.5
200k
Claude 3.7 Sonnet
200k
Claude Opus 4.8
1M
Claude 3 Opus
200k
Claude Opus 4
200k
Claude 3.5 Sonnet
200k
Claude Sonnet 4
200k
Benchmark / Metric
Claude Sonnet 5Anthropic · Article Model
Claude Opus 5Anthropic · Article Model
Claude Opus 4.5Anthropic · Article Model
Claude Opus 4.6Anthropic · Article Model
Claude Sonnet 4.5Anthropic · Article Model
Claude 3.7 SonnetAnthropic · Article Model
Claude Opus 4.8Anthropic · Article Model
Claude 3 OpusAnthropic · Article Model
Claude Opus 4Anthropic · Article Model
Claude 3.5 SonnetAnthropic · Article Model
Claude Sonnet 4Anthropic · Article Model
LMSYS Arena EloOverall human preference
1,556
1,624Top
1,568
1,574
1,524
1,362
1,598
1,248
1,490
1,280
1,455
Coding EloProgramming preference
1,562
1,618Top
1,574
1,570
1,531
—
1,604
—
1,502
—
1,468
SWE-bench VerifiedGitHub issue resolve %
73.8%
79.2%Top
76.8%
75.9%
70.1%
70.3%
78.4%
38.4%
72.5%
49%
62.8%
GPQA DiamondPhD-level science reasoning %
82.8%
87.1%Top
84.1%
84.8%
80.2%
78.5%
86.2%
50.4%
80.4%
65%
76.4%
LiveBenchContamination-free reasoning
70.4%
74.8%Top
71.2%
72%
67.4%
58.4%
73.6%
46.2%
66.8%
52.1%
63.2%
Throughput (Speed)Output tokens / second
118 tok/sTop
72 tok/s
62 tok/s
64 tok/s
88 tok/s
85 tok/s
66 tok/s
48 tok/s
58 tok/s
80 tok/s
92 tok/s
Time-to-First-TokenInitial latency (ms)
175 msTop
310 ms
380 ms
360 ms
240 ms
240 ms
330 ms
420 ms
390 ms
280 ms
220 ms
Output Token Price$ per 1M reply tokens
$10/1MTop
$25/1M
$75/1M
$75/1M
$15/1M
$15/1M
$25/1M
$75/1M
$75/1M
$15/1M
$15/1M
Input Token Price$ per 1M prompt tokens
$2/1MTop
$5/1M
$15/1M
$15/1M
$3/1M
$3/1M
$5/1M
$15/1M
$15/1M
$3/1M
$3/1M
Context WindowMax token capacity
1MTop
1M
200k
200k
200k
200k
1M
200k
200k
200k
200k
Executive Key Takeaway
Usually Sonnet 5 vs Opus 5. Coding delta versus output $/1M. The workhorse question.

Sonnet vs Opus is the intra-Anthropic question. After Jun 30 / Jul 24 that is Sonnet 5 vs Opus 5. Sonnet is the workhorse at $2/$10. Opus is the high-capability reasoning model at $5/$25. If the coding delta is small and you run thousands of agent turns, Sonnet is the default. If the task requires deep multi-step computer-use, pay Opus and measure on your eval.

Family pairs are not doorways

The numbers differ on price, SWE-bench, and TTFT. That is enough unique data. We still write this dispatch so the query has a dated explanation of the Aug 10 permanent Sonnet price.

Empirical Evaluation & Architectural Analysis

Standardized telemetry recorded in the verified CompareLLM Leaderboards captures distinct architectural priorities between Claude Sonnet 5 and Claude Opus 5. While blind human pairwise preference testing reflects instruction compliance and reasoning depth, domain-specific suites such as SWE-bench Verified and GPQA Diamond highlight coding execution and advanced STEM reasoning. For the live pairwise breakdown with historical trendlines, reference the interactive scorecard above or visit the Claude Sonnet 5 vs Claude Opus 5 Showdown.

Fable is the third sibling

If you are comparing invoices, include Fable 5 only when you actually need the $10/$50 X-High reasoning tier.

Strategic Deployment Recommendation

  • Production Workloads: For enterprise workloads requiring strict reliability, cross-reference Top Coding LLMs and Cheapest High-Quality LLMs.
  • Direct Pair Comparison: Explore the live pairwise breakdown at /compare/claude-opus-5-vs-claude-sonnet-5.
  • Architectural Stacks: Recommended architectural configurations can be evaluated on the CompareLLM Stack Engine.

Frequently Asked Questions

Is this an CompareLLM lab test score? No. Linked model dossiers and compare showdowns reflect dated snapshots gathered from named public evaluation benchmarks. Seed catalog entries remain transparently timestamped until daily ingest updates them.

Where can I see live comparisons for this model? View the showdown at /compare/claude-opus-5-vs-claude-sonnet-5.

What primary search query does this briefing answer? sonnet vs opus.

Share Analysis:
WhatsAppTelegramXLinkedInReddit

Head-to-Head Showdowns for Mentioned Models

Featured Article ShowdownClaude Sonnet 5 vs Claude Opus 5
Compare Benchmarks
Benchmark MatchupClaude Sonnet 5 vs GPT-5
Benchmark MatchupClaude Sonnet 5 vs GPT-5 mini
Benchmark MatchupClaude Opus 5 vs GPT-5
Benchmark MatchupClaude Opus 5 vs GPT-5 mini
Benchmark MatchupClaude Opus 4.5 vs GPT-5
Benchmark MatchupClaude Opus 4.5 vs GPT-5 mini
Benchmark MatchupClaude Opus 4.6 vs GPT-5
Benchmark MatchupClaude Opus 4.6 vs GPT-5 mini
Benchmark MatchupClaude Sonnet 4.5 vs GPT-5
Benchmark MatchupClaude Sonnet 4.5 vs GPT-5 mini
Benchmark MatchupClaude 3.7 Sonnet vs GPT-5
Benchmark MatchupClaude 3.7 Sonnet vs GPT-5 mini
Benchmark MatchupClaude Opus 4.8 vs GPT-5
Benchmark MatchupClaude Opus 4.8 vs GPT-5 mini
Benchmark MatchupClaude 3 Opus vs GPT-5
Benchmark MatchupClaude 3 Opus vs GPT-5 mini
Benchmark MatchupClaude Opus 4 vs GPT-5
Benchmark MatchupClaude Opus 4 vs GPT-5 mini
Benchmark MatchupClaude 3.5 Sonnet vs GPT-5
Benchmark MatchupClaude 3.5 Sonnet vs GPT-5 mini
Benchmark MatchupClaude Sonnet 4 vs GPT-5
Benchmark MatchupClaude Sonnet 4 vs GPT-5 mini

Related news

  • price · Aug 10, 2026

    Claude Sonnet 5 at $2/$10: the 2026 Anthropic workhorse

  • launch · Jun 20, 2026

    Keep Opus 4.8 on the board: a baseline, not a flagship

  • launch · Feb 18, 2026

    Claude Sonnet 4.5 is the previous workhorse — still a valid vs baseline

  • compare · Jul 24, 2026

    Claude vs Gemini: coding, Elo, and output price — not a brand mashup

Back to All DispatchesExplore Comparison Matrix