Models/claude-sonnet-5

Claude Sonnet 5

A different row from Sonnet 5.5. Supabase is 75.4% without skills and 79.7% with them. Next.js is 81% without AGENTS.md and 97% with it. 18th on DeepSWE v1.1 Best (max) at 53.8%.

Anthropic · Proprietary · Fetched 2026-10-10

LMArena WebDev

1541

#43 / 142 · ±6

LMArena Image-to-WebDev

1539

#33 / 64 · ±8

Android Bench

6.7%

#10 / 10 · 1.3–13.3

Benchmark scores

Boards older than 90 days are not counted as recommendations

BenchmarkSettingScoreRankCostAs of

LMArena WebDevhigh1541±6#43—2026-10-08

LMArena Image-to-WebDevhigh1539±8#33—2026-10-09

Android Benchclaude-code6.71.3–13.3#10$283.72026-10-06

Terminal-Bench 4.0max · Claude Code12.4±3.1#33$9.6k2026-10-06

The Slop Index—49#33—2026-10-11

Supabase Evalshigh · Claude Code75.4#5—2026-10-10

Next.js Agent EvalsClaude Code81.0#10$0.3862026-09-25

ReactBenchxhigh32.328.9–35.7#34$3.172026-10-10

DeepSWE v1.1max · mini-swe-agent53.849.6–58.1#18$26.402026-09-22

SWE-bench Verified—This board has stopped updating——2026-02-26

Aider Polyglot—This board has stopped updating——2025-11-20

Use it or not

Why

On Next.js, pass@4 moves from 81% to 97% once AGENTS.md is available.

When not

The Supabase baseline is 75.4%, behind Opus 5.5 at 91.3%. Do not merge this with Sonnet 5.5's other boards into one score.

Appears in these scenarios

Collected from scenario boards

Build a Web frontend#10Next.js 81.0%Build an Android app#10Android Bench 6.7%Write CLI, scripts, and DevOps#18DeepSWE 53.8%Write docs & copy#33Slop Index 49