Public benchmarkExternal source reviewReviewed by DanPublished Source checked Updated
Claude Sonnet 5.5: public coding benchmark receipts
The report
Verdict
Published benchmark evidence
Sonnet 5.5's headline terminal score and Cognition's code-mergeability score measure different work. A higher raw percentage is not evidence that one workflow succeeds more often.
On the tier list: The November 17 tier is Dan’s own ranking; public benchmark numbers are a separate evidence channel, not an automatic tier adjustment.
What the benchmarks report
- Terminal-Bench 4.0: 70.6%. Sonnet 5.5; headline benchmark result. Reported by Anthropic vendor release.
- FrontierCode 2.0 — Cognition: 42.2%. Sonnet 5.5; max; Claude Code. Reported by Cognition benchmark operator.
- CursorBench 4.0: 55.5%. Sonnet 5.5; multi-file coding. Reported by Anthropic vendor release.
Original sources
Source 1 — Terminal-Bench 4.0, CursorBench 4.0.
Source 2 — FrontierCode 2.0 — Cognition.
What this evidence tells a reader
Anthropic publishes 70.6% for Terminal-Bench 4.0 and 55.5% for CursorBench 4.0. Cognition separately reports 42.2% for FrontierCode 2.0 at max effort in a Claude Code harness. The operator, tasks, and scoring differ across all three; the percentages should remain attached to their named suites.
The Sonnet terminal figure is numerically above Opus 5.5's listed 66.4% on a same-named Terminal-Bench generation, but this report does not establish identical effort levels, prompts and aggregation. It would be premature to declare Sonnet universally stronger from that comparison alone.
For a developer choosing a working reviewer, the right question may be whether Sonnet catches the actual mistakes in a diff, explains actionable fixes and avoids unnecessary edits. Those capabilities require an evaluated review task, not the vendor's headline terminal number.
How to use these results
- Compare Sonnet and Opus only on identical repository tasks and documented effort/harness settings.
- Count real corrections found and false alarms if choosing a code reviewer.
- Keep Cognition's mergeability score separate from Anthropic's terminal score.
Limits of this report
- Anthropic and Cognition published different evaluations; no matched Sonnet-versus-Opus local trial was run for this page.
- The numbers cannot establish price-adjusted review quality or false-positive behavior on a user's changes.
Follow the lab
New test reports and tier-list changes get announced in my Discord's #news channel. The full videos are on YouTube.
Receipts
DanManREAL · AI lab
Test receipt
Claude Sonnet 5.5
- Terminal-Bench 4.0
- 70.6%Anthropic vendor release (model publisher). Sonnet 5.5; headline benchmark result. Not independently rerun by DanManREAL.
- FrontierCode 2.0 — Cognition
- 42.2%Cognition benchmark operator (independent operator or secondary publication). Sonnet 5.5; max; Claude Code. Not independently rerun by DanManREAL.
- CursorBench 4.0
- 55.5%Anthropic vendor release (model publisher). Sonnet 5.5; multi-file coding. Not independently rerun by DanManREAL.
- Source review
- November 17, 2026Date these source claims were reviewed; not a claim about when the published benchmark was executed.
Verdict
Published benchmark evidence
Tier list The November 17 tier is Dan’s own ranking; public benchmark numbers are a separate evidence channel, not an automatic tier adjustment.
danmanreal.com/tests/claude-sonnet-5-5-public-benchmarks
Models covered by external benchmarks
On the tier list's bench
- Claude Sonnet 5.5 · 2 receipts on file
Related
Independent testing. Sponsorships are always disclosed, and verdicts are never for sale.
Disagree? Bring your receiptsAll reports
- One inbox for my AI coding agents: triage took 34% less time, resuming work did not improveHead-to-headNov 20
- Mythos 5.2: research and security benchmark source reviewPublic benchmarkNov 17
- Fable 5.2: agentic coding benchmark source reviewPublic benchmarkNov 17
- Claude Opus 5.5: public coding benchmark receiptsPublic benchmarkNov 17
- Astra 6.2: new model and public task-evaluation evidencePublic benchmarkNov 17
- Sol 6.2: new model and public coding benchmark evidencePublic benchmarkNov 17
- Gemini 4 Carbon: benchmark source and usage profilePublic benchmarkNov 17
- Qwen 4: Qwen4-Max launch and benchmark sourcesPublic benchmarkNov 17
- Kimi K3.1 Pro: public coding benchmark receiptsPublic benchmarkNov 17
- GLM-5.3: public coding benchmark receiptsPublic benchmarkNov 17
- Gemini 4 Argon: public coding benchmark receiptsPublic benchmarkNov 17
- Claude Sonnet 5.5: public coding benchmark receiptsPublic benchmarkNov 17
- DeepSeek V4.1 Flash: public coding benchmark receiptsPublic benchmarkNov 17
- Muse Wonder Full 1.0: public coding benchmark receiptsPublic benchmarkNov 17
- Grok 4.7: separate release and public coding resultsPublic benchmarkNov 17
- ai-design-check on two of my FABLE screens: 10 warnings, 4 of them false alarmsField testNov 2
- Vercel Agent Desktop used a sample Trip app in 6 actions without moving the mouseField testNov 1
- Cerberus Guard on sample data: one crawler name, two verdicts, and a block earned by behaviorField testOct 31
- GPT-6.1 Terra vs Opus 5.5: three real projects, and the designs I picked blindHead-to-headOct 29
- Steps → Reduced: opening a terminal agent went from 98 seconds to 11AuditOct 28
- One ChatGPT chat directed Opus and Codex at once through DMR_REMOTEField testOct 28
- Claude Duet ran my channel for a week and caught 269 fake AI crawlersField testOct 27
- Kimi K3.1 Pro built my animation family, after a quota stop and 21 minutes of failed requestsField testOct 24
- ChatGPT + Codex built 33 new DMR_REMOTE toolsField testOct 23
- Gemini 4 Argon reviewed a finished episode: 14 clips, one flag, one inconsistencyField testOct 21
- Qwen 3.8 Flash Next on my RTX 5080: 81 tokens a second with 202K tokens readAuditOct 21
- Does an extra AI review step pay off? A small DeepSeek trial said noHead-to-headOct 19
- My real DeepSeek bill: 12.31 billion tokens for $122.28Cost receiptOct 17
- Gemini 4 Argon kept hanging in Gemini CLI. The cause was a 60-second timeoutIncidentOct 14
- I asked 10 AI coding agents for their biggest weaknessSelf-reportOct 14
- Jarvis 27B on my own PC: 30 tokens a second, and it still recalls at 251KAuditOct 10
- My agents' first game benchmark was wrong. Here's how it was caughtIncidentOct 10
- Pixal3D + Blender turned my 2D art into animated 3D charactersField testSep 22
- DeepSeek Flash answered 200 OK and sent nothing for two hoursIncidentSep 14
- Grok, Codex and DeepSeek played WorldOS for Rank 1. Grok won with 98Head-to-headAug 20
- DeepSeek V4 Flash audited 18 of my real coding sessions. Two said 'done' too earlyAuditAug 4