Fronset Benchmark

Claude Sonnet 4.6 (archived)

Provider
Anthropic
Model name
claude-sonnet-4-6
Last benchmarked
Qualified on
20 / 51 tasks (at the 90% bar)
Best value on
0 tasks

Good enough on 20/51 tasks at the 90% bar. Best value on 0 tasks. Best fit for: Long-form Content Generation.

Cost vs quality across all tasks, as of 15 August 2026

0%25%50%75%100%78910Quality score (7–10)Cost-efficiency vs best value (1.0 = best value)Engagement Triage — quality 8.87, 24.4x the cost of the best-value good-enough optionAuthor Matching — quality 8.13, 30.8x the cost of the best-value good-enough optionClaim Extraction — quality 7.11, 5.5x the cost of the best-value good-enough optionOnboarding Chapter Prompt Adaptation — quality 8.66, 53.8x the cost of the best-value good-enough optionTheme Generation — quality 8.77, 30.2x the cost of the best-value good-enough optionGeographic Region Identification — quality 9.10, 72.2x the cost of the best-value good-enough optionX Post Selection — quality 8.74, 20.4x the cost of the best-value good-enough optionAuthor Voice Generation — quality 9.12, 57.0x the cost of the best-value good-enough optionInvestment Panel Voting — quality 8.97, 22.6x the cost of the best-value good-enough optionPublication Title Generation — quality 8.37, 17.5x the cost of the best-value good-enough optionLanguage Detection — quality 9.95, 50.3x the cost of the best-value good-enough optionTrading Recommendation — quality 8.59, 24.7x the cost of the best-value good-enough optionTopic Report Relevance Scoring — quality 8.39, 14.7x the cost of the best-value good-enough optionPrompt Adaptation — quality 8.90, 15.0x the cost of the best-value good-enough optionTopic Cluster Naming — quality 8.63, 28.8x the cost of the best-value good-enough optionClaim-Referenced Analyst Writing — quality 8.87, 10.7x the cost of the best-value good-enough optionSocial Post Promotion — quality 8.08, 8.5x the cost of the best-value good-enough optionExecutive Summary Generation — quality 7.81, 61.3x the cost of the best-value good-enough optionActivity Feed Blurb Generation — quality 8.19, 23.7x the cost of the best-value good-enough optionTopic-to-Section Assignment — quality 8.10, 12.8x the cost of the best-value good-enough option

within ~1.3× of the best-value model · 1.3–2× · >2× · ★ this model is the best-value pick on that task. Top-right = best quadrant. Only tasks where this model qualifies at the 90% bar are plotted.

Per-task breakdown, as of 15 August 2026

TaskCategoryQuality (% of best)ConfidenceOverpay
Claim ExtractionStructured Data & Fact Extraction92%HIGH5.5x
Social Post PromotionSocial & Promotional Content91%HIGH8.5x
Claim-Referenced Analyst WritingLong-form Content Generation96%HIGH11x
Topic-to-Section AssignmentTopic Organization & Clustering92%HIGH13x
Topic Report Relevance ScoringRelevance, Classification & Matching90%HIGH15x
Prompt Adaptation bestInfrastructure & Utility100%RANKED15x
Publication Title GenerationContent Summarization & Synthesis92%RANKED18x
X Post SelectionRelevance, Classification & Matching99%RANKED20x
Investment Panel VotingFinancial Analysis & Trading Decisions99%RANKED23x
Activity Feed Blurb GenerationSocial & Promotional Content94%MEDIUM24x
Engagement TriageRelevance, Classification & Matching96%RANKED24x
Trading RecommendationFinancial Analysis & Trading Decisions97%HIGH25x
Topic Cluster NamingTopic Organization & Clustering96%RANKED29x
Theme GenerationLong-form Content Generation90%HIGH30x
Author MatchingRelevance, Classification & Matching95%MEDIUM31x
Language DetectionRelevance, Classification & Matching99%RANKED50x
Onboarding Chapter Prompt AdaptationInfrastructure & Utility96%HIGH54x
Author Voice GenerationLong-form Content Generation94%RANKED57x
Executive Summary GenerationContent Summarization & Synthesis98%MEDIUM61x
Geographic Region IdentificationStructured Data & Fact Extraction97%HIGH72x

Overpay — how much more you pay by running this model instead of the best-value model that clears the quality bar on that task (marked ★). "16x" means you overpay 16× — the same output for 16× the best-value good-enough option; ★ means this model is that option (no overpayment). Confidence — how sure we are about the quality score (more judgments + more agreement = higher confidence): RANKED many independent judges scored this model's outputs and their agreement is very high (most confident) — HIGH many judges have scored it and they mostly agree (well-pinned) — MEDIUM enough judges have weighed in to publish, but they disagree more than we'd like (treat with a small grain of salt). LOW-confidence cells are hidden everywhere on the site. See the methodology for the exact thresholds.