# Claude Sonnet 5 vs Claude Sonnet 5.5 (2026): side-by-side comparison Source: [GLAD-AI-TOR](https://glad-ia-tor.com) · Full page: https://glad-ia-tor.com/vs/claude-sonnet-5-vs-claude-sonnet-5-5 Arena: llm-models · Crowd scores are live visitor verdicts (one per person per tool, never paid, Bayesian-smoothed). ## At a glance | | Claude Sonnet 5 | Claude Sonnet 5.5 | |---|---|---| | Price | $10/1M out | $10/1M out | | Crowd score | 57% (3 votes) | 50% (0 votes) | | provider | Anthropic | Anthropic | | contextWindow | 1M tokens (128K max output) | 1M tokens (128K max output) | | priceIn | $2/1M in | $2/1M in | | priceOut | $10/1M out | $10/1M out | | modalities | text, vision (image input, text output) | text, vision (image input, text output) | | openWeights | no | no | | reasoning (1-5) | 4.5 | 4.5 | | coding (1-5) | 4.5 | 5 | | writing (1-5) | 4.5 | 4.5 | | speed (1-5) | 4 | 4.5 | | valueForMoney (1-5) | 4.5 | 4.5 | ### Claude Sonnet 5 > Anthropic's most agentic Sonnet: near Opus 4.8 quality on coding and agents at $2/$10 with 1M context Strengths: - Large agentic gains over Sonnet 4.6: Terminal-Bench 2.1 80.4% vs 67.0%, OSWorld-Verified 81.2% vs 78.5%, SWE-bench Pro 63.2% vs 58.1% - Matches Opus 4.8 on knowledge work (GDPval-AA v2: 1,618 vs 1,615) and nearly ties it on Humanity's Last Exam with tools (57.4% vs 57.9%) at 40% of Opus 4.8 pricing - 1M token context window and 128K max output; $2/$10 per 1M tokens, the launch price made permanent (the planned rise to $3/$15 was cancelled) - Persistent self-verifying agent behavior: hands-on reviews note it tests its own code and iterates on hard problems until solved, unlike Sonnet 4.6 Weaknesses: - New tokenizer inflates token counts roughly 30% for the same text (1.0-1.35x per Anthropic; ~1.4x English, ~1.28x Python measured by Simon Willison), raising effective cost and eating into the savings of the lower sticker price - Verbose and token-hungry: ~$2.29 per task vs ~$1.20 for Sonnet 4.6 in independent tests (ranked 101st of 161 for cost efficiency); at high effort cost-per-task can exceed Opus 4.8 - Measurably slower than Sonnet 4.6 on small routine edits and prone to over-engineering simple tasks (CodeRabbit hands-on review) Verdict: Choose Sonnet 5 if you run coding, terminal or computer-use agents and want near Opus 4.8 quality at Sonnet prices, at $2/$10, now its permanent price; it is a strict upgrade over Sonnet 4.6 at low and medium effort. Budget for the new tokenizer and its verbosity: real per-task costs run well above Sonnet 4.6, and at the highest effort levels Opus 4.8 can be the better deal per solved task. Avoid it for latency-sensitive small edits or pipelines that rely on temperature and top_p, which now error. Sonnet 4.6 remains the pragmatic pick for high-volume tiny-diff workloads. Full review: https://glad-ia-tor.com/tool/claude-sonnet-5 · Markdown: https://glad-ia-tor.com/tool/claude-sonnet-5.md ### Claude Sonnet 5.5 > Anthropic's fastest Sonnet: 70.6% on Terminal-Bench 4.0, within two points of Opus 5.5 on knowledge work, at $2/$10 Strengths: - 70.6% on Terminal-Bench 4.0 against 10.3% for Sonnet 5, ahead of Opus 5.5's 66.4% on the same Anthropic chart - Near-Opus knowledge work: 1844 Elo on GDPval-AA v2.1 vs 1846 for Opus 5.5, and 80.1% on OSWorld 2.1 vs 81.8% - $2/$10 per 1M tokens with cache reads at $0.10 (5% of input), and Anthropic measures up to 30% lower cost per task than Sonnet 5 - Fast: 138.7 output tokens per second on Artificial Analysis (#38 of 195), 30%+ faster than Sonnet 5 per Anthropic Weaknesses: - Opus 5.5 remains clearly stronger at complex, open-ended work requiring sustained judgment, Anthropic says so itself - Expensive at maximum effort: $10.02 per Intelligence Index task on Artificial Analysis, #54 of 58, as it spends many tokens when allowed to - Five breaking changes for Sonnet 5 code: forced tool use returns an error, thinking blocks are tied to the model, some computer-use tool versions are unsupported Verdict: Sonnet 5.5 is the default pick of the Claude lineup for most work: it beats Opus 5.5 on Anthropic's Terminal-Bench 4.0 chart, sits within two Elo points of it on GDPval-AA, and costs half as much per token. Take it for coding agents, computer use and office documents, and keep the effort at low or medium where Anthropic shows it beating Sonnet 5's best score for a tenth of the cost. Step up to Opus 5.5 for long, ambiguous tasks that need judgment, and watch the bill at maximum effort, where Artificial Analysis ranks it among the costliest models per task. Teams on Sonnet 5 should migrate: same price, faster, fewer tokens, but plan for the API breaking changes. Full review: https://glad-ia-tor.com/tool/claude-sonnet-5-5 · Markdown: https://glad-ia-tor.com/tool/claude-sonnet-5-5.md ## More Full llm-models ranking: https://glad-ia-tor.com/hall-of-fame/llm-models --- This markdown version exists for AI assistants; the canonical page is https://glad-ia-tor.com/vs/claude-sonnet-5-vs-claude-sonnet-5-5