Claude Sonnet
The Claude line that balances speed and intelligence.
Claude Sonnet is the middle of Anthropic's lineup, and on Anthropic's own Terminal-Bench 4.0 table its top scorer for agentic coding. Sonnet 5.5 turns in outputs more than 30% faster than Sonnet 5 while keeping the 1M-token context window, at half Opus 5.5's price.
What it is good at
- Agentic coding at speed — Terminal-Bench 4.0 70.6%, the top of Anthropic's own table (vendor-reported)
- Fast for its class — outputs 30%+ faster than Sonnet 5 (vendor-reported)
- Half Opus 5.5's price at $2 in / $10 out per MTok
What it is less good at
- Below Opus and Fable on the most demanding long-horizon work
- FrontierCode 1.1 trails Opus 5.5 — 46.2% at max effort against 54.4% (vendor-reported)
Reaching it with TeamCLI
TeamCLI runs Claude Sonnet as Claude Sonnet 5.5:
Ways to reach it
| Route | Spends | Status |
|---|---|---|
| Claude Code | An Anthropic subscription (Claude Pro or Max), or an Anthropic API key. | Verified |
| Antigravity | A Google account on a Google AI plan (the Antigravity / Google AI Ultra pools). | Verified |
| Cursor | A Cursor plan. | Listed — not yet verified |
The seat line
Each route maps to a TeamCLI seat line — the same fields, written generically. No machine paths, no launcher script of anybody's, no team's own seat names.
| Route | cli | vendor | model | version | display | launch |
|---|---|---|---|---|---|---|
| Claude Code | claude-code | anthropic | Claude Sonnet | 5.5 | Claude Sonnet 5.5 | claude |
| Antigravity | antigravity | anthropic | Claude Sonnet | 5.5 | Claude Sonnet 5.5 | agy |
| Cursor | cursor | anthropic | Claude Sonnet | 5.5 | Claude Sonnet 5.5 | cursor-agent |
Versions
One block per version, the current one first, each with its own anchor. The comparison table puts the costs and the family's benchmark columns side by side; the values are verbatim, and every one is sourced and dated in its version's section below and in the Sources list. A dash means the source publishes no figure, and nothing here is estimated.
No independent intelligence or speed index is republished here: Artificial Analysis measures both across labs, and its terms allow the link but not the copy — its figures live on its own pages.
| Version | Released | Context | Input price | Output price | Terminal-Bench 4.0 (agentic coding) | FrontierCode 1.1, Main (agentic coding) | CursorBench 4.0 (agentic coding) | GDPval-AA v2.1 (knowledge work, Elo) | OSWorld 2.1 (computer use) | Humanity's Last Exam (multidisciplinary reasoning) |
|---|---|---|---|---|---|---|---|---|---|---|
| Claude Sonnet 5.5 — current | September 28, 2026 | 1M tokens | $2 / MTok | $10 / MTok | 70.6% | 46.2% (max effort) | 55.5% | 1844 | 80.1% (partial) | 64.5% (with tools) |
| Claude Sonnet 5 | — | — | $2 / MTok | $10 / MTok | 10.3% | 42.4% | 34.1% | 1449 | 57.0% (partial) | — |
| Claude Sonnet 4.6 | — | — | $3 / MTok | $15 / MTok | — | — | — | — | — | — |
Claude Sonnet 5.5 — current
- Model id:
claude-sonnet-5-5
Abilities
- Released: September 28, 2026 — Anthropic — Claude Sonnet 5.5 model page (read 2026-10-19).
- Context window: 1M tokens — Anthropic — Claude Sonnet 5.5 model page (read 2026-10-19).
- Max output: 128K tokens — Anthropic — Claude Sonnet 5.5 model page (read 2026-10-19).
- Output speed: —
- Latency: Fast (comparative); outputs 30%+ faster than Sonnet 5 — Anthropic — Claude Sonnet 5.5 announcement (read 2026-10-19).
Costs
- Input price: $2 / MTok — Anthropic — pricing (read 2026-10-19). Batch API 50% off; cache reads $0.10
- Output price: $10 / MTok — Anthropic — pricing (read 2026-10-19).
Benchmarks — as reported by the lab
- Terminal-Bench 4.0 (agentic coding) — 70.6% (Anthropic — Claude Sonnet 5.5 announcement, read 2026-10-19)
- FrontierCode 1.1, Main (agentic coding) — 46.2% (max effort) (Anthropic — Claude Sonnet 5.5 announcement, read 2026-10-19)
- CursorBench 4.0 (agentic coding) — 55.5% (Anthropic — Claude Sonnet 5.5 announcement, read 2026-10-19)
- GDPval-AA v2.1 (knowledge work, Elo) — 1844 (Anthropic — Claude Sonnet 5.5 announcement, read 2026-10-19)
- OSWorld 2.1 (computer use) — 80.1% (partial) (Anthropic — Claude Sonnet 5.5 announcement, read 2026-10-19)
- Humanity's Last Exam (multidisciplinary reasoning) — 64.5% (with tools) (Anthropic — Claude Sonnet 5.5 announcement, read 2026-10-19)
This is the version the seat lines above run.
Claude Sonnet 5
- Model id: —
Abilities
- Released: —
- Context window: —
- Max output: —
- Output speed: —
- Latency: —
Costs
- Input price: $2 / MTok — Anthropic — pricing (read 2026-10-19).
- Output price: $10 / MTok — Anthropic — pricing (read 2026-10-19).
Benchmarks — as reported by the lab
- Terminal-Bench 4.0 (agentic coding) — 10.3% (Anthropic — Claude Sonnet 5.5 announcement, read 2026-10-19)
- CursorBench 4.0 (agentic coding) — 34.1% (Anthropic — Claude Sonnet 5.5 announcement, read 2026-10-19)
- FrontierCode 1.1, Main (agentic coding) — 42.4% (Anthropic — Claude Sonnet 5.5 announcement, read 2026-10-19)
- GDPval-AA v2.1 (knowledge work, Elo) — 1449 (Anthropic — Claude Sonnet 5.5 announcement, read 2026-10-19)
- OSWorld 2.1 (computer use) — 57.0% (partial) (Anthropic — Claude Sonnet 5.5 announcement, read 2026-10-19)
Claude Sonnet 4.6
- Model id: —
Abilities
- Released: —
- Context window: —
- Max output: —
- Output speed: —
- Latency: —
Costs
- Input price: $3 / MTok — Anthropic — pricing (read 2026-10-19).
- Output price: $15 / MTok — Anthropic — pricing (read 2026-10-19).
Sources
- Anthropic — Claude Sonnet 5.5 model page — read 2026-10-19
- Anthropic — pricing — read 2026-10-19
- Anthropic — Claude Sonnet 5.5 announcement — read 2026-10-19
Figures refreshed 2026-10-19; benchmarks are vendor-reported. Prices are per million tokens unless a figure says otherwise.