Claude Sonnet 5.5 (Sep 28, 2026): $2/$10, Platform breaks, Claude Code 2.1.284 default Sonnet

Sep 28 launch: Sonnet 5.5 at $2/$10 with $0.20 cache reads; five Platform migration breaks; Claude Code 2.1.284 default Sonnet + sessions start in auto mode.

On Sep 28, 2026, Anthropic launched Claude Sonnet 5.5 (claude-sonnet-5-5): same list price $2 / $10 per MTok and cache reads $0.20. Five Platform breaks vs Sonnet 5 (thinking between_tools not disabled; tool_choice any/tool → 400; account-/conversation-bound thinking blocks; computer_20251124 rejected on API/GCP; advisor rejects Opus 4.8 / 4.7 / Sonnet 5). Claude Code 2.1.284 makes it the default Sonnet and starts interactive/VS Code sessions in auto mode when no permission mode is set. Thin same-day N — soft bridge Opus 5.5; link auto mode. CASCADE auto-mode lean patch is after this N is live HTTP 200 — not today. Labs volumes for Sonnet 5.5 KWs: [GAP] (no Labs item).

Pricing and role

Product card pricing table (vs Opus 5.5):

Per 1M tokens	Sonnet 5.5	Opus 5.5
Input	$2	$4
Output	$10	$20
Cache reads	$0.20	$0.20
Cache writes	$2.50	$5

Same list price as Sonnet 5. Anthropic claims outputs 30%+ faster and up to 30% less cost per task vs Sonnet 5 because it typically needs fewer tokens — attributed product-card language, not a CF measurement. Positioning (thin): faster, lower-cost complement to Opus 5.5 for well-scoped everyday work, bugs, and polished docs/slides/spreadsheets; design eye. Haiku 5.5 is still “coming weeks” — launch date [GAP]. Optional one attributed bench line: Terminal-Bench 4.0 70.6% vs Sonnet 5 10.3% (product card). Model id on Platform: claude-sonnet-5-5 (API, Bedrock, Platform-on-AWS, GCP, Foundry).

Platform migration breaks

Platform RN September 28, 2026 + What’s new in Sonnet 5.5 — five breaks vs Sonnet 5 code:

Thinking off. Send thinking: {"type": "between_tools"} instead of "disabled" (at high effort or below). "disabled" → 400.
Forced tools. tool_choice types any / tool → 400. Use auto (with strict tool use) or none.
Thinking blocks. Tied to the model and conversation; Sonnet 5.5 blocks are also account-/conversation-bound (Preserved thinking). Another account → block dropped, request still succeeds.
Computer use. On Claude API and Google Cloud, earlier computer_20251124 is not accepted — migrate to the current computer toolset (computer_toolset_20260801 per what’s-new). On Amazon Bedrock, what’s-new states Sonnet 5.5 still accepts computer_20251124.
Advisor. Rejects Opus 4.8 / Opus 4.7 / Sonnet 5 as advisors for a Sonnet 5.5 executor.

Keep migration thin — follow Anthropic’s what’s-new + migration guide; no full agent-loop rewrite here.

Claude Code 2.1.284

Raw CHANGELOG head is ## 2.1.284. Must lines only:

Added Claude Sonnet 5.5 (claude-sonnet-5-5), now the default Sonnet — 1M context, $2/$10 per Mtok with $0.20/Mtok cache reads.
Changed interactive terminal and VS Code sessions to start in auto mode when no permission mode is configured; permissions.defaultMode still overrides — see How do I stop Claude Code auto mode without confirmation? and How do I configure Claude Code permissions?.

Do not dump the rest of the 2.1.284 UI/hooks/VS Code lists into this N. Soft note for HEAD: CASCADE lean patch on the live auto-mode Spec (+ soft permissions / effort / rate-limits) ships after this N is HTTP 200 — not a deliverable in this fiche.

Upgrade readout

Operator	Action
API / GCP integrators	Migrate thinking-off → `between_tools`; drop forced `tool_choice`; move computer use off `computer_20251124`; update advisor pairings
Bedrock computer-use	`computer_20251124` still OK on Sonnet 5.5 (what’s-new)
Claude Code	Upgrade to ≥ 2.1.284; check `/model` and permission mode after Sonnet default + auto-mode default
Opus job owners	Soft bridge only — do not reopen Opus N / Opus CASCADE rows: Claude Opus 5.5

Optional operator spokes (live): max effort · rate limits.

Sources (checked 2026-09-29)

Anthropic — Claude Sonnet 5.5 — Sep 28 launch, $2/$10, cache reads $0.20 / writes $2.50, 30%+ faster / up to 30% less $/task claims, Terminal-Bench 4.0 70.6% vs 10.3%
Platform release notes — September 28, 2026: model id, five migration breaks, preserved/account-bound thinking
What’s new in Claude Sonnet 5.5 — break detail, Bedrock computer keep, advisor pairings
Claude Code CHANGELOG.md — ## 2.1.284 (default Sonnet + sessions start in auto mode)
Live: /guides/claude-opus-5-5-2026-09-22 · /guides/how-do-i-stop-claude-code-auto-mode-without-confirmation · /guides/how-do-i-configure-claude-code-permissions