Anthropic 7/13 Claude Sonnet 4.5: mid-tier daily-driver with same $3/$15 pricing

3 sources

Anthropic on 2026-07-13 publishes Claude Sonnet 4.5 as the mid-tier flagship (between Haiku 4.5 and Opus 4.5); 200K context, native tool-use with extended-thinking mode; pricing unchanged at $3 input / $15 output per million tokens. Self-reported figures: SWE-Bench Verified 67.4, GPQA-Diamond 74.8, MMLU-Pro 82.9, BFCL v3 63.7, Terminal-Bench 2.1 74.2. Deployment via API + claude.ai + Bedrock + Vertex AI.

Why now

It lands one day after Anthropic's previous governance-pair content (07-13 hard-questions + LTBT Bernanke) and on the same day as five other vendor-aligned model drops — creators can pair the newsroom content with a model-line update to show the newsroom side AND the model side of Anthropic's week.

Why it is worth publishing

Best for daily-driver / cost-conscious creators. The same-price-as-prior framing is the structural hook; the Sonnet/Haiku/Opus tier positioning is the analyst hook.

Evidence basis

Same-day Decoder item (Independently verified) + same-pricing strategy + tier-positioning reveal. Three-stack heat.

“Anthropic just shipped a new Sonnet — same price as the last one, new spot in the tier stack.”

Angle

Frame Sonnet 4.5 as Anthropic's mid-tier repositioning (between Haiku 4.5 and Opus 4.5) at the same prior-Sonnet pricing — this is a tier-reframing move, not a price hike. Lead with the badge moment (Independently verified), then the tier positioning, then the same-pricing-as-Sonnet-4.5 fact.

Format

Carousel comparison + a single-card distillation

Demo idea

Three-card tier stack: row 1 'Haiku 4.5 (lightest)'; row 2 'Sonnet 4.5 (the new daily-driver — this card)'; row 3 'Opus 4.5 (deepest)'; columns: pricing, context, headline benchmark. This pairs naturally with the 07-13 governance pair to make a 'tier + governance' dual for the week.

Platform notes

Same-price-as-prior (medium risk): do not frame Sonnet 4.5 as a price cut OR a price hike — the $3/$15 anchor stays put. When viewers ask 'is Sonnet 4.5 better than Sonnet 4.5 (prior),' pin the link to the newsroom and quote only what the newsroom lists as deltas (vendor-stated, not reproduced).

Usable claims

Anthropic on 2026-07-13 published Claude Sonnet 4.5 as the mid-tier flagship positioning (between Haiku 4.5 and Opus 4.5); 200K context, native tool-use with extended-thinking mode; self-reported figures include SWE-Bench Verified 67.4, GPQA-Diamond 74.8, MMLU-Pro 82.9, BFCL v3 63.7, Terminal-Bench 2.1 74.2; pricing $3 input / $15 output per million tokens; deployment via API + claude.ai + Bedrock + Vertex AI; commercial-use license.

Evidence pipeline

From the news

Anthropic 7/13 Claude Sonnet 4.5: mid-tier flagship for everyday tasks at the same $3/$15 pricing

Breakdown

Walks the Sonnet 4.5 tier positioning (between Haiku 4.5 and Opus 4.5) and the same-pricing-as-prior-Sonnet framing (SWE-Bench Verified 67.4 / GPQA 74.8 / BFCL v3 63.7 / Terminal-Bench 2.1 74.2), paired with the second 'Independently verified' badge of the day from the Decoder tech-press item. Read alongside the 07-13 Anthropic governance pair for a 'governance + model' week framing.

Sources

Risks

Pin the link to each vendor page and the companion HF model card; quote only what those pages state; do not paraphrase self-reported benchmark numbers as third-party validation; for the Doubao topic specifically, pin the HF mirror card as a researcher-preview only and do not paraphrase the gating-to-API-only boundary.
Pin the link to each license text; quote only what the page states; do not paraphrase the baseline license as the sole operating condition; flag addenda by name without stating their scope; for Google Gemini 3.5 Pro, note that license lives with the API contract rather than a downloadable-weights license.

Demo ideas

Pair with the 07-13 Anthropic governance pair (hard-questions + LTBT Bernanke) — show 'this week's Anthropic beat' as governance + model-line, two layers, same vendor.
For long-form tier-by-tier coverage, build a Haiku 4.5 / Sonnet 4.5 / Opus 4.5 pricing-and-context matrix from the vendor pages only — no cross-vendor extrapolation.
Use the badge moment as the cluster hook: 'two of today's six topics surface the Independently verified badge — gpt-oss and Sonnet 4.5 — both because The Decoder picked them up today.'