Skip to content

Pricing Watch|Claude Sonnet 5 Price Hike Canceled — $2/$10 Becomes Permanent

Aug 17, 20261 min
TL;DRClaude Sonnet 5 was set to jump from its promo price of $2/$10 to $3/$15 on 9/1. On 8/10 Anthropic updated its pricing page to confirm the increase 'will not happen' — $2/$10 is now the permanent price. For a workload of 300K customer-service conversations per month, that avoids a $1,200/month cost increase (↓33%), and means Sonnet 5 is now permanently cheaper than its predecessor Sonnet 4.6 ($3/$15).

🌏 中文版

What Changed

When Claude Sonnet 5 launched in June, Anthropic explicitly labeled the $2/$10 (per million tokens input/output) pricing as an "introductory offer" — set to expire on 8/31, with the standard rate of $3/$15 taking effect on 9/1, matching the current price of the previous-generation Sonnet 4.6. Anthropic has now updated its official pricing page to announce that this increase "will not happen": the promo price is now the permanent price. In a month dominated by price hikes elsewhere (DeepSeek raised rates across the board on 8/16, AI demand continues to strain supply), this is a rare move in the opposite direction — Anthropic chose to lock in pricing on its workhorse production model, effectively canceling a planned 33% increase and making the newer model permanently cheaper than its predecessor, breaking the assumption that "newer models should cost more."

Before vs. After

ItemPromo Price (current, now permanent)Originally Planned 9/1 Standard Price (canceled)Avoided IncreaseOriginal Effective Date
Sonnet 5 Input$2.00/1M tokens$3.00/1M tokens50%Originally 2026-09-01, canceled
Sonnet 5 Output$10.00/1M tokens$15.00/1M tokens50%Originally 2026-09-01, canceled
Sonnet 5 Cache Hit Input$0.20/1M tokens (0.1x)No post-promo rate announcedPromo rate confirmed permanent—
Sonnet 5 5m Cache Write$2.50/1M tokensNo post-promo rate announcedPromo rate confirmed permanent—

Cost Estimate

Scenario: An agent handling 10,000 customer-service conversations per day (averaging 1,500 input tokens + 500 output tokens each), assuming continued use of Sonnet 5 from September onward.

Originally Planned 9/1 (if hike proceeded)Actual (after cancellation, price unchanged)Savings
Input cost/month (450M tokens)$1,350$900$450
Output cost/month (150M tokens)$2,250$1,500$750
Total$3,600/month$2,400/month$1,200 (↓33%)

Impact for Developers and Enterprises

Who Benefits Most

Teams already running Sonnet 5 in production that had budgeted for the 50% September increase benefit most directly — no renegotiations needed, no pricing model adjustments, the savings just land. Teams that were holding off on scaling their Sonnet 5 deployments because of the upcoming price hike now have one fewer reason to wait.

Competitive Landscape

Post-cancellation output pricing comparison for major models (USD/1M tokens):

ModelOutputNotes
GPT-5.6 Luna$1.20OpenAI 80% price cut on 7/30
DeepSeek V4-Pro (off-peak)$1.98Still one of the cheapest frontier options off-peak after 8/16 increase
Gemini 3.7 Flash (promo)$3.75Promo until 2026-12-31, doubles to $7.50 after
Claude Haiku 4.5$5.00—
Claude Sonnet 5$10.00Permanent price, no 9/1 hike risk
GPT-5.6 Terra$12.00—
Claude Sonnet 4.6 (previous gen)$15.00Sonnet 5 now permanently 33% cheaper than its predecessor
Claude Opus 5$25.00—

The most direct effect of the canceled hike is on Anthropic's own product line positioning: Sonnet 5 was supposed to converge to Sonnet 4.6's $15 output price in September. Instead, it stays permanently one-third cheaper than the previous generation — bundling "use the newer model" with "use the cheaper model" rather than forcing customers to choose between the two.

Action Items

  • If you're already running Sonnet 5 in production: no action needed — just delete the 9/1 price hike from your budget calendar.
  • If you're still on Sonnet 4.6: the case for switching to Sonnet 5 is now stronger and more durable — it's not just a newer model, it's permanently 33% cheaper, and you no longer have to rush a decision before the 8/31 promo window closes.
  • If you were delaying scale-up because of the planned September increase: that blocker is gone. You can build long-term architecture plans on the $2/$10 baseline.

Takeaway

Most pricing news covers "what changed." But the real story here is "something that was scheduled to happen didn't" — a publicly announced price increase was canceled. On the surface, "nothing changed," but for budget planning purposes, this matters just as much as an actual price cut, because it eliminates a future cost that was already locked in and just hadn't taken effect yet. This kind of "non-event" announcement is especially easy to miss, because headlines track things that happen, not things that were supposed to happen but didn't.

References