Anthropic makes Claude Sonnet 5's introductory pricing permanent at $2/1M input tokens and $10/1M output tokens, canceling its planned increase on September 1
We're making Claude Sonnet 5's introductory pricing permanent. We launched Sonnet 5 in June at $2 per million input tokens and $10 per million output tokens through August 31, and that price will remain unchanged.
@claudeaiClaude
Context & Ripple Effects
Anthropic launched Sonnet 5 with a temporary $2-per-million input-token and $10-per-million output-token rate, alongside a stated rise to $3 and $15 after August. It is now making that introductory Sonnet 5 rate the standing price instead.
The decision preserves a substantially lower price tier than Anthropic's $10/$50 pricing for Fable 5 and Mythos 5, sharpening the role of Sonnet in its model lineup.
First-order effects
Sonnet 5 customers avoid the planned 50% increase in both input and output token rates, keeping existing workload budgets and usage economics intact.
Anthropic forgoes the announced September repricing and establishes $2/$10 per million tokens as Sonnet 5's ongoing commercial rate.
Second-order effects
Developers building high-volume applications on Sonnet 5 can price products and commit usage with a stable token-cost baseline, rather than planning around a near-term step-up.
The lower permanent Sonnet tier makes Anthropic's own higher-priced Fable 5 and Mythos 5 offerings a clearer premium option on token price, increasing the importance of performance differentiation between them.
Third-order effects
Keeping launch pricing in place makes inference cost a more durable product-positioning lever: model vendors will increasingly be judged on the cost of useful workloads, not just published capability claims.
If permanent introductory rates become common, AI application economics will depend more on providers' ability to sustain capacity and efficiency at scale than on temporary launch discounts.
The trend: Frontier-model vendors are turning token pricing from a launch promotion into a durable lever for competing on application-level unit economics.
This means nothing because Sonnet 5's cost per task is absurd! On average it's more than TWICE as expensive as Opus 5!! And it's a significantly worse model. Absolute insanity from Anthropic I have no idea what tf they're doing. [image]
why would anyone choose @claudeai sonnet 5 at $2/$10 per million tokens when: @openai gpt-5.6 luna: $0.20/$1.20 @deepseek_ai v4 flash 0731: $0.14/$0.28 @Meta muse spark contributor: $0.10/$0.20 sonnet is 8x-50x more expensive on output tokens. and from my experience, these
I still can't quite make sense of Sonnet 5 pricing. GPT-5.6 Sol comes in cheaper, and if I specifically want Claude, Opus 5 is only ~$0.60 more. So what's the case for Sonnet here? 🤔 [image]
But but but people said prices wouldn't come down and tokens would only get more expensive and now Anthropic and OpenAI are lower prices of models that's simply not possible