Claude Sonnet 5: Pricing, What Changed from Sonnet 4.6, and How to Test It on Your Prompts
← All postsFix Your Prompts

Claude Sonnet 5: Pricing, What Changed from Sonnet 4.6, and How to Test It on Your Prompts

July 5, 2026·FixMyPrompt Team·8 min read
Share

Claude Sonnet 5 is Anthropic's new speed-and-intelligence model, at $3/$15 per million tokens with an introductory rate of $2/$10 through August 2026. Here is what changed from Sonnet 4.6, the exact pricing and limits, and how to run your own prompt against it before you commit.

#claude sonnet 5#claude sonnet 5 pricing#sonnet 5 vs sonnet 4.6#claude sonnet 5 review#claude sonnet 5 context window#claude sonnet 5 api#is claude sonnet 5 good#claude sonnet 5 vs opus

Anthropic released Claude Sonnet 5 (claude-sonnet-5) as the new middle tier of the Claude lineup, the one it describes as the best combination of speed and intelligence. If you were using Sonnet 4.6 for everyday work, this is the model that replaces it at the top of your list. Here is what actually changed, what it costs, and how to see its output on your own prompt before you switch anything.

The pricing, plainly

Sonnet 5 lists at $3 per million input tokens and $15 per million output tokens. That is the same headline rate as Sonnet 4.6, so a straight switch does not change your per-token math.

The part worth knowing is the introductory rate. Through August 31, 2026, Sonnet 5 runs at $2 input and $10 per million output, roughly a third cheaper than the standard rate. After that date it returns to $3/$15. If you have volume to move, the window before September is the cheap time to do it.

What changed from Sonnet 4.6

The context window and output cap are unchanged: a 1 million token context and up to 128,000 output tokens on the standard Messages API. So anything that fit in Sonnet 4.6 fits here.

Two differences matter in practice:

  • Adaptive thinking is on, and effort defaults to high. On the Claude API and in Claude Code, Sonnet 5 sets its effort level to high unless you tell it otherwise. That buys better answers on harder prompts, and it can also mean slower, more expensive responses on simple ones than you saw by default on older Sonnets. If you run high volume of easy calls, set the effort down.
  • A newer knowledge cutoff. Sonnet 5 has reliable knowledge through January 2026, later than Sonnet 4.6. For anything time-sensitive that still is not enough, and you should give the model the current facts in the prompt rather than trusting recall.

Anthropic has not published head-to-head numbers that we would repeat as fact, so the honest move is to judge it on your own work rather than a benchmark chart. Which brings us to the useful part.

How to test Sonnet 5 on your own prompt in FixMyPrompt

A model comparison only means something on the prompt you actually run. FixMyPrompt lets you take your prompt and run it against Sonnet 5 directly, so you see the real output instead of a leaderboard.

  1. Open your prompt in FixMyPrompt.
  2. Pick Claude Sonnet 5 from the model list in the test run.
  3. Run it, and read the output for your actual task. Run it against Sonnet 4.6 or another model in the same place if you want a side by side.

It is pay as you go, so a test run costs a small number of credits rather than a subscription. You get the answer to the only question that matters, which is whether Sonnet 5 writes a better response to your prompt than what you use now, before you rewire anything downstream.

Should you switch?

If Sonnet 4.6 was your default, Sonnet 5 is the natural upgrade at the same standard price, and cheaper until September. The one thing to watch is the high effort default, which can quietly raise cost and latency on lightweight calls. Set effort deliberately, test your real prompts, and let the output decide.

Run your prompt against Claude Sonnet 5 in FixMyPrompt. No signup needed to try it.

Found this useful? Share it.

Share

Read next

Run a free QA on your own prompt

Get a structured score, specific issues, and a rewritten prompt in seconds.

Run free QA