Claude 4.5 Sonnet vs GPT-5: pricing & cost comparison
On input tokens, GPT-5 is the cheaper of the two — 58% less per million ($3 vs $1.25). On output, GPT-5 is 33% cheaper ($15 vs $10) — and since output is usually the dominant cost driver, that gap matters more than it looks.
Side by side
| Claude 4.5 Sonnet | GPT-5 | |
|---|---|---|
| Input / 1M tokens | $3 | $1.25 |
| Output / 1M tokens | $15 | $10 |
| Context window | 200,000 | 400,000 |
| Token-count accuracy | ±2% | exact |
| Cost — 10,000 input + 2,000 output tokens | $0.06 | $0.0325 |
What a real request costs
Take a representative turn — 10,000 input + 2,000 output tokens. Claude 4.5 Sonnet comes to $0.06, GPT-5 to $0.0325. Across 100,000 requests that's a $2750 swing in favour of GPT-5. To run the numbers on your actual prompt, paste it into the calculator and toggle Compare across all models.
Close call — decide on output cost
Claude 4.5 Sonnet and GPT-5 are close enough on price that the tie-breakers matter more than the headline rate. Output tokens are usually the dominant cost driver, so weigh $15 vs $10 per 1M output first, then context window — GPT-5 carries the larger one at 400,000 tokens. They're different vendors, so factor in a small tokenizer calibration buffer on whichever side isn't exact.
See the full breakdown on the dedicated pages for Claude 4.5 Sonnet and GPT-5.
FAQ
- Is Claude 4.5 Sonnet or GPT-5 cheaper?
- For a typical request (10,000 input + 2,000 output tokens), GPT-5 is cheaper — about 46% less, or roughly $2750 saved per 100,000 requests. Claude 4.5 Sonnet runs $3/$15 per 1M input/output tokens; GPT-5 runs $1.25/$10.
- Which has the larger context window?
- GPT-5, at 400,000 tokens versus 200,000.
- How accurate are these token counts?
- Claude 4.5 Sonnet: Approximated with cl100k_base — drift typically <2% on English and code. GPT-5: Exact tokenization via the canonical OpenAI vocab (o200k_base). The dollar math itself is exact once the token count is known.
Both prices are computed from tokenmath's verified pricing table. Rates sourced from www.anthropic.com and openai.com, verified 2026-05-09. Vendor pricing changes often — confirm before you commit.