When you line up Anthropic’s Claude Sonnet 5 against OpenAI’s GPT‑5 nano, the token‑meter reads like night‑and‑day. Claude Sonnet 5 charges $2 per 1M input tokens, $0.2 per 1M cached input tokens and $10 per 1M output tokens, while GPT‑5 nano asks for $0.05 per 1M input, $0.005 per 1M cached input and $0.4 per 1M output. That makes the OpenAI offering roughly 40 × cheaper on input and 25 × cheaper on output, a gap that can flip a budget in seconds.
### Where the dollars actually sit
The table below strips away the marketing fluff and shows the pure per‑token rates for the two contenders. Anthropic’s pricing is uniform across the Sonnet 5 family, and OpenAI’s nano tier is the cheapest entry in the GPT‑5 lineup.
| Vendor | Model | Input | Cached input | Output |
|---|---|---|---|---|
| Anthropic | claude‑sonnet‑5 | $2 per 1M tokens | $0.2 per 1M tokens | $10 per 1M tokens |
| OpenAI | gpt‑5‑nano | $0.05 per 1M tokens | $0.005 per 1M tokens | $0.4 per 1M tokens |
### Who should pick which side of the price fence
If your product demands the highest quality output, tighter guardrails and you can absorb a premium, Claude Sonnet 5’s higher output fee may be justified – especially when you can cache a lot of prompt material to lean on the $0.2 cached‑input rate. For most developers, startups and anyone watching the burn rate, GPT‑5 nano delivers comparable latency at a fraction of the cost, making it the pragmatic choice for rapid iteration and scale. The decision comes down to whether you value Anthropic’s reputation for safety enough to pay a steep output surcharge, or you prioritize raw cost efficiency and are comfortable with OpenAI’s broader ecosystem.
Pricing verified on 2026-09-10.