AI
All toolsLLM Pricing Comparator
Compare model pricing for input and output tokens across providers.
Usage estimate
Prices in USD per 1M tokens. Reference only; check provider docs for current rates.
Cost ranking
| Model | Provider | $/1M in | $/1M out | Estimated cost |
|---|---|---|---|---|
| GPT-4o mini | OpenAI | $0.15 | $0.60 | $0.0270 |
| DeepSeek V3 | DeepSeek | $0.27 | $1.10 | $0.0490 |
| GPT-5 mini | OpenAI | $0.25 | $1.25 | $0.0500 |
| Gemini 2.5 Flash | $0.30 | $2.50 | $0.0800 | |
| Llama 3.3 70B | Meta (Together) | $0.88 | $0.88 | $0.1056 |
| Claude Haiku 4.5 | Anthropic | $1.00 | $5.00 | $0.2000 |
| Mistral Large 2 | Mistral | $2.00 | $6.00 | $0.3200 |
| Gemini 2.5 Pro | $1.25 | $10.00 | $0.3250 | |
| GPT-4o | OpenAI | $2.50 | $10.00 | $0.4500 |
| Claude Sonnet 4.6 | Anthropic | $3.00 | $15.00 | $0.6000 |
| GPT-5 | OpenAI | $10.00 | $30.00 | $1.6000 |
| Claude Opus 4.7 | Anthropic | $15.00 | $75.00 | $3.0000 |
Frequently asked questions
- Why is output more expensive than input?
- Generating tokens is autoregressive and uses far more compute per token than reading the prompt, which is processed in parallel. Output prices typically run 3-5x the input rate.
- How is pricing usually quoted?
- Per million tokens, split into input and output rates, sometimes with extra tiers for cached input, batch jobs, or extended context. Multiply rate by tokens then divide by one million.
- Do cheaper models always cost less overall?
- Not necessarily. A weaker model often needs more retries, longer prompts, or larger few-shot examples to hit the same quality, which can erase the per-token savings.