LLM pricing comparison shows cache costs can dominate
A side-by-side pricing sheet across about 7 AI s shows that the listed token price is not enough to estimate real cost. The comparison covers providers such as OpenRouter, DeepSeek, Together AI, Fireworks, and Groq, using public pricing pages and public APIs. It tracks input token prices, output token prices, , cached input prices, supported models, and provider-specific price differences.
It is not a speed or throughput benchmark, because no latency tests or load tests were run. The biggest finding is that cache pricing can change the cost picture sharply. For across different providers, a cache hit can make input tokens tens of times cheaper than the same input without cache use.
This matters for agents with large , RAG setups that reuse the same background material, multi-turn chats, and repeated s. The same model can also cost several times more or less depending on which provider serves it.
Key points
- The sheet compares public pricing from providers including OpenRouter, DeepSeek, Together AI, Fireworks, and Groq.
- It covers input and output token prices, , cached input prices, supported models, and provider-level price differences.
- It is not a performance benchmark, because no latency or throughput tests were run.
- For , cached input can be tens of times cheaper than uncached input depending on the provider.
- Agents, RAG setups, multi-turn chats, and repeated s may save more from cache policy than from a lower headline token price.