Compare $/1M token pricing across 10 LLM APIs — plus self-host GPU costs.
Should you use an API or run your own model on GPU cloud? Compare per-million-token pricing across 10 major LLMs. Open-source models show both API pricing and the estimated GPU cost to self-host.
| Model | Input $/1M tok | Output $/1M tok | Self-Host GPU cost/hr |
|---|---|---|---|
Llama 3.1 8BOpen★ | $0.15 | $0.15 | L4 ~$0.5/hr |
Gemini 1.5 Flash | $0.07 | $0.30 | — |
GPT-4o mini | $0.15 | $0.60 | — |
Llama 3.1 70BOpen | $0.59 | $0.79 | A100 80GB ~$1.5/hr |
Claude 3 Haiku | $0.25 | $1.25 | — |
Qwen 2.5 72BOpen | $0.50 | $1.00 | A100 80GB ~$1.5/hr |
Gemini 1.5 Pro | $1.25 | $5.00 | — |
Mistral Large 2 | $2.00 | $6.00 | — |
GPT-4o | $2.50 | $10.00 | — |
Claude 3.5 Sonnet | $3.00 | $15.00 | — |
LLM API pricing changes weekly. New models launch, old ones get cheaper, and the per-token math is opaque (input vs output, context length, batch discounts). Most teams end up overpaying because they don't comparison-shop — they use whatever API they started with.
Our LLM Pricing Comparison tracks $/1M token pricing across 10 major LLM APIs (GPT-4o, Claude 3.5 Sonnet, Llama 3.1, Mistral Large, Gemini 1.5 Pro, DeepSeek V3, and more) — plus a self-host GPU cost comparison so you can see when it's cheaper to run your own.
Prices are updated regularly. Use this to decide whether to use an API or self-host, and which API to use for which workload.
Adapt open-source LLMs (Llama, Mistral, Qwen, DeepSeek) to your domain — with LoRA, QLoRA, instruction tuning, and safe deployment guardrails.
Compare H100, A100, L40S, and consumer GPU pricing across 8+ providers — benchmarked against your actual workload, not affiliate payouts.
Production retrieval-augmented generation with vector stores, hybrid search, reranking, citation, and observability — so you can trust the answers.
Our consultants can turn tool output into a scoped engagement. Book a free 30-minute call.
Talk to an expert