The first thing that jumps out when you compare xAI’s Grok line to OpenAI’s GPT‑5‑nano and Anthropic’s Claude‑3‑haiku is how sharply the output prices diverge. Grok‑build‑0.1 charges $2 per 1M output tokens, while GPT‑5‑nano asks only $0.4 and Claude‑3‑haiku $1.25. Input prices follow a similar pattern: Grok‑build‑0.1 is $1, GPT‑5‑nano $0.05, and Claude‑3‑haiku $0.25.
### A single‑table snapshot
| Vendor | Model | Input | Cached input | Output |
|--------|-------|-------|--------------|--------|
| xAI | grok‑build‑0.1 | $1 per 1M tokens | $0.2 per 1M tokens | $2 per 1M tokens |
| OpenAI | gpt‑5‑nano | $0.05 per 1M tokens | $0.005 per 1M tokens | $0.4 per 1M tokens |
| Anthropic | claude‑3‑haiku | $0.25 per 1M tokens | $0.03 per 1M tokens | $1.25 per 1M tokens |
The table shows that for a workload that reads a lot of prompts but produces modest replies, OpenAI’s input price is a decisive advantage. But when the bulk of the token budget is spent on generating responses, GPT‑5‑nano’s output rate becomes the dominant factor, keeping it far cheaper than both Grok and Claude. Claude sits in between, offering a moderate input price and a middle‑range output rate, with a modest cached‑input discount that can add up for repeated prompt use.
### Who wins in practice
If your application is a data‑driven analytics engine that pushes millions of log snippets into a model and only asks for short, single‑line answers, the low input cost of GPT‑5‑nano makes it the obvious choice. For teams that need a balance between ingestion and output, especially when they reuse prompts, Claude‑3‑haiku provides a reasonable compromise, thanks to its $0.25 input and $1.25 output prices and a cached‑input discount. Grok’s higher rates are justified only when the model’s newer architecture delivers a clear functional advantage that outweighs the $2 output cost and the $1 input cost.
The recommendation is simple: for pure generation at scale, pick GPT‑5‑nano; for a middle ground with better prompt reuse economics, choose Claude‑3‑haiku; reserve Grok for niche cases where its capabilities are worth the premium.
Pricing verified on 2026-09-12.