OpenAI: o3 vs Grok 4
OpenAI: o3 is cheaper than Grok 4 by about 42% on a blended (75% input / 25% output) basis. OpenAI: o3 charges $2.000000 / $8.000000 per 1M input/output tokens; Grok 4 charges $3.000000 / $15.000000.
| Field | OpenAI: o3 | Grok 4 |
|---|---|---|
| Input $/Mtok | $2.000000 | $3.000000 |
| Output $/Mtok | $8.000000 | $15.000000 |
| Cache read $/Mtok | $0.500000 | – |
| Cache write $/Mtok | – | – |
| Input $/Mchar | 0.563428 | n/a |
| Output $/Mchar | 2.253712 | n/a |
| Context window | 200000 | 256000 |
| Tokenizer | o200k | – |
Price history
Same workload, monthly cost
Illustrative monthly cost at 10M tokens/month, split per workload preset. Not a substitute for the calculator, which uses your own volume, cache and batch assumptions.
| Workload | OpenAI: o3 | Grok 4 |
|---|---|---|
| Coding agent | $23.00 | $36.00 |
| RAG chat | $29.00 | $48.00 |
| Summarisation (batch) | $26.00 | $42.00 |
| Classification | $21.20 | $32.40 |
| Chat assistant | $44.00 | $78.00 |
| Content generation | $68.00 | $126.00 |
Cite this page
OpenAI: o3 vs Grok 4: $2.000000/$8.000000 vs $3.000000/$15.000000 per 1M tokens, as of 2026-10-04 (Token Barometer, https://tokenbarometer.com/compare/openai_o3-vs-xai_grok-4-0709).