Anthropic: Claude Opus 4.5 vs OpenAI: o3
OpenAI: o3 is cheaper than Anthropic: Claude Opus 4.5 by about 65% on a blended (75% input / 25% output) basis. Anthropic: Claude Opus 4.5 charges $5.000000 / $25.000000 per 1M input/output tokens; OpenAI: o3 charges $2.000000 / $8.000000.
| Field | Anthropic: Claude Opus 4.5 | OpenAI: o3 |
|---|---|---|
| Input $/Mtok | $5.000000 | $2.000000 |
| Output $/Mtok | $25.000000 | $8.000000 |
| Cache read $/Mtok | $0.500000 | $0.500000 |
| Cache write $/Mtok | $6.250000 | – |
| Input $/Mchar | n/a | 0.563428 |
| Output $/Mchar | n/a | 2.253712 |
| Context window | 200000 | 200000 |
| Tokenizer | claude | 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 | Anthropic: Claude Opus 4.5 | OpenAI: o3 |
|---|---|---|
| Coding agent | $60.00 | $23.00 |
| RAG chat | $80.00 | $29.00 |
| Summarisation (batch) | $70.00 | $26.00 |
| Classification | $54.00 | $21.20 |
| Chat assistant | $130.00 | $44.00 |
| Content generation | $210.00 | $68.00 |
Cite this page
Anthropic: Claude Opus 4.5 vs OpenAI: o3: $5.000000/$25.000000 vs $2.000000/$8.000000 per 1M tokens, as of 2026-10-04 (Token Barometer, https://tokenbarometer.com/compare/anthropic_claude-opus-4-5-vs-openai_o3).
Anthropic: Claude Opus 4.5 page | OpenAI: o3 page | Open in calculator