Anthropic: Claude Haiku 4.5 vs Meta: Llama 3.3 70B Instruct
Meta: Llama 3.3 70B Instruct is cheaper than Anthropic: Claude Haiku 4.5 by about 100% on a blended (75% input / 25% output) basis. Anthropic: Claude Haiku 4.5 charges $1.000000 / $5.000000 per 1M input/output tokens; Meta: Llama 3.3 70B Instruct charges $0.000000 / $0.000000.
| Field | Anthropic: Claude Haiku 4.5 | Meta: Llama 3.3 70B Instruct |
|---|---|---|
| Input $/Mtok | $1.000000 | $0.000000 |
| Output $/Mtok | $5.000000 | $0.000000 |
| Cache read $/Mtok | $0.100000 | – |
| Cache write $/Mtok | $1.250000 | – |
| Input $/Mchar | n/a | 0.000000 |
| Output $/Mchar | n/a | 0.000000 |
| Context window | 200000 | 128000 |
| Tokenizer | claude | llama3 |
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 Haiku 4.5 | Meta: Llama 3.3 70B Instruct |
|---|---|---|
| Coding agent | $12.00 | $0.00 |
| RAG chat | $16.00 | $0.00 |
| Summarisation (batch) | $14.00 | $0.00 |
| Classification | $10.80 | $0.00 |
| Chat assistant | $26.00 | $0.00 |
| Content generation | $42.00 | $0.00 |
Cite this page
Anthropic: Claude Haiku 4.5 vs Meta: Llama 3.3 70B Instruct: $1.000000/$5.000000 vs $0.000000/$0.000000 per 1M tokens, as of 2026-10-04 (Token Barometer, https://tokenbarometer.com/compare/anthropic_claude-haiku-4-5-vs-meta_llama-3.3-70b-instruct).
Anthropic: Claude Haiku 4.5 page | Meta: Llama 3.3 70B Instruct page | Open in calculator