Rerank 3.5 vs Granite Embedding 278M Multilingual
Side-by-side API pricing comparison · Cohere vs IBM
🏆
Rerank 3.5 is 430% cheaper on blended cost ($0.020 vs $0.106/Mtok)
Rerank 3.5
by Cohere
Current embeddingInput
$0.020/Mtok
Output
$0.020/Mtok
✓ Cheaper
| Blended avg | $0.020/Mtok |
|---|---|
| Context | — tokens |
| Modality | text |
| Parameters | Proprietary |
| Released | Mar 1, 2025 |
Granite Embedding 278M Multilingual
by IBM
Current embedding Open weightsInput
$0.106/Mtok
Output
$0.106/Mtok
| Blended avg | $0.106/Mtok |
|---|---|
| Context | — tokens |
| Modality | text |
| Parameters | 278M |
| Released | Jun 1, 2025 |
Cost at scale — 1M tokens (50/50 input/output)
| Volume | Rerank 3.5 | Granite Embedding 278M Multilingual | Savings |
|---|---|---|---|
| 1M tokens | $0.02 | $0.11 | $0.09 (84.9%) |
| 10M tokens | $0.2 | $1.06 | $0.86 (81.1%) |
| 100M tokens | $2 | $10.6 | $8.6 (81.1%) |
| 1000M tokens | $20 | $106 | $86 (81.1%) |
Summary
Rerank 3.5 by Cohere costs $0.020/Mtok input and $0.020/Mtok output, with a —-token context window. It supports text input.
Granite Embedding 278M Multilingual by IBM costs $0.106/Mtok input and $0.106/Mtok output, with a —-token context window. It supports text input.
On a blended cost basis, Rerank 3.5 is 430% cheaper than Granite Embedding 278M Multilingual.
Note: Pricing is per million tokens. Actual costs vary with usage patterns, prompt caching, and batch discounts. Always verify against official provider pricing pages.