Cohere rerankers - verified
Cohere: 5 reranker models, read from the provider's own pages on October 18, 2026. Cohere rerankers are billed in search units with no pay-as-you-go price stated in the page we read; Model Vault dedicated instances list from $5.00 per instance per hour.
Read from cohere.com, docs.cohere.com. Compare every reranker
Billing unit: search unit
“A single search unit is defined as one query with up to 100 documents to be ranked.”
Rerank documents are chunked internally; the models page says Rerank "will automatically chunk documents longer than 510 tokens" while the context column lists 4k or 32k. Billing unit quote: "A single search unit is defined as one query with up to 100 documents to be ranked." (cohere.com/pricing FAQ). The rerank-v2.0 family was shut down 2025-04-30 per the deprecations page.
rerank-v4.0-pro - API - Dedicated hourly only (Cohere Model Vault)
Note: Cohere bills Rerank in search units: "A single search unit is defined as one query with up to 100 documents to be ranked." cohere.com/pricing lists Model Vault hourly rates for Rerank but no per-search pay-as-you-go price on 2026-10-18, so it is shown as not stated.
Dedicated capacity
$5.00 per instance per hour (Model Vault), Medium
“Rerank 4 Pro Medium $5.00 $3,250”
$10.00 per instance per hour (Model Vault), Large
“Rerank 4 Pro Large $10.00 $6,500”
GA
“Rerank 4 is now available on Cohere’s Platform”
32,000 tokens
context length (models table lists 32k)
“rerank-v4.0-pro · A multilingual model that allows for re-ranking English and non-english documents and semi-structured data (JSON). This model is better suited for state-of-the-art quality and complex use-cases than its fast variant. · Text · 32k”
1,000
recommended maximum, not a hard limit
“For optimal performance we recommend against sending more than 1,000 documents in a single request.”
Over 100 languages
“Users can deploy Rerank in over 100 world languages, including state-of-the-art retrieval in 10 major business languages”
No (API only) inferred
Note: No official open weights are listed in the sources checked; the provider documents this as an API model.
4,096 tokens
“Defaults to 4096. Long documents will be automatically truncated to the specified number of tokens.”
Pro is the quality-focused Rerank 4 variant; the max_tokens_per_doc request parameter defaults to 4096, so long documents are truncated unless it is raised.
rerank-v4.0-fast - API - Dedicated hourly only (Cohere Model Vault)
Note: Cohere bills Rerank in search units: "A single search unit is defined as one query with up to 100 documents to be ranked." cohere.com/pricing lists Model Vault hourly rates for Rerank but no per-search pay-as-you-go price on 2026-10-18, so it is shown as not stated.
Dedicated capacity
$5.00 per instance per hour (Model Vault), Medium
“Rerank 4 Fast Medium $5.00 $3,250”
GA
“Rerank 4 is now available on Cohere’s Platform”
32,000 tokens
context length (models table lists 32k)
“This model is better suited for low latency and high throughput use-cases than its pro variant. · Text · 32k”
1,000
recommended maximum, not a hard limit
“For optimal performance we recommend against sending more than 1,000 documents in a single request.”
Over 100 languages
“Users can deploy Rerank in over 100 world languages, including state-of-the-art retrieval in 10 major business languages”
No (API only) inferred
Note: No official open weights are listed in the sources checked; the provider documents this as an API model.
4,096 tokens
“Defaults to 4096. Long documents will be automatically truncated to the specified number of tokens.”
Fast is the latency-focused Rerank 4 variant.
rerank-v3.5 - API - Dedicated hourly only (Cohere Model Vault)
Note: Cohere bills Rerank in search units: "A single search unit is defined as one query with up to 100 documents to be ranked." cohere.com/pricing lists Model Vault hourly rates for Rerank but no per-search pay-as-you-go price on 2026-10-18, so it is shown as not stated.
Dedicated capacity
$5.00 per instance per hour (Model Vault), Medium
“Rerank 3.5 Medium $5.00 $3,250”
GAinferred
“rerank-v3.5 · cohere.rerank-v3-5:0 · Unique per deployment · Cohere-rerank-v3.5 · cohere.rerank.3-5”
4,096 tokens
per query plus document
“This model has a context length of 4096 tokens.”
1,000
recommended maximum, not a hard limit
“For optimal performance we recommend against sending more than 1,000 documents in a single request.”
No (API only) inferred
Note: No official open weights are listed in the sources checked; the provider documents this as an API model.
Cohere's models table describes rerank-v3.5 as re-ranking "English Language documents", the same wording as rerank-english-v3.0; languages are left null rather than guessed. Also offered on Amazon Bedrock as cohere.rerank-v3-5:0.
rerank-english-v3.0 - API - Dedicated hourly only (Cohere Model Vault)
The provider's pages state no price for this model.
GAinferred
Status inferred: the source lists the model but does not label it.
4,096 tokens
per query plus document
“This model has a context length of 4096 tokens.”
1,000
recommended maximum, not a hard limit
“For optimal performance we recommend against sending more than 1,000 documents in a single request.”
English
“rerank-english-v3.0 · A model that allows for re-ranking English Language documents and semi-structured data (JSON).”
No (API only) inferred
Note: No official open weights are listed in the sources checked; the provider documents this as an API model.
Listed on the models page and absent from the deprecations page on 2026-10-18; no Model Vault rate is listed for it.
rerank-multilingual-v3.0 - API - Dedicated hourly only (Cohere Model Vault)
The provider's pages state no price for this model.
GAinferred
Status inferred: the source lists the model but does not label it.
4,096 tokens
per query plus document
“This model has a context length of 4096 tokens.”
1,000
recommended maximum, not a hard limit
“For optimal performance we recommend against sending more than 1,000 documents in a single request.”
Same languages as embed-multilingual-v3.0 (non-English)
“A model for documents and semi-structure data (JSON) that are not in English. Supports the same languages as embed-multilingual-v3.0.”
No (API only) inferred
Note: No official open weights are listed in the sources checked; the provider documents this as an API model.
Listed on the models page and absent from the deprecations page on 2026-10-18.
Rerank 4 Pro: pay-as-you-go price per search unit not stated; Model Vault Medium $5.00 per instance per hour (Model Vault), Large $10.00 per instance per hour (Model Vault). Rerank 4 Fast: pay-as-you-go price per search unit not stated; Model Vault Medium $5.00 per instance per hour (Model Vault). Rerank 3.5: pay-as-you-go price per search unit not stated; Model Vault Medium $5.00 per instance per hour (Model Vault). Rerank English 3.0: pay-as-you-go price per search unit not stated, and no Model Vault rate listed. Rerank Multilingual 3.0: pay-as-you-go price per search unit not stated, and no Model Vault rate listed. Prices are in Cohere's own units as of October 18, 2026. Billing unit: "A single search unit is defined as one query with up to 100 documents to be ranked."
Rerank 4 Pro: 32,000 tokens, context length (models table lists 32k). Rerank 4 Fast: 32,000 tokens, context length (models table lists 32k). Rerank 3.5: 4,096 tokens, per query plus document. Rerank English 3.0: 4,096 tokens, per query plus document. Rerank Multilingual 3.0: 4,096 tokens, per query plus document.
No official open weights were found in the sources checked; Cohere documents these as hosted models.
Last verified from each provider's own docs, pricing pages and Hugging Face cards; every quote was string-matched against the page that day. Prices are shown in the provider's own unit and change without notice, so check the linked source before you commit a budget.
Newsletter
One email per week. New model releases, retrieval and pricing updates and what they mean for what you build.