Cohere AI has released Rerank 5, a new reranker family in Pro and Fast tiers. Cohere Rerank 5 is a retrieval model that reorders search results before they reach an LLM. It adds 2 developer features: Context Focus, which returns relevant passages inside each document, and calibrated relevance scores. For RAG and agent builders, the message is simple. Send fewer, better tokens to the LLM.
TL;DR
- Size: Parameter count not disclosed. 32K token context window. Two tiers:
rerank-v5.0-proandrerank-v5.0-fast. - Runs on: Hosted API, Model Vault, Microsoft Foundry, Amazon SageMaker, or private VPC and on-prem. Cohere benchmarks speed on a single NVIDIA A100. No open weights or quantized builds.
- Performance: Pro ranks first on 39 of 46 composite datasets in Cohere’s own evaluation.
- Best: Pro scores 82.2 on BRIGHT (RCP-nDCG@10), 5.8 points above Voyage AI Rerank 3.
- Worst: Pro median latency is 11.7 s per 96-document request, versus 4.1 s for Fast.
What is new in Cohere Rerank 5?
Rerank 5 ships as two models. Pro targets ranking quality on reasoning-heavy queries and long, domain-specific documents. Fast targets low latency and high throughput for agents running many searches.
Both models support over 100 languages and a 32K context window. They also rerank semi-structured data such as JSON, tables and YAML records. Upgrading needs no re-indexing. Developers change the model parameter and keep their existing keyword, vector or hybrid search.
How does Context Focus work?
Context Focus is an optional feature, off by default. A single Rerank call returns document scores plus scored character spans. Each span carries start and end offsets into the original text. Developers set it through the relevant_spans field, with a threshold and max_spans value.
Cohere tested it on 30 QA datasets. Removing half the retrieved text kept 96% of answer quality. Removing three quarters kept 87%. Cohere says teams can push to 10x less context at some quality cost.
In a search-agent test across 17 datasets, BM25 alone scored 46% accuracy. Adding Rerank 5 Pro raised it to 49%. Context Focus at the default 0.1 threshold cut tokens per task by 19% and reached 57%. At 0.4, tokens fell 34% and accuracy hit 58%. Versus BM25 alone, the full stack used 35% fewer tokens. On 6 deep-search benchmarks, tokens fell 41% and accuracy rose from 39% to 52%.
What do calibrated relevance scores change?
Rerank 5 scores read as probabilities. A score of 0.7 means a 70% chance the document is useful.
Cohere team measured expected calibration error (ECE) against blind human graders and NIST TREC-DL assessors. Rerank 5 scores about 0.06 on both. Voyage AI Rerank 3 scores 0.234 and 0.307. Rerank 4 Pro scores 0.384 and 0.309. In practice, dev teams can filter by threshold instead of a fixed top-k. An agent can also detect when nothing relevant came back.
How does Rerank 5 perform on benchmarks?
All scores below are self-reported by Cohere using RCP-nDCG@10, a Cohere-developed variant of nDCG. Cohere ran the competitor models itself.
Pro leads all four main benchmarks. It scores 92.2 on ViDoRe V3, 90.4 on BEIR, 82.2 on BRIGHT and 87.3 on nanoMTEB. BRIGHT shows the largest jump: 12.4 points over Rerank 4 Pro.
Fast beats Voyage AI Rerank 3 Lite, Qwen3-Reranker-8B and Rerank 4 Pro on all 4. It ties Voyage AI Rerank 3 on ViDoRe V3 at 90.6. It trails it on BEIR, 88.1 versus 88.6.
On parsed PDFs, averaged over 47 datasets, Pro scores 89.9 and Fast 88.5. Voyage AI Rerank 3 scores 88.4. Pro ranks first on 14 of 19 non-English languages in Cohere’s internal suite.
‘}).join(”);
function drawB(b){var lo=50,f=bars.querySelectorAll(‘.fill’),v=bars.querySelectorAll(‘.val’);D[b].forEach(function(x,i){f[i].style.width=”0″;v[i].textContent=x.toFixed(1);(function(el,w){setTimeout(function(){el.style.width=((w-lo)/(100-lo)*100)+’%’},40+i*60)})(f[i],x)})}
R.querySelectorAll(‘#rrseg .btn’).forEach(function(btn){btn.onclick=function(){R.querySelectorAll(‘#rrseg .btn’).forEach(function(x){x.classList.remove(‘on’)});btn.classList.add(‘on’);cur=+btn.dataset.b;drawB(cur)}});
// context focus
var CF=[[‘46%’,’n/a’,’Baseline: keyword search only’],[‘49%’,’0%’,’Rerank 5 Pro reorders the top 100, Context Focus off’],[‘57%’,’-19%’,’Tokens per task vs Rerank 5 Pro without Context Focus’],[‘58%’,’-34%’,’Tokens per task vs Rerank 5 Pro without Context Focus’]];
R.querySelectorAll(‘#rrcf .btn’).forEach(function(btn){btn.onclick=function(){R.querySelectorAll(‘#rrcf .btn’).forEach(function(x){x.classList.remove(‘on’)});btn.classList.add(‘on’);var c=CF[+btn.dataset.c];document.getElementById(‘rracc’).textContent=c[0];document.getElementById(‘rrtok’).textContent=c[1];document.getElementById(‘rrtokd’).textContent=c[2]}});
function drawQ(){R.querySelectorAll(‘.rrq’).forEach(function(el,i){el.style.width=”0″;setTimeout(function(){el.style.width=el.dataset.w+’%’},60+i*120)})}
// ECE
var EM=[‘Cohere Rerank 5 Pro’,’Cohere Rerank 5 Fast’,’Cohere Rerank 4 Pro’,’Voyage AI Rerank 3′,’Voyage AI Rerank 3 Lite’];
var ED=[[0.058,0.058,0.384,0.234,0.226],[0.057,0.058,0.309,0.307,0.325]],EC=[‘var(–coral)’,’var(–lilac)’,’#6b6878′,’var(–purple)’,’#8f7fe8′];
var ece=document.getElementById(‘rrece’),curJ=0;
ece.innerHTML=EM.map(function(m,i){return ”}).join(”);
function drawE(j){var f=ece.querySelectorAll(‘.fill’),v=ece.querySelectorAll(‘.val’);ED[j].forEach(function(x,i){f[i].style.width=”0″;v[i].textContent=x.toFixed(3);(function(el,w){setTimeout(function(){el.style.width=(w/0.4*100)+’%’},40+i*70)})(f[i],x)})}
R.querySelectorAll(‘#rrjd .btn’).forEach(function(btn){btn.onclick=function(){R.querySelectorAll(‘#rrjd .btn’).forEach(function(x){x.classList.remove(‘on’)});btn.classList.add(‘on’);curJ=+btn.dataset.j;drawE(curJ)}});
window.addEventListener(‘load’,post);window.addEventListener(‘resize’,post);setTimeout(post,300);
})();
