Cohere
Cohere provides large language model APIs for developers and enterprises building natural language processing applications.
About this data
Updated June 29, 2026
Overall Pulse Score
-8 over this period
A 0-100 index summarizing the tone of 104 relevant public mentions gathered from public online communities across 7 weeks in the selected period. It measures online sentiment, not a rating of the product's quality.
Weekly Sentiment Trend
Pulse Score by week over the selected period. Each point is one complete week of mentions.
This week in public discussion
Discussion around Cohere over the recent period was largely shaped by bug reports, which dominated mentions by a wide margin. Commenters frequently flagged reliability issues including rerank costs appearing as null, silent billing failures on Bedrock and Vertex, and a training-loss bug in the ASR implementation. On the positive side, several mentions praised integration flexibility and feature quality, with some discussions highlighting Cohere being incorporated into RAG and load-balancing workflows. The pulse score held steady, suggesting neither meaningful recovery nor further decline in community sentiment.
Read the deeper analysisAI-generated summary of public online discussion during this period. It reflects the tone of that discussion, not facts about the product or our views.
Sentiment mix by week
How the tone of public discussion splits each week.
Most-discussed praise
Most-discussed complaints
Themes across the selected period, with mention counts.
How Cohere compares
Pulse Score over the selected period versus the top tracked competitors in Software.
Where the mentions come from
Share of the 104 relevant public mentions in the selected period, by source.
Sample public mentions
Showing 5 of 104 analyzed public mentions in this period, with links to the original source. We do not reproduce full threads.
“[Bug]: OpenRouter rerank models not available in v0.26.1 UI, but /api/v1/rerank works. ### Self Checks - [x] I have searched for existing issues search for existing issues, including closed ones. - [x] I confirm that I am using English to submit this report (Language Policy). - [...”
“I have been working on multiple projects lately involving AI endpoints (including some I run locally) and I found I needed a way to easily load balance across multiple. Sometimes my on-prem would not be able to handle to load and Id have to crank up the z.ai usage or Anthropic de...”
“AI Gateway: cohere/embed-v4.0 intermittently ignores outputDimension (returns 1536 instead of requested 512). ### Summary When requesting reduced-dimension embeddings from cohere/embed-v4.0 through the AI Gateway, the gateway **intermittently** ignores the requested output dimens...”
“bug: Instruction payload is prefixed with safety instructions. **Version:** e.g. 0.5.x-xxx Describe the Bug Context is being prefixed with a pre-instruction card: Causing certain models to preamble about safety and or what they are allowed to do, or they won't embody their assist...”
“Cohere on AWS Bedrock ignores memory_action and embeds search queries with search_document input_type. ### Bug mem0/embeddings/aws_bedrock.py hardcodes Cohere input_type="search_document" for every call and ignores memory_action. mem0 passes memory_action="search" for queries, so...”
1,144+ more analyzed mentions, full history, and theme breakdowns are part of Pro.
Get ProDeeper analysis
- Bug reports and reliability concerns dominated the conversation, accounting for the large majority of negative mentions across the window.
- Sentiment fell sharply as mention volume spiked in early June before showing only a small partial recovery in the most recent week.
- Opinion was divided on integrations, with some commenters praising connectivity while others flagged gaps and confusing error behavior.
- A smaller but consistent thread of feature praise and AI quality appreciation suggested a portion of users remained positively engaged despite the broader negativity.
| Praise theme | Mentions |
|---|---|
| Strong features | 24 |
| Good integrations | 13 |
| Feature requests | 11 |
| AI quality | 8 |
| New releases | 7 |
| Complaint theme | Mentions |
|---|---|
| Bugs | 40 |
| Reliability | 15 |
| Missing features | 12 |
| Lacking integrations | 10 |
| UI frustrations | 3 |
Discussion of Cohere over the past four weeks was dominated by a heavy volume of bug reports, which formed the single largest theme in the conversation by a significant margin. Commenters frequently surfaced issues tied to cost tracking, usage mapping, and silent failures in rerank and embedding workflows, with several mentions pointing to billing-related data never being populated correctly. The tone around these reports was technical and frustrated, suggesting users were encountering problems in production or near-production environments rather than casual experimentation.
The score trajectory tells a clear story of deterioration followed by a faint stabilization. Early in the window, discussion was sparse but relatively positive, with scores in the high-60s to mid-80s. As mention volume surged sharply in early to mid June, sentiment dropped hard, bottoming out in the high 30s. The most recent period showed a modest uptick, though commenters who drove that recovery still appeared to be working through reliability and integration concerns rather than celebrating improvements.
Praise did exist in the conversation, and several mentions highlighted specific features and integration possibilities in appreciative terms. A thread of enthusiasm around AI output quality and new feature releases suggested a subset of users remained genuinely engaged with the product. Feature requests also surfaced with some regularity, which discussion suggested reflects ongoing user investment rather than abandonment.
Opinion was most divided around integrations. Some commenters described smooth and useful connections with broader tooling, while others flagged integration gaps and confusing error handling for non-standard model configurations. This split gave the integration theme an ambivalent character overall. UI complaints were relatively minor in volume, drawing little sustained attention compared to the bug and reliability threads that anchored most of the negative sentiment.
AI-generated summary of public online discussion during this period. It reflects the tone of that discussion, not facts about the product or our views.
Member perspectives
Individual opinions from Pro members, posted over time. These are personal member views, not aggregated sentiment data.
Overall Pulse Score
-8 over this period
A 0-100 index summarizing the tone of 104 relevant public mentions gathered from public online communities across 7 weeks in the selected period. It measures online sentiment, not a rating of the product's quality.
Data summary
Compare with another tool
Cohere
45
Trainual
88
Score-level preview from live weekly tracking.
Are you Cohere?
Get a private enterprise dashboard for your product - full history, every source, theme deep-dives, and weekly alerts. You can also respond to the data shown here.
Explore the enterprise dashboardAffiliate disclosure
Some links on this site may be affiliate links. If you click one and make a purchase, we may earn a commission at no extra cost to you. Learn more.
Is Cohere your product?
See everything behind this page - full history, every source, theme deep-dives, and weekly alerts - in a private enterprise dashboard.
Request early accessCompare with similar tools
Mail Memories
Mail Memories is an email-based service that delivers users their past emails or memories on a scheduled basis.
Paid plans available
View DetailsLookAway
LookAway is a Mac app that reminds users to take regular breaks from their screen to reduce eye strain.
Free tier; paid plans available
View Details