Denser vs Chatbase
Both build a chatbot from your own content. Chatbase runs standard vector search, does not cite its sources, and caps each agent at 20 MB. Denser links every answer to the passage behind it, retrieves with a reranker benchmarked above vector search on MTEB, and holds 1 GB per bot at the same tier.
2,000+ businesses · free plan, no card required · 5-minute setup
Trusted by 2,000+ businesses worldwide
City of San José
Onondaga County
EOS Australia
Portfoliobox
Upsiide
Valk WeldingCity of San JoséOnondaga CountyEOS AustraliaPortfolioboxUpsiideValk WeldingDenser links each reply to the exact passage it came from, so a customer or an agent can confirm it in one click. Chatbase does not advertise source citations on its product or pricing pages. An answer nobody can check is an answer nobody should act on.
A chatbot can only answer from the passage retrieval hands it, so retrieval quality sets the ceiling on everything downstream. Denser Retriever combines keyword search, vector search, and an XGBoost reranker, and on MTEB retrieval benchmarks that combination beats a vector-search baseline outright, by 13.07% NDCG@10 over top vector baselines on MS MARCO. Chatbase runs standard vector search. In support, that gap is the difference between a correct answer and a confident wrong one.
A Denser bot holds 2,000 pages or 1 GB on Standard and 50,000 pages or 25 GB on Enterprise, and the reranking stage is what keeps the right passage findable as the corpus grows. Chatbase caps training data at 20 MB per agent on Standard and 40 MB on Pro, so a documentation estate of any size does not fit before quality is even discussed.
Denser grades every answer and flags the ones that fell short, labelled by cause and clustered by topic, so knowledge gaps arrive as a ranked list. Chatbase has no equivalent, so a gap surfaces only when a customer complains about it.
Chatbase runs voice, telephony, and email channels that Denser does not. That breadth is priced accordingly: Standard is $150/mo for 4,000 credits against Denser's $119 for 7,500 queries, extra agents are $25/mo each, and removing the badge is $99/mo. None of those channels change what an answer is built on, which is where this comparison is actually decided.
| Denser | Chatbase | |
|---|---|---|
| Source citations | Every answer cited to its exact passage | Not advertised |
| Retrieval | Hybrid keyword + vector + XGBoost reranking, MTEB-benchmarked above vector search alone. Open source (MIT) | Vector search |
| Content per bot | 2,000 pages / 1 GB (Standard) up to 50,000 pages / 25 GB | 20 MB (Standard), 40 MB (Pro) |
| Billing unit | One query per question on the latest Claude, GPT, or Gemini | 1–5 credits per reply depending on model |
| Entry paid plan | $39/mo, 1,500 queries | $40/mo, 700 credits |
| Mid tier | $119/mo, 7,500 queries | $150/mo, 4,000 credits |
| Bots included | 2 (Starter) to 8 | 1 on every published tier, +$25/mo each |
| Remove vendor badge | Included | $99/mo |
| Overage | Upgrade tier | Auto-recharge $40 per 1,000 credits |
| Deployment | Widget, iframe, React, REST API, TS/Python SDKs, WordPress, Shopify, Slack, WhatsApp | Widget, Slack, WhatsApp, phone, email, API |
| Free plan | 20 queries/mo, no card | 50 credits/mo, agent deleted after inactivity |
| Voice / telephony | Not supported | Supported |
| Email channel | Not supported | Supported |
| Best for | Accurate, cited answers over a large knowledge base | Multi-channel CX including voice |
Real cost
Take a mid-sized site fielding 5,000 customer questions a month, on each vendor's published rates.
| Denser | Chatbase | |
|---|---|---|
| Plan needed | Standard, $119/mo (7,500 queries) | Standard, $150/mo (4,000 credits) |
| Overage | None — 2,500 queries still unused | 1,000 credits short: +$40 auto-recharge |
| If you use a premium model | No change — a query is a query | 5,000 answers ≈ 10,000+ credits, so Pro at $500/mo |
| Remove vendor badge | Included | +$99/mo |
| Monthly total | $119 | $289 on a standard model, $599 on a premium one |
Rates read from chatbase.co/pricing on 19 August 2026. Credit burn per model is Chatbase's own published figure.
Switching
Chatbase trains on files and URLs you already have. Collect the same PDFs, docs, and sitemap you gave it — there is nothing proprietary to extract.
Add the URLs for a crawl and upload the files. Indexing is automatic; a typical knowledge base is ready in minutes.
Replace the Chatbase script tag with Denser's. The widget position, colours, and greeting are configurable to match what you had.
Run your twenty most common customer questions through both before you cancel. Denser shows the source passage behind each answer, so the comparison is verifiable rather than a vibe.
Questions
It is the stronger choice when accuracy and knowledge-base size decide the evaluation. Denser cites every answer to its exact source passage, holds 1 GB per bot on its $119 Standard plan against Chatbase Standard's 20 MB, and bills one query per customer question instead of credits that burn faster on premium models. Chatbase remains the better pick if you need voice, email, and outbound campaigns in one platform.
Chatbase does not advertise source citations on its product or pricing pages. Denser links every answer to the exact passage it came from, so your team and your customers can verify a reply in one click rather than trusting it.
Chatbase caps training data at 1 MB on Free, 10 MB on Hobby, 20 MB on Standard, and 40 MB on Pro, per agent. A Denser bot holds 100 pages or 50 MB on Starter, 2,000 pages or 1 GB on Standard, 10,000 pages or 5 GB on Business, and 50,000 pages or 25 GB on Enterprise.
At comparable volume, yes. Denser Standard is $119/mo for 7,500 queries; Chatbase Standard is $150/mo for 4,000 credits, and a premium model spends 2 to 5 credits per reply. Chatbase also charges $99/mo to remove its badge and $25/mo per additional agent, both of which Denser includes.
A Denser query is one customer question, and the latest Claude, GPT, and Gemini models all cost exactly that one query. A Chatbase credit is consumed per reply at a rate set by the model — 1 credit on standard models, up to 5 on Claude Opus — so identical conversation volume produces a different bill depending on model choice.
Yes, and there is nothing to export. Chatbase trains on files and URLs you already own, so you point Denser at the same sources, let it index, and swap the embed snippet. Most teams are running both side by side within an hour.
The retrieval engine is. Denser Retriever is MIT-licensed on GitHub and combines keyword search, vector search, and an ML reranking stage. You can audit exactly how answers are found, which is not possible with Chatbase's retrieval.
Point Denser at your website or upload your documents, ask the questions your customers actually ask, and check the source behind every answer. It takes about five minutes.
