Glean crawls connected SaaS applications into a unified index that mirrors each source's access controls, then exposes it as a search page, an assistant, an agent builder and a REST API. The differentiator against building is the connector fleet — more than 250 connectors with ACL sync — plus Glean Protect for governance and an AI gateway that routes requests across models. Unlike Onyx or Dashworks it is sold as a company-wide platform rather than a team tool, and unlike Vectara or Ragie the end-user application is the product rather than something you assemble.
| Deployment | Vendor SaaS only |
|---|
| Prebuilt SaaS connectors | 250+ |
|---|
| Source ACL enforcement | Yes, synced and enforced at query time |
|---|
| Open source | No |
|---|
| Primary interface | End-user search and chat app, plus REST API |
|---|
Watch out: No published pricing at all, which makes budgeting a sales cycle rather than a calculation, and reported deals start in the six figures with seat minimums — badly matched to powering a single application feature. There is no self-hosted option, you cannot inspect or replace the chunking and ranking internals, and index freshness is a crawl schedule you do not control.
unknown — the pricing page publishes no figures and routes to a demo request; deals are seat-based annual contracts quoted by sales.
Onyx, renamed from Danswer, is an open-source chat and search layer over connected company apps that you can run in your own Docker or Kubernetes environment or buy hosted at $20 per user per month. It ships around 40 connectors, custom agents, a code interpreter and developer APIs, and works with any LLM provider. It is the practical answer when Glean is the right shape but the price or the data-residency requirement is not; the trade is that you operate the index and identity plumbing yourself.
| Deployment | Self-host (Docker/K8s) or vendor cloud |
|---|
| Prebuilt SaaS connectors | 40+ |
|---|
| Source ACL enforcement | Yes, via connector permission sync (enterprise-licensed code) |
|---|
| Open source | Yes — MIT core, proprietary ee/ modules |
|---|
| Primary interface | End-user search and chat app, plus developer API |
|---|
Watch out: Connector depth and the reliability of permission syncing lag Glean and the hyperscalers, so ACL edge cases become your problem. SSO/SAML, permission sync and other enterprise controls live in ee/ directories that are not MIT-licensed, so the genuinely free build is missing the features an enterprise rollout needs, and self-hosting means running and tuning the search and database infrastructure yourself.
Cloud Business $20 per user per month billed annually; Enterprise is quote-only. Self-hosting the MIT-licensed core is free.
· open source
Vectara is an end-to-end retrieval API: you POST documents, it handles chunking, embedding, hybrid retrieval, reranking and grounded generation, and returns answers with citations plus a factual-consistency score from its own hallucination evaluation model. It is a retrieval service rather than an employee search app, so there is no connector fleet to Slack or SharePoint. The notable 2026 change is commercial, not technical: the published entry price is now $100,000 per year for SaaS, replacing the low-cost self-serve tiers it was previously known for.
| Deployment | Vendor SaaS, customer VPC, or on-premises |
|---|
| Prebuilt SaaS connectors | Minimal — API ingestion is the primary path |
|---|
| Source ACL enforcement | No — you pass filter attributes and enforce them yourself |
|---|
| Open source | No |
|---|
| Primary interface | REST API / SDK |
|---|
Watch out: The pricing floor is the headline problem — at $100K/year for SaaS it is now unavailable to startups and small teams that used its earlier self-serve tiers, and any tutorial or comparison written before 2026 quotes prices that no longer exist. It is also a retrieval API, not an enterprise search product: there is no meaningful catalogue of SaaS connectors and no end-user application, so ingestion from Drive, Slack or Confluence is code you write.
SaaS from $100,000/year (1 deployment); VPC from $250,000/year; on-premises from $500,000/year. 30-day full-feature free trial.
Contextual AI sells an assembled RAG agent platform whose individual components — a document parser, a reranker, a grounded generation model and the LMUnit evaluation model — are also callable separately and priced separately. That makes it the unusual member of this category where you can see exactly what each stage costs: $3 per 1,000 pages for text parsing, $40 per 1,000 pages for multimodal, $0.05 per million tokens to rerank. It targets accuracy-sensitive document workloads rather than employee search, and competes with Vectara on grounding quality rather than with Glean on connectors.
| Deployment | Vendor SaaS; enterprise/private options by contract |
|---|
| Prebuilt SaaS connectors | Minimal — API ingestion is the primary path |
|---|
| Source ACL enforcement | No — metadata filtering only |
|---|
| Open source | No |
|---|
| Primary interface | REST API / SDK |
|---|
Watch out: There is no end-user search application and effectively no SaaS connector catalogue, so anything involving Drive, Slack or Confluence is ingestion code you own. Component-level pricing sounds transparent but pushes cost modelling onto you, and the $40 per 1,000 pages multimodal parse tier gets expensive fast on scanned or diagram-heavy corpora — re-parsing a large archive is a bill worth estimating before you start.
$25 free credits on signup. Parse $3 per 1,000 pages (text) or $40 per 1,000 pages (multimodal); Rerank-v2 $0.05 per 1M tokens, rerank-v2-mini $0.02 per 1M tokens; Generate $3 per 1M input tokens, $15 per 1M output tokens. Enterprise plan quote-only.
Ragie is a developer-facing managed pipeline: connectors pull from SaaS sources, it parses, chunks, embeds, reranks and serves hybrid retrieval, with partitions for multi-tenant isolation and whitelabel connectors for products that need to embed their customers' OAuth flows. It sits between the pure retrieval APIs and the employee search apps — it has real connectors, but no end-user UI. The pricing structure is unusual and worth reading carefully: retrievals are unlimited on paid plans, but each connector past the first is a separate $250 per month line item.
| Deployment | Vendor SaaS only |
|---|
| Prebuilt SaaS connectors | Yes, billed per connector ($250/mo each after the first) |
|---|
| Source ACL enforcement | Partial — partitions and metadata filters; per-user source ACLs are your responsibility |
|---|
| Open source | No |
|---|
| Primary interface | REST API / SDK |
|---|
Watch out: The $250 per month per connector charge is the real cost driver — a product touching six sources pays $1,250/month in connector fees on top of the plan, which can exceed the plan price several times over. It is also a small vendor with no end-user application, so you build the search and chat surface yourself, and there is no self-hosted option if data residency is a hard constraint.
Developer free (1,000 pages, 1,000 retrievals, first connector free). Starter $100/month (10,000 pages). Pro $500/month (60,000 pages). Extra pages $0.02 fast / $0.05 hi-res; storage $0.12/GB/month; audio $0.0067/min, video $0.025/min. Every connector after the first is $250/month.
Bedrock Knowledge Bases turns a data source into a queryable index and a Retrieve/RetrieveAndGenerate API, with the managed option bundling document parsing, embedding generation and reranking at no separate charge. A self-managed mode lets you point it at your own OpenSearch, Aurora or Pinecone index instead. It is the natural choice if your application already runs on AWS and your documents already sit in S3; it is not an employee search product and has no chat UI of its own.
| Deployment | AWS-managed service in your account |
|---|
| Prebuilt SaaS connectors | Limited (S3, SharePoint, Confluence, Salesforce, web crawler) |
|---|
| Source ACL enforcement | Partial — metadata filtering; per-user source ACLs largely your responsibility |
|---|
| Open source | No |
|---|
| Primary interface | REST API / SDK |
|---|
Watch out: The connector list is short compared with the workplace-search products — S3, SharePoint, Confluence, Salesforce and a web crawler — and per-user permission propagation is limited, so multi-tenant ACL enforcement is largely on you. Per-call retrieval pricing punishes agentic loops that fire many retrievals per turn, feature availability varies by region, and you still write the application, evaluation and chunking-strategy work around it, which makes this closer to build than buy.
Managed knowledge bases: $5.00 per GB of raw data indexed per month, $1.00 per 1,000 Retrieve API calls, $4.00 per 1,000 agentic Retrieve calls plus $1.00 per 1,000 underlying Retrieve calls. Managed parsing, embedding and reranking are included at no extra charge. Self-managed mode bills the underlying vector store separately.
This is the product previously sold as Vertex AI Search, Enterprise Search and Gen App Builder, now called Agent Search. You point it at unstructured documents, structured data or a website, and it returns ranked results plus grounded generative answers with citations. Two incompatible pricing models exist: General (pay per query and per GB indexed) and Configurable (a monthly subscription for query-per-minute throughput plus per-query add-ons), and a data store is locked to whichever model it was created under. Vertical variants exist for healthcare and media.
| Deployment | Google Cloud managed service |
|---|
| Prebuilt SaaS connectors | Limited — third-party connectors are largely in Gemini Enterprise instead |
|---|
| Source ACL enforcement | Partial — filtering and identity-mapped ACLs on supported sources |
|---|
| Open source | No |
|---|
| Primary interface | REST API / SDK, plus a console widget |
|---|
Watch out: The naming churn is genuinely disruptive — Gen App Builder to Vertex AI Search to Agent Search means most third-party documentation and blog posts you find are describing this product under a name it no longer uses, and SDK imports have moved with it. The two pricing models are mutually exclusive per data store, so picking wrong means rebuilding the index, and at $4.00 per 1,000 Enterprise queries plus $4.00 per 1,000 for advanced answers, a chatty agent doing several retrievals per turn gets expensive quickly.
General model: Search Standard Edition $1.50 per 1,000 queries; Search Enterprise Edition $4.00 per 1,000 queries; Advanced Generative Answers +$4.00 per 1,000 user-input queries; data indexing $5 per GiB per month with the first 10 GiB free. 10,000 free queries per account per month. Configurable model: $6.00 per QPM per month plus $1/GB/month storage and per-query add-ons (Semantic $0.75 per 1,000, KPI & Personalization $0.20 per 1,000).
Gemini Enterprise is Google's packaged answer to Glean and Microsoft 365 Copilot: a seat-licensed application that indexes connected company data, answers questions over it, and lets staff build no-code agents in a workflow builder. It absorbed the Agentspace product line and sits on top of the same retrieval stack as Agent Search, with pooled index storage sized per seat. A pay-as-you-go edition with a $0 seat fee is rolling out to a limited set of customers for organisations that prefer usage billing.
| Deployment | Google Cloud SaaS |
|---|
| Prebuilt SaaS connectors | Yes — Workspace plus third-party enterprise sources |
|---|
| Source ACL enforcement | Yes for Workspace and identity-mapped connected sources |
|---|
| Open source | No |
|---|
| Primary interface | End-user chat and search app, plus agent builder |
|---|
Watch out: It is the newest of the three big workplace assistants and the edition structure is still moving — Agentspace was folded into it, the pay-as-you-go edition is only rolling out to selected customers, and Gemini Notebook is not yet included there. Value depends heavily on already being a Workspace shop; if your documents live in SharePoint and Teams the connector story is weaker than Microsoft's, and the Business edition's 300-seat ceiling forces a migration to Standard/Plus mid-rollout for larger companies.
Business from $21 per seat per month (up to 300 seats, 25 GiB pooled index storage per seat). Standard/Plus from $30 per seat per month (unlimited seats, up to 75 GiB per seat). Frontline edition available as an add-on to Standard/Plus. Pay-as-you-go edition: $0 seat fee for 20+ seats, billed on tokens, memory, compute and storage (limited availability).
Microsoft 365 Copilot grounds its answers in the Microsoft Graph — your SharePoint, OneDrive, Outlook and Teams content — with each user's existing permissions applied automatically, and extends to external systems through Graph connectors, of which Microsoft advertises over 100. It is sold as a per-user add-on requiring a qualifying Microsoft 365 licence underneath, and a cheaper Business SKU now exists for smaller organisations. Because the ACL model is inherited from the tenant rather than reconstructed by a crawler, it is the lowest-effort permission-correct option for organisations already on M365.
| Deployment | Microsoft 365 tenant SaaS |
|---|
| Prebuilt SaaS connectors | 100+ Graph connectors |
|---|
| Source ACL enforcement | Yes — inherits Microsoft Graph permissions per user |
|---|
| Open source | No |
|---|
| Primary interface | End-user chat and search inside Microsoft 365 apps |
|---|
Watch out: Answer quality is a direct function of SharePoint hygiene — over-shared sites and stale document libraries surface as confidently wrong or inappropriately visible answers, and cleaning that up is usually a bigger project than the deployment itself. It requires a qualifying M365 licence underneath, so the real cost is well above $30 per user, and grounding on data outside the Graph means building and paying for Graph connectors with their own indexing quotas rather than pointing at a database.
Microsoft 365 Copilot $30.00 per user per month paid yearly, or $31.50 per user per month on an annual commitment billed monthly; a qualifying Microsoft 365 licence is required separately. Microsoft 365 Copilot Business is listed from $21.00 per user per month paid yearly (currently promoted at $18.00).
Azure AI Search provides the index, hybrid and vector retrieval, an indexer and skillset pipeline for enrichment, a semantic reranker, and a newer agentic retrieval mode that spends reasoning tokens to decompose a query into subqueries. Capacity is bought as search units billed hourly, with a serverless option that separates compute and storage. It is the most build-shaped entry here: nothing about chunking strategy, evaluation or the end-user experience is decided for you, which is either the point or the disqualifier depending on your team.
| Deployment | Azure managed service in your subscription |
|---|
| Prebuilt SaaS connectors | Limited — mostly Azure data sources plus SharePoint Online |
|---|
| Source ACL enforcement | Partial — security trimming via document-level filters you populate |
|---|
| Open source | No |
|---|
| Primary interface | REST API / SDK |
|---|
Watch out: Calling this managed RAG oversells it — you still design chunking, write indexers and skillsets, build evaluation, and construct the entire application layer, so the engineering saving versus assembling components is modest. There is no end-user chat or search application, SaaS connector coverage outside Azure data sources and SharePoint is thin, and capacity billed per search unit per hour means you pay for a provisioned index whether or not anyone queries it.
Free tier available. Basic $0.101/hour (~$74/month), Standard S1 $0.336/hour (~$245/month), S2 $1.344/hour, S3 $2.688/hour, Storage Optimized L1 $3.839/hour (East US). Serverless: compute $0.24/hour, storage $0.20/GB/month. Semantic ranker $1.00 per 1,000 queries ($2.00 per 1,000 overage). Agentic retrieval reasoning tokens $0.000022 per 1K (low/minimum) to $0.0001 per 1K (medium).
Quick Suite is AWS's employee-facing assistant and the stated successor to Amazon Q Business, combining chat over connected company data, research agents, automation flows and QuickSight analytics under one seat licence. Existing Q Business customers are being moved onto it. Two licensing paths exist: consumer-style Free/Plus/Max plans capped at 300 users, and AWS-account Professional and Enterprise plans with pooled index storage, agent-hour metering and a flat monthly infrastructure fee.
| Deployment | AWS-managed SaaS |
|---|
| Prebuilt SaaS connectors | Yes — Slack, Microsoft 365, Google, SharePoint, Confluence, ServiceNow and others |
|---|
| Source ACL enforcement | Yes — permission-aware responses via IAM Identity Center |
|---|
| Open source | No |
|---|
| Primary interface | End-user chat and research app, plus embeddable APIs |
|---|
Watch out: You are buying into a product mid-transition — Q Business is being superseded, migration paths and feature parity are still settling, and documentation written for either name may not describe what you get. The $250 per account per month infrastructure fee is a floor that makes small pilots disproportionately expensive, agent-hour metering at $3-$6 per hour makes spend hard to forecast, and the enterprise path requires standing up AWS IAM Identity Center before anyone logs in.
Free $0 (1 user, 1GB index). Plus $20/user/month billed annually ($25 monthly). Max $100/user/month billed annually ($125 monthly); both capped at 300 users. AWS-account plans: Professional $20/user/month, Enterprise $40/user/month, plus a flat $250 per account per month infrastructure fee, $3 per agent hour ($6 per research agent hour), and $5/GB/month index storage overage. Legacy Amazon Q Business: Lite $3/user/month, Pro $20/user/month, plus index units at $0.264/hour.
Dify is an open-source platform for building LLM applications visually — workflows, agents, prompt management — with a built-in knowledge base that handles upload, chunking, embedding and retrieval. Its appeal in this category is that a non-specialist can stand up a working RAG chatbot in an afternoon and self-host the whole thing. It is an application builder that includes retrieval rather than a retrieval platform, which shows in the document limits and the absence of enterprise connector and permission machinery.
| Deployment | Self-host or vendor cloud |
|---|
| Prebuilt SaaS connectors | Few — mostly file upload, web crawl and plugin integrations |
|---|
| Source ACL enforcement | No |
|---|
| Open source | Yes — Apache 2.0 with additional use restrictions |
|---|
| Primary interface | Visual app builder, plus API for built apps |
|---|
Watch out: The knowledge base is deliberately simple — hard document caps of 500 on Professional and 1,000 on Team, no SaaS connector fleet, and no per-user ACL synchronisation — so it does not scale into company-wide search. The licence is not plain Apache 2.0: it forbids offering Dify as a multi-tenant service and removing branding, which rules it out as the backbone of a product you resell.
Sandbox free (200 message credits, 5 apps, 50 knowledge documents, 50MB storage). Professional $590/year (5,000 monthly message credits, 500 knowledge documents, 5GB). Team $1,590/year (10,000 monthly message credits, 1,000 knowledge documents, 20GB, 50 members). Enterprise quote-only. Community edition free to self-host.
· open source
Morphik focuses on documents where the layout carries meaning — engineering drawings, medical scans, financial statements — using multimodal ingestion rather than text extraction alone, plus knowledge-graph construction and a research agent over the corpus. The core is on GitHub and self-hostable. It is a narrower bet than Ragie or Vectara: strong on hard document types, thin on connectors and end-user search.
| Deployment | Vendor cloud, dedicated VPC instance (Team), or self-host under BSL |
|---|
| Prebuilt SaaS connectors | Few — file and API ingestion is the primary path |
|---|
| Source ACL enforcement | No |
|---|
| Open source | No — Business Source License 1.1 with a revenue cap |
|---|
| Primary interface | REST API / SDK, plus a console |
|---|
Watch out: The Business Source Licence is the catch: production use is only granted while the revenue attributable to it stays under $2,000 per month, so this is not free software for a commercial product and calling it open source is inaccurate. Page-based pricing with $0.03 overage penalises large archives that are queried infrequently, the team is small with correspondingly limited support, and there is no SaaS connector fleet or ACL model for workplace search.
Free $0 (200 pages, 3 research-agent calls/month, shared GPU queue). Pro $59/month (2,000 pages, 30 research-agent calls, $0.03/page overage). Team $799/month (10GB storage, unlimited queries, dedicated isolated instance, $2.50/GPU-hour autoscale). Enterprise custom.
LlamaCloud is the hosted side of LlamaIndex: LlamaParse for document parsing with several quality tiers and auto-routing, structured extraction to a schema, classification and splitting, plus managed indexes and a workflow builder. Everything is metered in credits at 1,000 credits per $1.25, so parsing, indexing and extraction all draw from one pool. Its real strength is the parser — it is frequently used purely as an ingestion stage in front of someone else's vector store.
| Deployment | Vendor SaaS (US or EU); hybrid/VPC on Enterprise |
|---|
| Prebuilt SaaS connectors | Limited — 50 external data sources on Starter, 100 on Pro |
|---|
| Source ACL enforcement | No |
|---|
| Open source | No — the LlamaIndex framework is MIT, the cloud service is not |
|---|
| Primary interface | REST API / SDK, plus a console |
|---|
Watch out: The per-index file caps are the blocking constraint — 2,000 files per index even on the $500/month Pro tier means any real document archive needs the Enterprise plan or sharding across indexes. Credit-based billing spanning parse, extract and index makes forecasting awkward, and the hosted index and connector features are noticeably less developed than the parsing side, so this is usually a component purchase rather than the assembled thing.
Free $0 (10,000 credits, 5 indexes, 50 files per index). Starter $50/month (40,000 credits, pay-as-you-go to $500/month, 50 indexes, 500 files per index). Pro $500/month (400,000 credits, pay-as-you-go to $5,000/month, 100 indexes, 2,000 files per index). Enterprise custom. Credits are $1.25 per 1,000.
Dashworks answers questions across connected wikis, chat, CRM, ticketing and code with document-level access control, and its most-used surface is a Slackbot rather than a separate search portal. It is the cheapest credible per-seat option in this category, with no seat minimum on the Team plan. The company has announced it is joining HubSpot, which should factor into any multi-year commitment.
| Deployment | Vendor SaaS only |
|---|
| Prebuilt SaaS connectors | Yes — wikis, messaging, CRM, project management, support, code |
|---|
| Source ACL enforcement | Yes — document-level access control with instant updates |
|---|
| Open source | No |
|---|
| Primary interface | Slackbot, browser extension and web app; API as an Enterprise add-on |
|---|
Watch out: The company has announced it is joining HubSpot, so the standalone roadmap, pricing and long-term availability are genuinely uncertain — a bad moment to sign a multi-year deal without an exit clause. The connector catalogue is narrower than Glean's or Microsoft's, there is no self-hosted or VPC option, and API access is a paid Enterprise add-on rather than part of the standard plans.
Team $12 per seat per month ($10 billed yearly), no seat minimum. Business $15 per seat per month ($12 billed yearly), minimum 10 seats. Enterprise quote-only. 14-day free trial, no credit card.
Moveworks, now formally acquired by ServiceNow, is an AI assistant for employee support that resolves IT and HR requests conversationally and includes enterprise search across connected systems, with Agent Studio and Knowledge Studio for building and curating on top. It is sold as a front door to service workflows rather than as a retrieval platform, and post-acquisition it is being positioned alongside ServiceNow's own agentic products. Choose it when the goal is deflecting service tickets, not when the goal is general document retrieval.
| Deployment | Vendor SaaS |
|---|
| Prebuilt SaaS connectors | Yes — ITSM, HRIS, ticketing and knowledge systems |
|---|
| Source ACL enforcement | Yes — permission-aware answers across connected systems |
|---|
| Open source | No |
|---|
| Primary interface | Chat assistant in Slack/Teams, plus enterprise search |
|---|
Watch out: The ServiceNow acquisition means you are buying into a platform consolidation rather than an independent product, and the go-forward roadmap points at ServiceNow's unified AI experience — pricing and packaging should be expected to move. It is scoped for employee service, so using it as a general-purpose developer retrieval layer fights the product, and deployments are professional-services heavy with long implementation timelines rather than something a team self-serves.
unknown — no public pricing; quoted by ServiceNow/Moveworks sales, typically as an annual enterprise agreement.
Coveo is the long-established relevance platform in this space, with unified indexing across enterprise sources, machine-learned ranking tuned on behavioural signals, a Passage Retrieval API and generative answering, packaged for commerce, customer service, websites and workplace. Its behavioural relevance tuning and analytics are more developed than anything the newer RAG vendors ship. It is a public company sold through enterprise agreements, with implementation typically running through partners.
| Deployment | Vendor SaaS |
|---|
| Prebuilt SaaS connectors | Yes — Salesforce, ServiceNow, SharePoint, Zendesk, Adobe, SAP and others |
|---|
| Source ACL enforcement | Yes — early-binding permission model on indexed sources |
|---|
| Open source | No |
|---|
| Primary interface | Search UI components and APIs (Passage Retrieval, generative answering) |
|---|
Watch out: No pricing is published anywhere, so evaluation begins with a sales cycle and the entry cost is enterprise-scale. Deployments carry real implementation weight — indexing pipelines, relevance tuning and front-end integration usually mean professional services — which makes it over-scoped for an internal employee Q&A bot where Onyx or Dashworks would land in days rather than quarters.
unknown — the pricing page publishes no figures and routes to a sales conversation; a free trial is offered.
Rovo indexes Confluence, Jira and connected third-party tools into Atlassian's Teamwork Graph and exposes search, chat and agents across them, including inside Slack and Teams. Its distinguishing commercial fact is that it is no longer a separate purchase — it is automatically activated in Standard, Premium and Enterprise cloud subscriptions with a monthly credit allowance per user, so many organisations already own it. Usage beyond the allowance is metered at $0.01 per credit.
| Deployment | Atlassian Cloud SaaS |
|---|
| Prebuilt SaaS connectors | Yes — Atlassian products plus third-party tools via Teamwork Graph |
|---|
| Source ACL enforcement | Yes — inherits Atlassian product permissions |
|---|
| Primary interface | Search and chat in Atlassian apps, Slack and Teams; agents and MCP server |
|---|
| Open source | No |
|---|
Watch out: The included credit allowances are small — 25 credits per user per month on Standard is a handful of interactions, not daily use — so realistic adoption means either a plan upgrade or metered overage, and the credit-per-action model makes spend hard to predict across a large org. It is also strongly Atlassian-centric: coverage of non-Atlassian sources is thinner than Glean's or Microsoft's, and it is not a general-purpose retrieval API you can point your own application at.
Included in Jira and Confluence Standard, Premium and Enterprise cloud plans with allowances of 25, 70 and 150 Rovo credits per user per month respectively; Teamwork and Service Collections include more. Overage is $0.01 per credit (e.g. 7,000 credits over allowance = $70). Rovo Dev Standard is $20 per developer per month including 2,000 Rovo Dev credits.