Select a name to see it on the map. Open a profile to see every fact and source.
What the source marks mean Each mark shows what kind of source backs a fact. Open the fact to read the source, the date it was checked and any conditions.
The provider's own documentation supports this fact, for the product, plan and date stated.
The provider says this, for example on a marketing page. Read the linked source and its conditions before you rely on it. ETHORITY's assessment, based on the linked sources.
An independent source supports this one fact, within the scope stated. Check the other facts separately. Figures marked 'Scenario' are calculated from stated assumptions. Results measured in practice are shown separately.
Voyage AI's voyage-code family (code-4 current; code-3, code-2 older) turns source code into Matryoshka-sized, quantisable embedding vectors so coding agents and code search can retrieve relevant files.
$0.12–0.18 /1M tokens ● Conditions apply – open for details voyage-code-4: 32,000; … ● Proprietary hosted mode… ● Conditions apply – open for details Voyage API and MongoDB … ● VA voyage-code
Voyage AI (MongoDB) · Model family
Facts, sources and conditions Voyage AI's voyage-code family (code-4 current; code-3, code-2 older) turns source code into Matryoshka-sized, quantisable embedding vectors so coding agents and code search can retrieve relevant files.
The provider’s release article states that voyage-code-4 is available and recommends it for new code-retrieval uses. https://blog.voyageai.com/2026/08/13/voyage-code-4/
Where it sits on the map · 1 · 1 Code embedding model for code retrieval / RAG
Derived from model overview.
Per 1M tokens embedded
Unit: USD Billed to the Atlas account linked to the API key.
voyage-code-4 $0.12 / 1M tokens; voyage-code-3 $0.18 / 1M; voyage-code-2 $0.12 / 1M. EU/US-scoped endpoints +10%.
Unit: USD 10% uplift from https://www.mongodb.com/docs/voyageai/api-reference/overview/.
Europe Geography endpoint eu.ai.mongodb.com (EEA cloud regions; GA 29 Sep 2026), inference and retained payload logging within Europe; requires paid tier; 'Scoped endpoints don't serve every model'
Geography details from https://www.mongodb.com/company/blog/product-release-announcements/europe-data-residency-for-atlas-embedding-reranking. Availability of voyage-code models on the EU endpoint NOT confirmed on reviewed pages. Account metadata shared globally.
Used unless you opt out — Customer Content may be used "to train, improve, and otherwise further develop the Service" unless you opt out
Opt-out applies prospectively and requires a payment method. Atlas has a separate model-improvement setting.
Voyage API and MongoDB Atlas Embedding and Reranking API (ai.mongodb.com); positioned as complement to grep-style search for coding agents. No IDE, Git host or CI/CD integrations named.
Quantization and shorter embeddings reduce retrieval quality; free-trial rate limits 3 RPM / 10K TPM without payment method; scoped endpoints have lower rate limits and don't serve every model
Rate-limit statements from https://www.mongodb.com/docs/voyageai/api-reference/overview/.
Context window voyage-code-4: 32,000; voyage-code-3: 32,000; voyage-code-2: 16,000
Unit: tokens Licence and access rights Proprietary hosted models; no licence stated; open weights: no (code models not listed as open-weight)
Only voyage-4-nano is open-weight on that page. Exact licence name not applicable/not stated.
Model access Voyage API and MongoDB Atlas Embedding and Reranking API (ai.mongodb.com; eu./us. scoped endpoints); not downloadable; marketplaces not confirmed
Deployment requirements Hosted API (Voyage API / MongoDB Atlas Embedding and Reranking API, SaaS); no open weights for code models
Only voyage-4-nano listed as open-weight. EU post lists air-gapped/self-managed deployment as out of scope.
Published evaluations and scope voyage-code-4 (NDCG@10, avg relative improvement): vs voyage-code-3 "27.54%" agentic / "13.98%" traditional; vs OpenAI v3 large "48.58%" / "40.06%"; vs Gemini Embedding 2 "31.03%" / "16.01%"; vs Cohere Embed v4 "28.25%" / "19.21%". voyage-code-3: outperforms OpenAI-v3-large and CodeSage-large by "13.80% and 16.81%" across 32 datasets.
Where the provider's own sources disagree The provider’s own sources say different things here. Provider claims, not verified. voyage-code-3 figures from https://blog.voyageai.com/2024/12/04/voyage-code-3/. Minor Source conflict: code-3 blog cites 32 datasets, code-4 blog says code-3 was evaluated on 28 datasets.
Documented limitations Quantization and shorter embeddings reduce retrieval quality; free-trial rate limits 3 RPM / 10K TPM without payment method; scoped endpoints have lower rate limits and don't serve every model
Rate-limit statements from https://www.mongodb.com/docs/voyageai/api-reference/overview/.
Grok Build (grok-code-fast-1) xAI · Model
xAI's grok-code-fast-1 (now aliased to grok-build-0.1) is a fast reasoning model with native tool calling, built to run multi-step agentic coding loops inside coding tools via an OpenAI-compatible API.
$1 in · $2 out /1M tok ● Conditions apply – open for details 256,000 ● Conditions apply – open for details xAI API (OpenAI-compati… ● Conditions apply – open for details Hosted API (SaaS) only … ● Conditions apply – open for details Grok Build (grok-code-fast-1)
xAI · Model
Facts, sources and conditions xAI's grok-code-fast-1 (now aliased to grok-build-0.1) is a fast reasoning model with native tool calling, built to run multi-step agentic coding loops inside coding tools via an OpenAI-compatible API.
The model page is now headed Grok Build0.1 (grok-build-0.1), with grok-code-fast-1 listed as an alias. Provider pages name SpaceXSI. https://docs.x.ai/developers/models/grok-code-fast-1
Where it sits on the map · 1 · 1 Agentic coding model (tool-calling code agent backend)
Derived from prompt-engineering guide and model card.
Per 1M tokens (input, cached input, output)
Unit: USD Requests above 200K context charged at different, unlisted rates; regional pricing mentioned but not shown.
Input $1.00 / cached input $0.20 / output $2.00 per 1M tokens
Unit: USD Grok Build 0.1, alias grok-code-fast-1, as described on the model page.
No EU region stated; model page lists regions us-east-1 and us-west-2. Default endpoint api.x.ai 'does not guarantee a processing region'; US-only endpoint (us.api.x.ai) serves only grok-4.7 and grok-4.6.
Region statements from https://docs.x.ai/developers/faq/security. No EU processing option found in reviewed pages.
Not used for training — "never trains on your API inputs or outputs without your explicit permission"
Enterprise ToS s.3.1: will not use any User Content to train foundation models/LLMs (https://x.ai/legal/terms-of-service-enterprise). Qualifier: 'without your explicit permission'.
OpenAI-compatible API with native tool/function calling; described as a pair-programmer 'inside most common coding tools' and offered via unnamed 'launch partners' / third-party partners. No IDE, Git host or CI/CD named on reviewed pages.
The model card describes availability through third-party partners.
SOC 2 Type 2 (self-stated: "We are SOC 2 Type 2 compliant"); scope and auditor not stated
Provider describes SOC 2 access under NDA. SOC 2 is an attestation.
Reasoning traces only accessible in streaming mode; XML-based tool-call output 'may hurt performance'; Grok 4 positioned as better for one-shot Q&A and deep debugging; Batch API not supported; model card reports MASK dishonesty rate 71.9 (higher than Grok 4) and agentic misuse risk.
Batch API from model page; MASK and risk statements from https://data.x.ai/2025-08-26-grok-code-fast-1-model-card.pdf.
Context window 256,000
Unit: tokens Pricing differs above 200K. Model card does not state context window.
Model access xAI API (OpenAI-compatible) and third-party partners; regions us-east-1, us-west-2; no cloud marketplace or download stated
Partners from model card https://data.x.ai/2025-08-26-grok-code-fast-1-model-card.pdf; not named.
Deployment requirements Hosted API (SaaS) only on reviewed pages; no open weights or self-hosting stated
No self-hosted/VPC option found in reviewed sources.
Published evaluations and scope Model card reports safety/dual-use evals only, e.g. "AgentHarm answer rate: 17.0", "AgentDojo attack success rate: 26.9", "MASK dishonesty rate: 71.9", "CyBench unguided success rate: 22.5". Guide claims "up to 4x the speed and 1/10th the cost of other leading agentic models". No coding benchmark (e.g. SWE-bench) on reviewed pages.
Performance and cost figures are provider claims; test against your own workload.
Documented limitations Reasoning traces only accessible in streaming mode; XML-based tool-call output 'may hurt performance'; Grok 4 positioned as better for one-shot Q&A and deep debugging; Batch API not supported; model card reports MASK dishonesty rate 71.9 (higher than Grok 4) and agentic misuse risk.
Batch API from model page; MASK and risk statements from https://data.x.ai/2025-08-26-grok-code-fast-1-model-card.pdf.
MO Morph Fast Apply Morph · Model
Model that replaces the file-writing layer in AI coding agents and IDE plugins, available as an MCP tool for Claude Code, Cursor and similar environments.
$0.80–1.90 /1M tokens ○ 262K (both morph-v3-fas… ○ Proprietary; open weigh… ○ Conditions apply – open for details Morph API (OpenAI-compa… ● Conditions apply – open for details MO Morph Fast Apply
Morph · Model
Facts, sources and conditions Model that replaces the file-writing layer in AI coding agents and IDE plugins, available as an MCP tool for Claude Code, Cursor and similar environments.
The provider describes a 7B model and claims 10,500 tokens/second with 98% accuracy for merging code edits. These are provider figures, not independent results for your repository.
Where it sits on the map · 1 · 1 Code edit merge (apply) model for coding agents
Derived from product page.
Per 1M tokens (input, output); credit-based plans
Unit: USD morph-v3-fast $0.80 input / $1.20 output per 1M; morph-v3-large $0.90 / $1.90 per 1M; paid plans from $20/month incl. $20 credits
Unit: USD No EU region named; 'primary data processing occurs in the United States'; regional data centers available to paid/enterprise customers with residency requirements; self-hosted keeps processing in customer environment
Specific regions for regional data centers not stated.
Not used for training — "We do not use your data to train our models" (except support, debugging, service-quality fixes)
ToS s.5 (https://www.morphllm.com/privacy/tos): 'We do not use your code or content to train our models' with same support/debugging exception. Enterprise: 'Zero training'.
OpenAI-compatible API (api.morphllm.com/v1); TypeScript and Python SDKs; MCP setup (edit_file, codebase_search) for Claude Code, Codex, Cursor, VS Code
Product page names JetBrains, Vercel, Webflow as users (customer logos, not integrations). No Git host or CI/CD integrations.
Self-stated: "SOC 2 Type II, ISO 27001 compliance maintained" (auditor/scope not stated); docs list 'SOC2' for enterprise
Compliance wording describes alignment; confirm certificate scope. SOC 2 is an attestation.
Sections missing the '// ... existing code ...' marker are treated as deletions; ToS: 'Model outputs may be inaccurate, incomplete, or misleading'
ToS s.9.2 https://www.morphllm.com/privacy/tos.
Context window 262K (both morph-v3-fast and morph-v3-large)
Unit: tokens Licence and access rights Proprietary; open weights: no (listed under 'Specialized', not 'Open Source Models'); ToS: 'No license is granted except as explicitly described in these Terms'
Paid/Enterprise terms include bounded IP defence; outputs have a separate IP warranty exclusion. Confirm the applicable model licence.
Model access Morph API (OpenAI-compatible), SDKs, MCP; models morph-v3-fast, morph-v3-large, auto router; no cloud marketplace or download stated
Access paths also from https://docs.morphllm.com/introduction.
Deployment requirements Hosted API (SaaS); enterprise: dedicated instances, self-hosted deployments, zero data retention
Dedicated-inference page (https://www.morphllm.com/dedicated-inference) does not list Fast Apply in its dedicated catalog and mentions no self-hosted option — self-hosting of Fast Apply not specifically confirmed.
Published evaluations and scope "98% accuracy" merge accuracy (v3-large) vs 82–94% for other listed models; "10,500 tokens per second" (v3-fast), 5,000+ tok/s (v3-large); ~0.8s for a 500-line file; '100% merge success' with five frontier planning models
Provider claims, not verified; benchmark name/method not stated on page.
Documented limitations Sections missing the '// ... existing code ...' marker are treated as deletions; ToS: 'Model outputs may be inaccurate, incomplete, or misleading'
ToS s.9.2 https://www.morphllm.com/privacy/tos.
SN Snyk DeepCode AI Snyk · Product
According to Snyk: “Snyk's DeepCode AI code analyzer was designed to find, autofix, and prioritize vulnerabilities”
Monthly · credits (Ent.) ○ Conditions apply – open for details Free · from $25 /mo ○ Conditions apply – open for details SN Snyk DeepCode AI
Snyk · Product
Facts, sources and conditions According to Snyk: “Snyk's DeepCode AI code analyzer was designed to find, autofix, and prioritize vulnerabilities”
The product page was inaccessible. Documentation names the autofix function Snyk Agent Fix, formerly DeepCode AI Fix; confirm the relevant product and plan.
Where it sits on the map · 1 · 1 SAST vulnerability detection and AI autofix
Free/Team per month (Team for up to 10 developers); Enterprise prepaid credits (“1 credit = $1”) per “Active Contributor / per Day”
DeepCode AI/Agent Fix inclusion per plan not stated
Free $0/month (Snyk Code 100 tests/month); Team “Starting at” $25/month (1,000 tests/month); Enterprise “Contact Sales”
Unit: USD Platform plans, not DeepCode-specific
“Snyk EU” region; Snyk Code source cache 24–48 h for AWS tenants in EU, AU and Private Tenant
Not used for training – “Snyk does not use customer proprietary software code to train, optimize, fine-tune, or improve any AI models”
Fix model trained on permissively licensed public repositories (https://docs.snyk.io/snyk-data-and-governance/how-snyk-handles-your-data)
IDE plugins: VS Code, Visual Studio, Eclipse, JetBrains IDEs; Jira (Team plan); Snyk Code repository scanning
ISO 27001, SOC2 Type II (“We uphold ISO 27001, SOC2 Type II, and GDPR compliance”)
Company-level; scope and auditor not stated; SOC 2 is an attestation; GDPR is a law
“Snyk Agent Fix focuses on single-file fixes”; some large-file requests up to 2 minutes; no suggestion shown if it cannot meet “security and functional benchmarks”; “Human review required”
AC Auggie CLI Augment Code · Agent
Terminal agent from Augment Code that brings its codebase context engine to the command line, running interactively or in --print mode for CI pipelines such as GitHub Actions PR reviews.
Credits · monthly plans ○ Conditions apply – open for details $20–200 /mo ○ Conditions apply – open for details AC Auggie CLI
Augment Code · Agent
Facts, sources and conditions Terminal agent from Augment Code that brings its codebase context engine to the command line, running interactively or in --print mode for CI pipelines such as GitHub Actions PR reviews.
Description follows provider documentation.
Where it sits on the map · 1 · 1 Terminal coding agent with codebase context
credits (“A credit is the unit we use to measure usage in Augment Code”), monthly plans
Blog dated Oct 2025 announcement; current pricing page
Indie $20 (40,000 credits); Standard $60 (130,000); Max $200 (450,000); Enterprise custom
Unit: USD/month Per Oct 2025 pricing-change blog; not confirmed against current pricing page
Not used for training – “We never train on our customer's proprietary data.”
Confirm which plan includes the feature you need.
GitHub Actions (PR Description, PR Review workflows); native integrations GitHub, Linear, Notion; MCP (client and --mcp server); Augment for VS Code / JetBrains IDEs; Python SDK
SOC 2 Type II “Attested” (auditor not named); “ISO/IEC 42001-certified Artificial Intelligence Management System”
Provider claim, not registry-verified; GDPR/CCPA/HIPAA listed as “Compliant” are laws, not certifications
“Non-interactive mode is an additional feature that may be disabled if it is not included in your agreement (enterprise)”; requires Node 20+; permissions “not enforced in the Augment code extension”
StackBlitz tool, from the team behind WebContainers, that builds full-stack web applications using a wide range of JavaScript web frameworks.
AI tokens per month ○ Free · $18 /mo (annual) ○ Facts, sources and conditions StackBlitz tool, from the team behind WebContainers, that builds full-stack web applications using a wide range of JavaScript web frameworks.
The accessible introduction also describes mobile apps through Expo. The originally listed introduction URL was inaccessible. https://support.bolt.new/building/intro-bolt
Where it sits on the map · 1 · 1 Prompt-to-app full-stack builder
AI tokens per month; Teams per member; Enterprise “billed by consumption”
Free $0 (1M tokens/month, 300K daily limit); Pro $18/month billed annually (10M tokens); Teams $27/month per member billed annually; Enterprise custom
Unit: USD Used unless you opt out – “Unless you opt out … we may use Bolt Model Development Content”; EEA/UK/CH accounts excluded; Team/Enterprise under MSA/DPA excluded
Bolt Forge: prompts/code “de-identified, then stored and used by Bolt's partners to train open-source AI models” (opt-out not applicable while Forge enabled) – https://support.bolt.new/account-and-subscription/bolt-forge
GitHub (version control, repo import), Supabase, Stripe, Figma, Expo, Google SSO, MCP servers, Google Stitch, Lovable import; bolt.host publishing and custom domains
No CI/CD integrations stated
SOC 2 Type 2 (“Bolt is SOC 2 Type 2 certified”); auditor not named
SOC 2 is an attestation; enterprise page says “HIPAA, FedRAMP, and SOC 2 ready” – “ready” is not a certification; GDPR/CCPA are laws
“LLMs are predictive, not deterministic”; “Sometimes, they generate false information”; may not know latest framework versions; Code view desktop-only; AI Output “may be inaccurate, incomplete, insecure”
GI GitLab Duo Agent Platform GitLab · Platform
GitLab's agent layer for GitLab.com, Self-Managed and Dedicated where agents and flows turn issues into merge requests, review code and fix CI/CD pipelines, billed via GitLab Credits.
GitLab Credits (usage) ● $1 per credit ● Conditions apply – open for details GI GitLab Duo Agent Platform
GitLab · Platform
Facts, sources and conditions GitLab's agent layer for GitLab.com, Self-Managed and Dedicated where agents and flows turn issues into merge requests, review code and fix CI/CD pipelines, billed via GitLab Credits.
Description follows the cited product documentation.
Where it sits on the map · 1 · 1 DevOps lifecycle AI agents and flows
GitLab Credits (usage-based); on-demand “list price of $1 per credit used”; Agent Platform Self-Hosted: “flat fee per customer” (ELA)
$1 per credit on-demand; e.g. on GitLab-managed models one credit covers 4 Code Review Flow executions; Premium/Ultimate credited $12/$24 per user per month (“for a limited time”)
Unit: USD Included-credit amounts stated in GA blog only; docs say the number per seat is not stated
Not used for training – “GitLab does not train generative AI models”; sub-processors “restricted from using model input and output to train models”
GitLab (issues, MRs, pipelines); IDEs: VS Code, JetBrains IDEs, Cursor, Windsurf; external agents Claude Code and Codex CLI; MCP; LDAP/SAML
ISO/IEC 27001:2022 (GitLab.com, GitLab Dedicated); ISO/IEC 27017:2015 and 27018:2019 “compliance”; SOC 2 Type 2 report (Security, Confidentiality, Availability) for GitLab.com and Dedicated; CSA STAR
Company/SaaS-level, not Agent-Platform-specific; SOC 2 is an attestation; auditor not named in text (certificate links point to Schellman directory – not verified); GDPR is a law, not a certification
“Beta features that don't consume credits can change to generally available with usage billing at any time”; some agents/flows beta or experimental; “In GitLab 18.9 and earlier, you cannot use the Agent Platform with the GitLab Duo Enterprise add-on”
Mellum JetBrains · Model family
JetBrains' 4B-parameter multilingual model, trained from scratch for code completion, powering cloud-based completion in JetBrains IDEs.
Free (Apache 2.0) ○ Conditions apply – open for details Apache 2.0 (open-source… ○ Sources disagree – open for details Downloadable (Hugging F… ○ Open weights (Hugging F… ○ Mellum
JetBrains · Model family
Facts, sources and conditions JetBrains' 4B-parameter multilingual model, trained from scratch for code completion, powering cloud-based completion in JetBrains IDEs.
This description concerns Mellum-4B. The family also includes Mellum2, described as 12B MoE with 2.5B active parameters for routing, RAG and subagents. The later Mellum2.1 page was inaccessible. https://blog.jetbrains.com/ai/2026/06/mellum2-goes-open-source-a-fast-model-for-ai-workflows/
Where it sits on the map · 1 · 1 Code-completion / fast workflow LLM (open weights)
Bedrock: compute only (“no additional model fees”); in IDE: AI Pro subscription
Open weights “free to use under the Apache 2.0 license”; Bedrock “no additional model fees”
No AI Pro price read (not in scope pages)
Detailed code-related data (if collected) “remains stored within the European Economic Area (EEA)”
Concerns collected telemetry data, not inference hosting region for cloud completion; no inference region found
Only with opt-in – commercial users: “explicit, informed consent has been obtained in advance”; orgs: “disallowed by default”
Where the provider's own sources disagree The provider’s own sources say different things here. Contradiction: data-collection notice says individual non-commercial users are opted in by default (can opt out); data-handling page says setting is off by default. Model trained “solely on publicly available, permissively licensed code” (https://blog.jetbrains.com/blog/2024/10/22/introducing-mellum-jetbrains-new-llm-built-for-developers/)
JetBrains AI Assistant / cloud completion in JetBrains IDEs; local completion in paid IntelliJ-based IDEs; Amazon Bedrock Marketplace; Hugging Face
No Git host/CI integrations (model)
“Mellum isn't a plug-and-play solution”; Mellum2 “is not multimodal”; 2024 languages: Java, Kotlin, Python, Go, PHP
Licence and access rights Apache 2.0 (open-source Hugging Face versions, Mellum-4B and Mellum2) – open weights: yes
Where the provider's own sources disagree The provider’s own sources say different things here. Contradiction over time: 2024 launch post calls Mellum “a proprietary large language model”; open-sourced 2025. HF model card https://huggingface.co/JetBrains/Mellum-4b-base
Model access Downloadable (Hugging Face); Amazon Bedrock Marketplace; JetBrains AI Assistant in IDEs
Deployment requirements Open weights (Hugging Face), self-hosted/local (“Run Mellum2 locally or self-host it”); Amazon Bedrock Marketplace deployment on chosen GPU; JetBrains cloud completion (SaaS)
Published evaluations and scope Provider-stated Mellum-4B-base: HumanEval Infilling single-line 66.2, multi-line 38.5; RepoBench 1.1 (2K, py) 28.2; SAFIM avg 38.1. In-product: “acceptance rate … about 40%”, latency “one third of its previous time”
Provider benchmark charts; evaluate on your own workload.
Documented limitations “Mellum isn't a plug-and-play solution”; Mellum2 “is not multimodal”; 2024 languages: Java, Kotlin, Python, Go, PHP
SO SonarQube AI CodeFix Sonar · Product
According to Sonar: “AI CodeFix then proposes an edit that resolves the problem without changing the code's functionality.”
In Team/Enterprise plans ● Conditions apply – open for details From $34 /mo ○ Sources disagree – open for details SO SonarQube AI CodeFix
Sonar · Product
Facts, sources and conditions According to Sonar: “AI CodeFix then proposes an edit that resolves the problem without changing the code's functionality.”
The description is supported by the SonarQube Server documentation; the Cloud documentation was inaccessible. Confirm which deployment the terms cover. https://docs.sonarsource.com/sonarqube-server/ai-capabilities/ai-codefix
Where it sits on the map · 1 · 1 LLM-generated fixes for static-analysis issues
Derived from Server AI CodeFix doc.
AI CodeFix available within SonarQube Cloud Team/Enterprise plans and Server Enterprise/Data Center editions; Cloud Team priced per month by lines of code analysed; monthly AI CodeFix usage allowance
Unit: USD Plan pricing basis from https://www.sonarsource.com/plans-and-pricing/. No separate AI CodeFix price found.
SonarQube Cloud Team 'Starts at $34 monthly' for up to 100k LOC; Enterprise custom. No separate AI CodeFix charge stated.
Unit: USD Where the provider's own sources disagree The provider’s own sources say different things here. CONTRADICTION on same page: FAQ says 'prices start at $32 per month (previously listed at $65)'. Pricing page does not mention AI CodeFix.
SonarQube Cloud EU region (default) hosted in Frankfurt, Germany (AWS eu-central-1), backups in AWS eu-west-1; US region (sonarqube.us) Enterprise-only. Code snippet sent to the selected LLM; LLM processing location not stated. Server: self-hosted LLM endpoint option.
Region list also at https://docs.sonarsource.com/sonarqube-cloud/getting-started/choosing-your-region (AI CodeFix availability per region not stated). AI terms mention only 'cloud servers' — LLM inference location not confirmed as EU.
SonarQube Cloud / Server UI ('Generate AI fix'); SonarQube for IDE connected mode in VS Code and IntelliJ ('Open in IDE'); open fix as pull request in Azure DevOps (Cloud bound to Azure DevOps)
ISO 27001:2022 certification and SOC 2 Type II attestation 'for all products and services' (auditor not named)
Self-stated; not registry-verified. AWS certifications on same page are the host's, not Sonar's. Iron Bank acceptance is not a certification. SOC 2 report under NDA.
Monthly usage allowance (resets 1st of month; error when reached); Suggestions 'may be incorrect, incomplete, or contain vulnerabilities'; non-recommended models 'may produce unexpected fix suggestions'; self-hosted LLMs subject to provider rate limits
Suggestion caveat from AI Annex s.5.1 https://www.sonarsource.com/legal/ai/; model caveat from enable-ai-codefix doc.
TR Tricentis Tosca AI agents Tricentis · Product
According to Tricentis: “AI agents built for the Tosca environment generate complete, end-to-end tests from plain-language descriptions.”
AI credits per action ● Conditions apply – open for details Creates new test cases … ● Conditions apply – open for details TR Tricentis Tosca AI agents
Tricentis · Product
Facts, sources and conditions According to Tricentis: “AI agents built for the Tosca environment generate complete, end-to-end tests from plain-language descriptions.”
The product page was inaccessible. The Tosca Cloud documentation describes test cases from natural language: https://docs.tricentis.com/tosca-cloud/en-us/content/ai_integration/agentic_test_automation/landing_page.htm
Where it sits on the map · 1 · 1 Agentic generation of end-to-end UI test cases
Derived from docs.
AI credits — each agent interaction (click, text input, verification, buffer action) costs one AI credit
Unit: credits 'All interactions count, even if they don't match your original test intent.' Tosca Cloud licence required.
Yes — data processed in region chosen at Tosca Cloud tenant setup; Azure OpenAI in Europe (France, EU-WEST-3); Anthropic Claude region 'Europe' (no code); also US and Australia
Trust page https://www.tricentis.com/trust/ai: administrators may select EU, US or Australia region. Doc lists Japan under AP-SOUTHEAST-2 and Canada under US-EAST-1 (provider labelling).
Not used for training — "we don't use your queries for model training purposes"; opt-out available for anonymised product-improvement use
Trust AI page (https://www.tricentis.com/trust/ai): inputs/outputs not used to improve third-party models; customers can opt out of anonymised inputs/outputs being used for Tricentis product improvement (i.e. default use for product improvement, scope vs 'training' not specified).
Tosca Cloud; qTest (imported manual tests via 'Automate with ATA'); Test outline API; TBox and Vision AI engines; supported apps: generic web, SAP Fiori, SAP GUI, Salesforce. No IDE/Git/CI integration stated for the agent.
qTest/API from landing page doc.
ISO/IEC 42001 scope includes Agentic Test Automation; ISO/IEC 27001:2022, ISO 27701, ISO 9001 and SOC 2 Type 2 cover Tosca Cloud (ISO also Tosca On Prem); auditors not named
Provider statement; SOC 2 is an attestation.
Creates new test cases only (cannot update existing); one AI credit per interaction, tool unavailable when credits run out; requires 1920x1080 at 100% scale (pixel-level screen reading), Chrome/Edge; attachments ≤4 MB text-based; AI Solutions prohibited in 'high risk' deployments
Credit/display limits from landing page doc; high-risk prohibition from https://www.tricentis.com/trust/ai.
ZE Zencoder Zencoder · Platform
Zencoder runs multi-model coding agents in VS Code and JetBrains IDEs and as headless CI workers, indexing several repositories to turn Jira, Linear or GitHub Issues tickets into multi-file changes.
Per seat/mo + credits ● Conditions apply – open for details $40–195 /mo · free BYOK ● ZE Zencoder
Zencoder · Platform
Facts, sources and conditions Zencoder runs multi-model coding agents in VS Code and JetBrains IDEs and as headless CI workers, indexing several repositories to turn Jira, Linear or GitHub Issues tickets into multi-file changes.
Headless CI agents are described in the provider’s CI/CD automation guide. https://docs.zencoder.ai/learn/enterprise-best-practices/ci-cd-automation.
Where it sits on the map · 1 · 1 AI coding agent platform (IDE and CI)
Derived from product page and docs.
Per seat per month plus monthly credit allowance consumed by LLM usage (premium models consume more credits); BYOK calls don't consume credits
Unit: USD Model multipliers 0.25x–5x on https://docs.zencoder.ai/features/models.
Free $0 (BYOK only); Pro $45/mo ($40 annual), 30,000 credits; Pro Plus $95 ($85), 80,000 credits; Pro Max $195 ($175), 180,000 credits; Enterprise custom (pre-paid commit)
Unit: USD per seat/month Only with opt-in (trial/paid: training only "with Your further consent"); free accounts: used — data may be used to "fine-tune or train AI models"
Where the provider's own sources disagree The provider’s own sources say different things here. ToS s.4.1, last updated June 15, 2026 (entity: For Good AI Inc., dba Zencoder Inc.). Source conflict: enterprise page states "zero model training on your data" and FAQ "does not train its models on your code or data" (https://zencoder.ai/enterprise) without the free-account caveat.
IDEs: VS Code, IntelliJ, PyCharm, WebStorm, Android Studio (JetBrains). Issue trackers: Jira, Linear, GitHub Issues. Git hosts: GitHub, GitLab, Bitbucket (CI triggers). CI/CD: headless agents in GitHub/GitLab/Bitbucket pipelines via workflow file or webhook. CLIs: Zencoder CLI, Claude Agent, OpenAI Codex.
Git/CI from https://docs.zencoder.ai/learn/enterprise-best-practices/ci-cd-automation and https://zencoder.ai/enterprise.
Self-stated: SOC 2 Type II, ISO 27001, ISO 42001 (auditors/scope not stated)
Where the provider's own sources disagree The provider’s own sources say different things here. The page claims SOC 2 Type II but its FAQ describes the report as in progress. Confirm the current attestation; ISO 27001 is a separate claim.
Model availability depends on plan and is 'subject to change'; model providers 'may retain prompts, code inputs, files, context, outputs' for limited periods; trial credits don't carry over; ToS: no warranty of 'noninfringement of the Output'
Trial carry-over follows the pricing FAQ; warranty terms follow the service agreement.
CO CodeRabbit CodeRabbit · Product
AI pull-request reviewer for GitHub, GitLab, Bitbucket and Azure DevOps that automatically reviews every PR, ranks incoming PRs via Triage and opens security fixes as PRs.
Per user per month ○ $24–72 /user/mo ○ CO CodeRabbit
CodeRabbit · Product
Facts, sources and conditions AI pull-request reviewer for GitHub, GitLab, Bitbucket and Azure DevOps that automatically reviews every PR, ranks incoming PRs via Triage and opens security fixes as PRs.
Where it sits on the map · 1 · 1 AI pull-request code review
per user per month (developer seat); annual plans “saves 20%”
Essentials $24/user/mo; Team $48/user/mo; Advanced $72/user/mo; Enterprise “Contact Us”
Unit: USD Not used for training – “Neither CodeRabbit nor its Third-Party AI Model Providers use Customer's code to train any AI models.”
Consistent with docs FAQ: “CodeRabbit never uses customer code for training”
GitHub, GitLab (cloud and self-hosted), Bitbucket, Azure DevOps; VS Code extension (also Cursor, Windsurf); CLI
CI/CD integrations not stated on reviewed pages
“100% accuracy isn't guaranteed due to AI's evolving nature”; works with all languages “with varying proficiency”
