Sonar
A series of large language models developed in-house by Perplexity, optimized for its AI search products and designed primarily for low-cost, high-frequency search question-answering scenarios.
Timeline (last 90 days)
The research distills three factors that help organizations scale AI: redesigning processes, the toolchain, and people.
Enterprises that have deployed AI end-to-end across the organization have already seen increased PR throughput, shortened PR cycle time, improved developer self-reported productivity, and acceleration of tasks across all parts of the SDLC.
Sonar is an agentless cloud security scanning engine launched by Orca Security, capable of discovering assets, configurations, and potential risks across cloud environments.
Sonar queries generated by Discovery AI undergo syntax validation and permission checks to ensure that no invalid or unauthorized queries are produced.
Orca Security built a natural language query agent named Discovery AI based on Amazon Bedrock, automatically converting plain English questions into Sonar query statements.
Version 5.0.0 replaces the original Sonar model IDs and provider options with the Agent API's presets, models, and tools.
Version 5.0.0 no longer supports Sonar's PDF input capability and no longer returns image and video results.
The core Sonar model series (such as sonar-small and sonar-large) layers real-time web retrieval capabilities on top of large language models.
Sonar has proposed the ACDC (Agent-Centric Development Cycle) framework, comprising three core phases: Guidance, Verification, and Solve.
Anirban Chatterjee is the Head of Product Marketing at Sonar, a code quality platform.
7 more timeline events
All Facts (17)
The research distills three factors that help organizations scale AI: redesigning processes, the toolchain, and people.
50%UnverifiedEnterprises that have deployed AI end-to-end across the organization have already seen increased PR throughput, shortened PR cycle time, improved developer self-reported productivity, and acceleration of tasks across all parts of the SDLC.
50%UnverifiedSonar is an agentless cloud security scanning engine launched by Orca Security, capable of discovering assets, configurations, and potential risks across cloud environments.
50%UnverifiedSonar queries generated by Discovery AI undergo syntax validation and permission checks to ensure that no invalid or unauthorized queries are produced.
50%UnverifiedOrca Security built a natural language query agent named Discovery AI based on Amazon Bedrock, automatically converting plain English questions into Sonar query statements.
50%UnverifiedThe core Sonar model series (such as sonar-small and sonar-large) layers real-time web retrieval capabilities on top of large language models.
50%UnverifiedVersion 5.0.0 replaces the original Sonar model IDs and provider options with the Agent API's presets, models, and tools.
50%UnverifiedVersion 5.0.0 no longer supports Sonar's PDF input capability and no longer returns image and video results.
50%UnverifiedAnirban Chatterjee is the Head of Product Marketing at Sonar, a code quality platform.
50%UnverifiedSonar is a search-augmented model launched by Perplexity that integrates real-time web retrieval capabilities with the generative capabilities of large language models
50%UnverifiedSonar is a search-augmented model launched by Perplexity, whose core feature is the deep integration of real-time web retrieval capabilities with the generative capabilities of large language models, enabling it to cite the latest web sources in its responses.
50%UnverifiedSonar（前身为SonarSource）由Olivier Gaudin和Freddy Mallet于2008年在瑞士日内瓦创立，核心产品SonarQube支持30多种编程语言
50%UnverifiedGitar 现已成为 Sonar（SonarQube/SonarCloud 母公司）的一部分
50%UnverifiedSonar（前身为SonarSource）成立于2008年，其核心产品SonarQube是全球最广泛使用的开源代码质量管理平台之一，支持30多种编程语言
50%Unverified2023年，Sonar完成了4.12亿美元的融资，估值达到47亿美元
50%UnverifiedSonar has proposed the ACDC (Agent-Centric Development Cycle) framework, comprising three core phases: Guidance, Verification, and Solve.
50%UnverifiedSonar has established a public LLM code quality leaderboard that evaluates models using approximately 4,000 coding tasks and the SonarQube metrics system.
50%