Cisco Introduces Antares AI Models for Source-Code Vulnerability Triage
Summary
Cisco announced open-weight Antares models for locating known vulnerabilities in codebases and released a 500-entry benchmark. Cisco says the models cost less than GPT-5.5 and GLM-5.2 in its tests.
Key points
- Cisco Foundation AI announced two open-weight small language models, Antares-350M and Antares-1B, designed to locate known vulnerabilities in codebases.
- The models use a repository-search process starting from a vulnerability description and return a ranked list of likely relevant source files to support initial triage.
- Cisco says the open-weight models can be used without moving the analyzed repository out of the organization’s possession, supporting data-sovereignty needs.
- Cisco released a 500-entry Vulnerability Localization Benchmark involving unfamiliar codebases and vulnerability patterns associated with specific CWE categories.
- In Cisco's benchmark comparison, Antares reportedly cost 172 times less than GPT-5.5 and 15.2 times less than GLM-5.2; these are Cisco-reported results.
Article Details
- Event Type
- Cisco announced the release of the Antares-350M and Antares-1B open-weight small language models for source-code vulnerability localization, alongside a new benchmark.
- Impact
- The models are intended to help security teams identify files that may contain known vulnerabilities and prioritize triage while keeping code under the user's control. Cisco's benchmark claims Antares runs at lower cost than the compared models—172 times cheaper than GPT-5.5 and 15.2 times cheaper than GLM-5.2. The article does not independently verify these performance and cost claims.
People
Vendors
CiscoCisco Foundation AI has announced Antares, a small language model (SLM) named after the brightest star in the Scorpius constellation.OpenAIUsing this benchmark, Cisco tested Antares against a top performing closed LLM frontier model (OpenAI’s GPT-5.5) and a top performing open weight model (Z.ai’s GLM-5.2) and found that it completes faster at lower costZ.aia top performing closed LLM frontier model (OpenAI’s GPT-5.5) and a top performing open weight model (Z.ai’s GLM-5.2) and found that it completes faster at lower cost than its rivals (172x cheaper than the LLM and
Products
Antares-1BJust as the Antares star is bright because it is a binary star, so Cisco’s Antares family currently comprises two models: Antares-350M and Antares-1B.Antares-350MJust as the Antares star is bright because it is a binary star, so Cisco’s Antares family currently comprises two models: Antares-350M and Antares-1B.GLM-5.2performing closed LLM frontier model (OpenAI’s GPT-5.5) and a top performing open weight model (Z.ai’s GLM-5.2) and found that it completes faster at lower cost than its rivals (172x cheaper than the LLM and 15.2xGPT-5.5Using this benchmark, Cisco tested Antares against a top performing closed LLM frontier model (OpenAI’s GPT-5.5) and a top performing open weight model (Z.ai’s GLM-5.2) and found that it completes faster at lower cost