Jul 22 - Antares-350M + 1B - open-weight - 500-task vulnerability benchmark - ~$0.01/scan vs $3-$15/scan - local inference - Antares-3B coming
The decision · Benchmark your vulnerability scanning cost per commit by end of week. If you pay per-token for a frontier model to scan code, run the same repo through Antares-1B and compare precision at 1/300th the cost. For teams that avoid cloud-based code scanning for compliance reasons, Antares changes the math: it runs locally, on-prem, with no code leaving your environment. The open-weight license means you can fine-tune it on your own vulnerability taxonomy. For founders building dev-tool or security companies, Cisco's broader move (open-weight models plus Foundry Security Spec plus CodeGuard) signals a shift toward composable, auditable AI security infrastructure that can be integrated directly into CI/CD pipelines. The integration window is open.
Jul 11 - independent eval - safety test - no usable score
The decision · Stop quoting vendor benchmark numbers in your diligence. Run your own task-specific eval before locking a model into production. The gap between advertised and actual quality is where your deployment risk lives.