capability update
Codex local models and Kandev agent support rechecked
The October 4 review corrects Codex local-provider support, updates its model and pricing details, and expands Kandev's verified agent list.
Codex CLI can run local models through Ollama or LM Studio. It also supports configured providers with compatible Responses API endpoints. The profile now records this model choice and local deployment option. Codex cloud remains an OpenAI-managed service. The stack builder limits self-hosted Codex recommendations to its terminal interface, or to that terminal coding layer beneath an orchestrator.
The model list now includes GPT-6.1 Sol, GPT-6 Astra, GPT-6 Sol, and GPT-6 Luna alongside the GPT-5.6 models that remain available during rollout. Availability depends on plan, client, workspace settings, and rollout. GPT-5.3-Codex-Spark retired on September 14 and is excluded.
Codex still starts at $0. The plan details now include Go at $8/month and the two-user minimum for Business. API usage is billed separately and excludes cloud features.
Kandev’s official README lists 22 supported agent CLIs. The profile now includes all of them, with iFlow marked as beta. The separate experimental Codex app-server integration is not included as a generally available agent CLI.
This was a targeted review. Other evidence dates and the existing full-profile review deadlines remain in place.
Affected profiles
Primary sources
- Codex models ↗ · checked 2026-10-04
- Codex pricing ↗ · checked 2026-10-04
- Codex provider configuration ↗ · checked 2026-10-04
- Kandev supported agent CLIs ↗ · checked 2026-10-04