Model control center

Providers & models.

Veyra orchestrates across the best models for each task. Use defaults, pin a provider per agent, or bring your own.

Supported

Currently available

Codex 5.3

Stable

Coder · Reviewer

Best-in-class code generation with strong refactor reasoning. Ideal for repo-scale edits.

Speed

Balanced

Reasoning

High

ChatGPT 5.4

Stable

Planner

Long-horizon reasoning and complex task decomposition.

Speed

Standard

Reasoning

Very High

ChatGPT 5.4 Mini

Stable

Operator

Fast, low-cost execution for tool calls and short edits.

Speed

Fast

Reasoning

Medium

DeepSeek V3.2

Stable

Coder

Open-weight code model tuned for repo-scale tasks. Strong cost/perf.

Speed

Fast

Reasoning

High

Grok 2 Reasoning

Beta

Reviewer

Aggressive critique and edge-case discovery.

Speed

Standard

Reasoning

Very High

Claude Sonnet 4

Stable

Planner · Reviewer

Reliable reasoning with excellent instruction following.

Speed

Balanced

Reasoning

Very High

Gemini 2.5 Pro

Stable

Coder

Massive context window for whole-repo reasoning.

Speed

Standard

Reasoning

High

Llama 3.3 70B

Beta

Operator

Self-hostable open-weight workhorse.

Speed

Fast

Reasoning

Medium

+ Custom endpoint

Open

Custom

Connect any OpenAI-compatible API.

Speed

—

Reasoning

—

Routing

Adaptive provider routing.

Veyra selects the cheapest model that meets the task's reasoning bar — and escalates only when needed.

Task class	Default	Escalation	Notes
Trivial edit	ChatGPT 5.4 Mini	Codex 5.3	One-line, low-risk
Refactor	Codex 5.3	Claude Sonnet 4	Symbol-graph aware
Architecture	ChatGPT 5.4	Grok 2 Reasoning	Long horizon
Review	Claude Sonnet 4	Grok 2 Reasoning	Adversarial critique
Operator/Tools	ChatGPT 5.4 Mini	Llama 3.3 70B	Cost-sensitive