FRONTIER MODEL DIRECTORY & BENCHMARKS

Frontier Intelligence, Calibrated for Code

Switch models on the fly directly in your terminal without managing multiple API provider keys or paying separate monthly subscriptions. All models consume from your sovereign LEGECY CLI credit pool.

Recommended 200K ctx

Claude 3.5 Sonnet

Anthropic's flagship agentic model. Industry benchmark for multi-file refactoring, autonomous tool execution, and complex debugging.

Token Cost ~2.0 Cr / 1k
Fast Speed 128K ctx

GPT-4o Omniverse

High-throughput multi-modal model with sub-second time-to-first-token. Ideal for inline autocompletion and rapid prototyping.

Token Cost ~1.8 Cr / 1k
Ultra Value 64K ctx

DeepSeek Coder V3

State-of-the-art open weights coding specialist. Unmatched price-to-performance for large automated unit test generation suites.

Token Cost ~1.0 Cr / 1k
Deep CoT 64K ctx

DeepSeek R1 Reasoner

Reinforcement-learning tuned reasoning engine. Excels at finding subtle race conditions, memory leaks, and cryptanalysis bugs.

Token Cost ~1.5 Cr / 1k
High Reasoning 128K ctx

OpenAI o1 Reasoning

Advanced deliberate thinking model engineered for complex algorithms, system architecture design, and competitive math.

Token Cost ~4.5 Cr / 1k
Mega Context 2M ctx

Gemini 1.5 Pro

Massive 2,000,000 token context window. Ingest whole enterprise repositories, architectural diagrams, and documentation in one prompt.

Token Cost ~1.5 Cr / 1k
terminal — on-the-fly model flag ● 100% Reverse Proxy Protected
# Switch model instantly without changing keys:
user@workstation:~/repo$ legecy --model claude-3.5-sonnet "Refactor state store"
user@workstation:~/repo$ legecy --model deepseek-coder "Write unit test suite"
user@workstation:~/repo$ legecy --model gemini-1.5-pro "Ingest full monorepo"