INTEGRATIONS ARCHITECTURE

Supported AI Model Integrations

CodeXa Studio IDE connects directly to leading model providers through official APIs and local on-device runtimes. All paid plans include universal access to enabled models with transparent token weighting and BYOK support.

AI ECOSYSTEM • MULTI-MODEL INTEGRATION

Intelligence, connected.

CodeXa Studio IDE integrates supported AI models into one development workspace.

HOST ECOSYSTEM

CodeXa v2.4

CodeXa Agency

IDE Architecture & Core

CodeXa Agency orchestrates multi-agent coding workflows, local in-memory AST symbol caching, and universal access across frontier AI model families without vendor lock-in.

Dual-Agent Planner

Deterministic code synthesis with atomic hunk review

Zero Telemetry BYOK

Direct encrypted endpoints with local OS Keychain storage

In-Memory AST Index

Sub-millisecond monorepo navigation & rename propagation

Agency RuntimeArchitecture Docs

CodeXa Engine

SUPPORTED PROVIDERIntegrated Model Runtime

OpenAI

GPT Models

Deconstruct deep monorepo refactors, execute Directed Acyclic Graph (DAG) task planning, and generate cross-file code updates with automatic rollback safeguards.

VERIFIED CONFIGURED MODELS

GPT-6 AstraConfigured Frontier Preview

256k

GPT-5.6 SOLHigh-Throughput Synthesis

128k

GPT-4oMultimodal Code & Visual UI

128k

Integrated supported model runtime • Deducts from token allowanceCatalog

MODEL SPECIFICATIONS

Flagship Model Catalogue

Verified model families supported in CodeXa Studio IDE. Access is universal across all four plan tiers.

Universal Model Access Policy: Every CodeXa subscription tier (Free, Starter, Pro, Max) includes access to all enabled models. Cloud model usage is governed by transparent monthly token quotas, while native NEXA models run locally with unlimited zero-token usage.

GPT-6 Astra

OpenAI

256k Tokens

Configured Frontier Preview

Multi-file DAG task decomposition & whole-systems architecture

OPTIMIZED WORKFLOWS:

Autonomous monorepo refactoring

Architectural planning with rollback safety

Full-repo syntactic code synthesis

Monthly Token Quota (All Tiers: Free, Starter, Pro, Max)

GPT-5.6 SOL

OpenAI

128k Tokens

Active Supported Integration

High-throughput code synthesis & comprehensive test generation

OPTIMIZED WORKFLOWS:

End-to-end test suite creation

Rapid boilerplate generation

API contract synchronization

Monthly Token Quota (All Tiers)

GPT-4o

OpenAI

128k Tokens

Active Supported Integration

Multimodal visual reasoning & UI component synthesis

OPTIMIZED WORKFLOWS:

Figma design to production code

Visual UI layout debugging

Split preview diff review

Monthly Token Quota (All Tiers)

Claude 3.7 Sonnet

Anthropic

200k Tokens

Active Supported Integration

Hybrid reasoning & complex algorithmic state machine debugging

OPTIMIZED WORKFLOWS:

Complex concurrency bug isolation

Mathematical logic proofs

Multi-agent task orchestration

Monthly Token Quota (All Tiers)

Claude 3.5 Sonnet

Anthropic

200k Tokens

Active Supported Integration

Precision AST refactoring & whole-codebase consistency audits

OPTIMIZED WORKFLOWS:

In-place TypeScript/Rust refactors

Deterministic diff staging

ESLint/Clippy gate verification

Monthly Token Quota (All Tiers)

Claude 3.5 Haiku

Anthropic

200k Tokens

Active Supported Integration

Sub-second latency in-line completions and docstrings

OPTIMIZED WORKFLOWS:

Real-time typing autocompletion

Fast docstring generation

Quick snippet repairs

Monthly Token Quota (All Tiers)

Gemini 2.0 Flash

Google

1,000,000+ Tokens

Active Supported Integration

Ultra-low latency code search & massive semantic indexing

OPTIMIZED WORKFLOWS:

Whole-repo semantic code search

Cross-file reference lookups

Rapid multimodal inspection

Monthly Token Quota (All Tiers)

Gemini 1.5 Pro

Google

2,000,000+ Tokens

Active Supported Integration

Massive repository context ingestion & legacy codebase migration

OPTIMIZED WORKFLOWS:

Million-line monorepo comprehension

Legacy codebase modernization

Full API specification auditing

Monthly Token Quota (All Tiers)

Grok 3

xAI

128k Tokens

Supported via BYOK & Enterprise / Staged Integration

Dense mathematical computing & low-level performance kernels

OPTIMIZED WORKFLOWS:

High-speed algorithmic debugging

Numerical simulation kernels

Low-level memory optimization

Direct BYOK API Key or Enterprise Plan

NEXA-7B Fast

CodeXa Native

32k Tokens

Native Core • Included On All Tiers

100% air-gapped on-device autocomplete with zero cloud telemetry

OPTIMIZED WORKFLOWS:

Air-gapped offline coding

Classified / proprietary codebases

Zero-latency keystroke prediction

Unlimited Zero-Token Usage (No Quota Deductions)

NEXA-Code Instruct

CodeXa Native

32k Tokens

Native Core • Included On All Tiers

Local syntax repair, AST node formatting, and hunk staging

OPTIMIZED WORKFLOWS:

Offline syntax error fixes

Code formatting & AST linting

Local hunk diff creation

Unlimited Zero-Token Usage (No Quota Deductions)

Provider Engineering Specifications

Detailed technical contracts, context limits, and security guarantees.

OpenAI — GPT Engines

Direct integration with OpenAI foundation models. Powers deep architectural planning, multi-file code synthesis, and autonomous agent loops.

Available on All Paid Plans

Model Name	Context Window	Token Weighting	Primary Workload
GPT-6 Astra	256k	1.5x units	Flagship Reasoning & Planning
GPT 5.6 SOL	128k	2.0x units	High-Throughput Monorepos

KEYRING & PRIVACY

BYOK: Supported via Local OS Keychain

Data Privacy: Zero customer code retained for model training. API zero-data retention enabled.

SUPPORTED WORKFLOWS

Directed Acyclic Graph (DAG) task formulation
Cross-file refactoring runs
Autonomous tool orchestration

Anthropic — Claude

Supported integration for Anthropic Claude models. Specialized in long-context repository comprehension, nuanced code refactoring, and AST verification.

Available on All Paid Plans

Model Name	Context Window	Token Weighting	Primary Workload
Claude 3.7 Sonnet	200k	1.8x units	Nuanced Code Synthesis
Claude OPUS 5.5	200k	3.5x units	Deep Frontier Logic

KEYRING & PRIVACY

BYOK: Supported via Local OS Keychain

Data Privacy: Anthropic commercial terms: source context never used for training.

SUPPORTED WORKFLOWS

Complex algorithm redesign
Syntactic diff validation
Long-context architecture analysis

Google — Gemini

High-throughput, low-latency integration powered by Google foundation models. Delivers rapid iterative code edits and multi-million token context windows.

Available on All Paid Plans

Model Name	Context Window	Token Weighting	Primary Workload
Gemini 3.8 Flash	1M	1.2x units	Low-Latency Synthesis
Gemini 3.6 Flash	1M	1.0x units	Fast Inline Autocomplete

KEYRING & PRIVACY

BYOK: Supported via Local OS Keychain

Data Privacy: Enterprise API privacy guarantee. Data isolated within regional boundaries.

SUPPORTED WORKFLOWS

Sub-second code completion
Million-token context exploration
Instant documentation generation

xAI — Grok

Scheduled integration for xAI Grok developer API. Designed for rigorous mathematical problem solving, formal logic checks, and algorithmic verification.

Planned Integration (Pre-Beta Wave 2)

Model Name	Context Window	Token Weighting	Primary Workload
Grok 3 Reasoning	128k	TBD on launch	Mathematical & Logic Synthesis

KEYRING & PRIVACY

BYOK: Planned on public release

Data Privacy: Will adhere to xAI API developer agreement terms.

SUPPORTED WORKFLOWS

Algorithmic complexity analysis
Mathematical proof checking
Formal specification generation

CodeXa Native — NEXA Edge

100% on-device offline runtime. Executes directly on local CPU/GPU using native Ollama detection. Zero cloud latency, zero external data transit, and costs 0x tokens.

Unlimited Fair-Use on All Plans

Model Name	Context Window	Token Weighting	Primary Workload
NEXA-8B Edge	32k	0x units (Free)	Local Offline Inference
CodeXa AST Indexer	Unlimited	0x units (Free)	Sub-50ms Symbol Cache

KEYRING & PRIVACY

BYOK: Not needed (Runs 100% locally)

Data Privacy: Air-gapped operation. Zero telemetry leaves your machine.

SUPPORTED WORKFLOWS

Offline code completion
In-memory AST symbol cache
Local git diff inspection