Alibaba Cloud
Model Studio and Qwen for long-context autonomous engineering tasks, with a native Forge provider and BYOK control.
Explore Alibaba Cloud →One Forge workflow. Your preferred AI and cloud underneath.
Forge keeps provider complexity out of the developer workflow while supporting direct, bring-your-own-key integrations with leading AI infrastructure.
FEATURED
Featured integrations
Launch integrations are technically connected through OpenAI-compatible provider adapters and remain customer-controlled.
Qwen
Use qwen3.8-max as a first-class Forge model while retaining Forge planning, build, test, repair and verification.
Explore Qwen →NVIDIA
Connect NVIDIA NIM as a Forge model provider and use NVIDIA-hosted or approved self-hosted NIM endpoints.
Explore NVIDIA →NVIDIA NIM
OpenAI-compatible NIM inference with a coding-focused default model and Forge's verification workflow above it.
Explore NIM →DEVELOPER WORKFLOW
Editors, CI, SDKs and cloud runtimes
Every integration uses the same authenticated Forge run/status/cancel contract.
VS Code + JetBrains
Run Forge from the editor while keeping credentials in the editor's secure storage.
Download editor plugins →GitHub + GitLab + Generic CI
Use Forge in repository pipelines with secret-managed API keys and idempotent runs.
Download CI adapters →Node.js + Python SDKs
Call Forge directly with retry, polling and cancellation helpers.
Download SDKs →AWS · Azure · Google Cloud
Cloud adapters also cover Alibaba Cloud and NVIDIA workloads through the same Forge API.
Download cloud adapters →Automatic for most users. Explicit when enterprises need control.
Forge can route tasks automatically, while enterprise teams can pin a provider, region, endpoint or model when policy requires it.
- Bring your own provider credentials
- Provider keys stored encrypted when configured
- Per-workspace permissions and audit trail
- Custom OpenAI-compatible endpoints for private infrastructure
Built for measurable adoption.
Each provider integration is designed to support attributable calls, model usage and verified developer outcomes as the partnership program expands.
FORGE CORE 1.0
Autonomous engineering with a verification moat
Forge Core 1.0 combines repository IR, dependency/change-impact reasoning, atomic safe patching, risk-based specialist verification, tamper-evident proof receipts and privacy-safe learning. Forge 0.2.2 remains the automatic fallback for internal Core failures.