Augment Code Alternatives in 2026
Your completions are gone. Ours run on your machine.
Augment removed inline completions and Next Edit from its Indie, Standard, and Legacy plans on March 31, 2026. Their replacement, Intent, is macOS-only. Windows waitlist opened in v0.2.6, no Linux roadmap. Here is what the sunset means for developers. Bodega One runs local FIM completions on all three platforms through models you own. No subscription. No sunset date.
What you lost. What you can get.
Augment Code (Non-Enterprise)
- Inline completions and Next Edit removed March 31, 2026 for Indie, Standard, and Legacy plans
- AI Chat and Code Review continue. Augment pivoted first to Intent, then to Cosmos (public preview May 4, 2026) -- their agentic SDLC orchestration product, Max plan only
- Intent is macOS-only. Windows waitlist as of v0.2.6, no Linux roadmap
- Standard and Max are now priced per developer ($60/dev/mo, $200/dev/mo) -- team costs scale with headcount
- Monthly subscription still required for remaining features
- Enterprise plans kept everything. Individual plans lost the editing flow
Bodega One
- Local FIM completions run on your machine, permanently
- Full IDE with Monaco editor, AI chat, and autonomous agent
- One-time purchase: $79 Personal, $149 Pro. No renewal.
- BYOLLM: 10+ LLM providers. Swap models in seconds.
- Air-gap mode: 9 enforcement layers. Zero bytes leave your machine.
What Augment Code plans cost now.
Restructured March 2026. Inline completions removed March 31 for all non-Enterprise plans. Cosmos public preview launched May 4, 2026 in the Max plan.
Indie
$20/mo
Single developer. 40,000 credits/mo. Completions removed March 31, 2026.
Standard
$60/dev/mo
Per developer, up to 20 users. 130,000 credits/mo. Completions removed.
Max
$200/dev/mo
Per developer, up to 20 users. 450,000 credits/mo. Now includes Cosmos (public preview, May 4, 2026).
Source: augmentcode.com/pricing. Check for current rates.
Everything Augment dropped. And what it never offered.
Local FIM Autocomplete
Fill-in-the-Middle completions powered by models running on your hardware through Ollama. qwen2.5-coder, codellama, deepseek-coder, and more. No API calls. No rate limits.
Zero Usage Caps
Your machine, your tokens. Generate as many completions as your GPU can handle. No daily limits, no throttling, no "fair use" policies that kick in when you actually need it.
Works With Your Hardware
6 GB of VRAM gets you solid completions with qwen2.5-coder 7B. 16 GB runs the 32B model. Apple Silicon users get MLX-optimized inference. We detect your hardware and recommend the right model.
Privacy by Default
Air-gap mode enforces 9 independent layers of network isolation. Tool filtering, shell blocking, auto-updater blocking, git IPC blocking, and more. Zero bytes leave your machine. Not a promise. An architecture.
Full IDE, Not Just Completions
Monaco editor, AI chat, autonomous coding agent, 23 built-in tools. Completions are one piece. You also get a full development environment with an agent that can read, write, and verify code.
Quality Enforcement Layer
QEL catches what raw completions miss. Structural completeness, contract compliance, language-specific patterns. Every code change gets verified before it hits your workspace.
Built for completions, not bolted on.
FIM-Compatible Models
How FIM Works
FIM splits your cursor position into prefix (code before) and suffix (code after). The model generates the middle. This is how tab-complete works in tools like Augment and Cursor. The difference: Bodega One runs the model locally.
Debounce and Latency
Adaptive debounce adjusts to your typing speed. On fast hardware (16+ GB VRAM, qwen2.5-coder 7B), expect under 200ms latency. No network round-trip means no variable latency spikes.
Air-Gap Architecture
9 independent enforcement layers. Not one kill switch. Nine separate systems that each block network access independently. Disable one and the other eight still hold.
See our full local LLM guide | Learn about air-gap mode Explore BYOLLM providers
Completions that land clean.
Every file Bodega One writes or completes passes through three verification levels before the change lands. Pattern and compile checks after every write. Micro-proof gates every second write. A full structural verifier at loop end. Not a linter. A verification pipeline.
Incremental Verification
Pattern and compile check after every file write.
Micro-Proof Gates
tsc / py_compile runs every second write.
Full Verification
Structural verifier post-loop. Pass threshold 80 for new files.
Replacing what
Augment Code removed.
What are the current Augment Code pricing tiers?+
As of May 2026: Indie ($20/month, 40,000 credits, up to 1 user), Standard ($60/developer/month, 130,000 credits, up to 20 users), Max ($200/developer/month, 450,000 credits, up to 20 users, includes Cosmos), Enterprise (custom). Standard and Max are billed per developer, not flat. There is no free Community plan. Augment switched from message-based to credit-based pricing in October 2025. Inline completions and Next Edit were removed for all non-Enterprise plans on March 31, 2026. Enterprise plans retain completions. See augmentcode.com/pricing for current rates.
What is Cosmos? Should I switch plans for it?+
Cosmos is Augment's agentic SDLC orchestration product, public preview launched May 4, 2026, bundled in the Max plan ($200/developer/month). It consolidates eight SDLC checkpoints (spec, prioritization, code review) into three, with agents running across your environment and Augment's cloud. Bodega One ships the agent and verification inside the local IDE for $79 once -- no cloud orchestration layer required.
Is Bodega One free?+
No. Personal is $79 and Pro is $149. Both are one-time purchases. No subscription, no monthly fees, no renewal. You buy it once and own it.
Do I need Ollama for local completions?+
Yes. Ollama runs the local models that power FIM completions. It is free and takes about 2 minutes to install. Once running, Bodega One detects it automatically and lists your installed models.
What models work best for FIM?+
qwen2.5-coder is our top recommendation. It ships in sizes from 1.5B to 32B, so it fits most hardware. codellama, deepseek-coder, codestral, starcoder2, and codegemma all support FIM as well.
Can I still use cloud LLMs?+
Yes. Bodega One supports BYOLLM with 10+ providers including OpenAI, Anthropic, Groq, Together AI, and OpenRouter. Use local for completions and cloud for heavy reasoning. Or go fully local. Your call.
Is my data actually safe?+
Air-gap mode enforces 9 independent enforcement layers. Zero bytes leave your machine. This is not a privacy policy. It is an architecture with nine separate systems that each block network access independently.
When is Bodega One available?+
Beta is live now for the first 200 users. Full launch coming later this year. Join the waitlist at bodegaone.ai to be first in line.
Keep your completions. Lose the subscription.
One-time purchase. Local FIM autocomplete. 23 built-in tools. An autonomous agent that verifies its own work. Your code stays on your machine.
Join the WaitlistWindows, macOS, Linux. Requires Ollama for local completions.