AI Tools
More from AI tools and models: Gemma 4 and Gemini selection, local LLMs with Ollama and LM Studio, and practical API usage with Antigravity.
Trying a Local LLM Without Slowing Down Your Mac — The First 30 Minutes of Running Ollama on an SSH Linux Host from Antigravity
Using the SSH host connection added in Antigravity App v2.33.0, I put Ollama on a separate Linux machine and had the agent write a check script that runs on that host. Why you do not need to open any port, and how to tell what runs where.
Gemini 4 Carbon Doubles in Price on December 1: Estimating One Antigravity Session from Cache Ratio and Thinking Level
Antigravity v2.31.2 lets you pick Gemini 4 Carbon when signed in with Google Cloud credentials. Before the introductory price ($2 input, $10 output) ends on November 30, here is a small script that estimates what one session costs from tokens, cache ratio, and thinking level, plus where the price increase can and cannot be absorbed.
Attachments Went Up to 20MB. Here Is What I Hand the Agent, and What I Keep Back
Antigravity app v2.28.0 lets you attach any file up to 20MB. Three questions to decide what to hand over as-is, what to trim first, and what to keep out entirely.
A 54-Line Front Door for the Days I Can't Decide Between agy and gemini-api
The Gemini API CLI (gemini-api) landed in beta on October 27. Here is a small shell script that sends repo-editing requests to the Antigravity CLI (agy) and read-only requests to gemini-api, including the 14 MB input ceiling, exit codes 0-3, and timeout's 124, all checked with stubs.
When All You See Is "An Error Occurred": Suspect a 429 First, and Check in This Order
How to tell whether a generic Antigravity error is really a usage limit (HTTP 429) or a network or sign-in problem. Open the debug details, wait the right way, and skip the retry spiral. A sequence a non-engineer can follow.
Antigravity, AI Studio, or Stitch First? Pick by the Shape of What You're Building
A way for non-coders to decide which Google tool to open first: split Antigravity, AI Studio, and Stitch by the shape of what you want to make — a screen, a small app, or a tool for your own files.
The Terminal Stayed on Running Forever: Three Ways the Finish Signal Never Arrives, Checked Before Raising the Timeout
When a command you handed to an agent sits on Running, it may not be slow at all; the finish signal may simply never arrive. I measured three shapes in a plain shell: waiting on stdin, a child process holding the output pipe, and waiting on a prompt. One run was cut off at 3.0 seconds, another held for 30.0, and I note where the fixes break.
Does Splitting Work Across AI Tools Really Save Money? Put the Monthly Gap and the Switching Cost in One Table
Is it cheaper to use different AI tools for different tasks? A small Python script that puts per-task cost and switching effort side by side so you can find the break-even point.
A 50-Line Wrapper That Hands Japanese Proofreading to agy's Headless Mode
Wrap the Antigravity CLI (agy) headless mode in a small proofreading script: how to feed stdin, why exit code 0 can still mean an empty answer, the argument-length ceiling, and a cost estimate that accounts for the January 2027 price change.
A report of a cache cleanup that wiped a drive: why I now route every delete through a checkpoint
After reading about a cache cleanup that wiped a drive, I built a small Python checkpoint that narrows how far a delete can reach. Here is v2, which closes four holes I found in November (a root taken from the current directory, nested .git, half-finished --apply, and inflated plan sizes), with real outputs, a symptom table, and where I draw the approval line.
You Don't Need to Read Commands: Three Places to Check When an Agent Works
If strings of terminal text make the approve button feel risky, here is a way to check just three things: a way back before you start, the first word of a command, and the diff afterward.
The Day I Stopped Splitting by Skill — Three Questions for Routing Work Between Antigravity CLI and Claude Code
An unattended job sat waiting on an approval prompt for three days without a single error line. Here is how I stopped dividing work between two agent CLIs by capability, and started dividing it by where the approval boundary falls.
The October 16 date I could not find — taking stock of Gemini model shutdown dates yourself
A widely repeated shutdown date for the Gemini 2.5 family is not on the official deprecations page. Here is how I pull every model ID out of a repository and reconcile it against a ledger I maintain myself, with working scripts.
Starting with AGY_CLI_HIDE_LOGO: tuning Antigravity CLI for narrow terminals, screen readers, and recordings
Logo art, your email address, output that swallows the pane, a copy you never asked for. Here are the Antigravity CLI environment variables and settings that quiet the display, grouped by the three situations where they matter.
Three things I had to fix before my status line could tell me what a session cost
Getting real session cost into a custom Antigravity CLI status line took three fixes: how the script reads stdin, how to tell whether your build sends a cost field, and how to keep a usage ledger that does not double-count.
When Antigravity Swaps Its Default Model, Only the Jobs You Narrowed First Survive
Gemini 3.7 Flash is now the default model for Antigravity agents. The places you never configured are exactly the places that shift silently. Here is how to inventory your default-model exposure, then split your jobs into move-now and hold, scored by how much output freedom you left open.
When a Fast Model Feels Slow, Look at Reasoning Effort Before Switching Models
How I choose between Low, Medium, and High reasoning effort in Antigravity by the shape of the task, how that compares with switching models or rewriting the request, and a triage table for when changing the level makes no difference. The logging function is now fixed to run under zsh, and I re-tested it before publishing.
Half My Tasks Went to Pro — and So Did Only 61% of the Tokens
A singular model setting became a models collection, which means routing across models is now something you define yourself. Here is how I re-measured a task-type routing rule against the actual context-size distribution of 67 tasks in my own repository.
Routing /effort by Task Class in Antigravity CLI: Six Weeks of Measurements
I built a small router that picks an /effort level from the shape of the task, then aggregated six weeks of run logs. Here is where raising effort helped, where it actively hurt, and what mattered more than effort.
Routing Between Local Gemma and Cloud Gemini 3.5 Flash by How Easily You Can Verify the Output
When I split local and cloud work by whether it was sensitive, or by which was faster, my decisions wobbled every time. The axis that finally held was different: if the output is wrong, can I catch it cheaply and undo it cheaply? Here is a router that chooses a model from verifiability and recovery cost, with working code and a measurement ledger.
Is the $100 AI Ultra Tier Worth It Solo? Measure the Break-Even from Limits and Parallelism
Whether the $100/month AI Ultra tier (5x the Pro limit) is worth it for an indie developer, framed as a break-even from how often you hit the cap and the effective throughput of parallel agents, with a calculator script.
Measure Before You Trim: A Context Ledger for Antigravity CLI Token and Latency Costs
Prompted by the ~70% token reduction reported for the Android CLI agent, I built a thin wrapper and a weekly review to measure my own agent runs. Here is how I replaced whole-file context with line ranges and cut wait times.
When an Overnight Local Agent Crawls by Dawn — Keeping Ollama's Latency Flat by Working Backward from Context Length
Why each step of a long-running local agent gets heavier toward the end, how to measure it from Ollama's timing fields, and how a fixed num_ctx plus a rolling summary keep per-step latency flat.
A Schedule That Survives 429s: Backoff and Jitter for Agent Automation
Run agents in parallel and rate-limit 429s can cascade until everything dies. Here is how to design exponential backoff and jitter so the retries themselves don't create new congestion, from an indie developer's automation setup.
Using the v2.1.4 Quota Screen for a Weekly Reckoning: Reading Used and Remaining to Run an Indie Budget
How to turn the used/remaining display in the reworked Antigravity v2.1.4 quota screen into a weekly reckoning instead of a gut feeling. Baseline recording, burn-rate math, and allocation across multiple projects, written as an indie-dev operating routine.
Your Antigravity LLM App Drifts on Cost and Quality While the Dashboard Stays Green — Instrumentation Field Notes
Watching only total cost and latency hides the slow drifts that hurt. These are field notes on attributing telemetry by feature, tenant, and prompt version so you catch quality regressions and cost spikes early.
LM Studio Not Starting or Model Won't Load? Fixes for Mac & Windows (2026)
Fix LM Studio when it will not start, a model will not load, or the local API stays silent: find the stuck stage first, then check requirements, memory estimates, port conflicts and logs on Mac and Windows.
Fix PyTorch CUDA Errors: torch.cuda.is_available() False & Version Mismatches (2026)
Struggling with PyTorch or CUDA installation errors? This guide covers version mismatches, dependency conflicts, and GPU detection failures with step-by-step solutions.
Pairing a Local LLM With Antigravity to Keep Sensitive Code Off the Cloud
Should you really let a cloud agent read code that holds your billing keys and revenue logic? For indie developers that worry is concrete. Here I pair Ollama and Gemma as a local LLM with Antigravity, routing sensitive parts to local and general parts to the cloud, with the decision rules and measurements.
Routing thinking_level by Task in Gemini 3.5 Flash — Re-measuring My Token Spend
After Gemini 3.5 Flash became Antigravity's default, my thinking tokens crept up quietly. Here's how I measured thinking_level per task type and landed on a setup that cuts spend without losing accuracy.
Parallel Agents Quietly Burn Through Your Quota — A Self-Defense Circuit Breaker When Limits Are Invisible
Even on AI Ultra's high ceiling, running parallel agents can exhaust your allowance without warning and leave later runs half-failing. Assuming the limit is invisible from outside, here is a circuit breaker that records consumption on your side and applies the brakes, drawn from real operation.
How Far Can Android Studio's Migration Agent Take iOS Code Into Kotlin? A Hands-On Evaluation
I fed one screen of my ukiyo-e wallpaper app to Android Studio's migration agent. What the generated Kotlin got right, the 12 fixes it needed, and the review workflow I now run before accepting any of it.
Cutting Down 'Plausible but Wrong' RAG Answers — A Retrieval Evaluation Harness for Gemma 4 and Antigravity
Replace gut feeling with recall@5, MRR and faithfulness scores — a 30-question golden dataset and a small Python harness for evaluating a local Gemma 4 RAG stack.
Chrome DevTools for agents 1.0 goes stable — what changes when it ships inside Antigravity 2.0
Google has shipped Chrome DevTools for agents 1.0 as a stable release and bundled it into Antigravity 2.0. Here is what becomes possible, how it differs from Playwright MCP and Claude in Chrome, and what to watch out for on day one.
Three Months Using Antigravity as a Creative Assistant: An Artist's Honest Review
An indie creator who develops apps while maintaining an international art practice shares an honest, three-month account of using Antigravity for the work that surrounds creation. What can you delegate? What must stay in your own hands? Here's the dividing line I found.
Gemma 4 on Antigravity: Picking Q4 vs Q5 — What I Found After a Week on M2 Mac
A hands-on comparison of Gemma 4 quantization variants (Q4_K_M / Q5_K_M / Q8_0 / fp16) running locally with Antigravity on a 16GB M2 Mac, measured across speed, memory, and output quality.
3 Weeks with Gemini 2.5 Flash as My Default in Antigravity — Speed, Accuracy, and Cost in Practice
What actually changes when you switch from Gemini 2.5 Pro to Flash as your default model in Antigravity? Three weeks of real usage data, including where Flash falls short and when Pro still earns its place.
VSCode 1.118: The Quiet Upgrade That Matters for AI-Assisted Development
VSCode 1.118's pay-as-you-go prep, prompt caching, and context compression, revisited after a month-plus on Copilot's consumption billing: how to measure your tokens, a break-even comparison, and settings that keep your bill from creeping up.
Antigravity vs Cursor vs Bolt for Monetization Projects — 2026 Comparison
A practical comparison of Antigravity, Cursor, and Bolt from a revenue generation perspective. Which AI development tool should indie developers and freelancers choose for projects designed to make money?
Antigravity vs Cursor vs Bolt vs Lovable — An Indie App Developer's Honest Cost and Quality Comparison
A six-month real-world comparison of Google Antigravity, Cursor, Bolt, and Lovable for indie app development. Actual monthly costs, code quality for App Store submissions, implementation speed, and a project-type framework for choosing the right tool.