Sign in

Kilo

@kilocode.ai
198 followers 48 following 756 posts

Kilo is an open-source all-in-one agentic platform. 3M+ Kilo Coders. 500+ models. No markup.

PostsRepliesMedia
Kilo @kilocode.ai · 28/07/2026
Once a good plan exists, the model that builds it moves the cost far more than the result. blog.kilo.ai/p/auto-model...
blog.kilo.ai
Auto Model vs Picking Your Own: We Tested Kilo Code's Router on a Backend Build
The pace of new model releases has made “which model should I use for this?” a real problem.
010
Kilo @kilocode.ai · 28/07/2026
All three built the same working service, but Kilo's Auto Model built it for less than half the cost: - Auto Model (frontier plan → efficient build): $1.26 - GPT-5.6 Sol everywhere: $2.90 - Claude Sonnet 5 everywhere: $2.47
111
Kilo @kilocode.ai · 28/07/2026
We keep arriving at the same conclusion in these experiments: you don't need the most expensive model for every step of your workflow. So we tested whether Kilo Code's Auto Model router could act on that better than picking models by hand.
511
Kilo @kilocode.ai · 23/07/2026
More ways to control your AI coding spend are live: 📊 Cost Insights: what's driving your cost, broken down by product & user 🔔 Spend Alerts: get notified about unusual usage spikes 💡 Cost Suggestions: inference suggestions based on usage Read more: blog.kilo.ai/more-ways-to-control-ai-coding-spend
010
Kilo @kilocode.ai · 21/07/2026
Everyone's waiting for a second "DeepSeek moment." We don't think it's coming. Open models like Kimi K3 top the intelligence charts, but serving them made throughput fall off a cliff. The scarce thing was never the model, but the compute to run it. Read more: blog.kilo.ai/p/no-second-...
000
Kilo @kilocode.ai · 10/07/2026
Kilo Code is now available for JetBrains IDEs as a native Kotlin/Swing plugin built on the IntelliJ Platform. It includes chat, slash commands, file mentions, MCP servers, and model selection. Available now on the JetBrains Marketplace. blog.kilo.ai/p/kilo-code-...
blog.kilo.ai
Kilo Code Goes Native on JetBrains
A ground-up rebuild in Kotlin brings first-class AI coding assistance to IntelliJ, WebStorm, PyCharm, and every JetBrains IDE
000
Kilo @kilocode.ai · 08/07/2026
Build for model freedom, not vendor dependency. That's the main takeaway underneath the noise this week.
000
Kilo @kilocode.ai · 08/07/2026
Which is exactly why it matters more than ever that your work isn't tied to a single model vendor. The routing, the cost optimization, the data and workflows your team built, none of that should disappear because one company or government changed a policy.
110
Kilo @kilocode.ai · 08/07/2026
Strip away the drama and you'll see a clear pattern: enterprises are starting to realize that you can't blindly place all your bets on a single frontier model. Terms, access and pricing keep moving. Days later, Mistral's CEO made the same case, don't let closed AI providers control your data.
110
Kilo @kilocode.ai · 08/07/2026
The AI race isn't heating up. It's on fire. Pricing changes. Models getting pulled out from under your feet with no warning. And now Palantir's CEO went on CNBC to call the whole thing "effing insane."
100
Kilo @kilocode.ai · 30/06/2026
Auto Efficient is a next-generation model router inside Kilo. It's session-aware and informed by real benchmark data. On average, it's 77% cheaper than Claude Opus 4.8 while retaining nearly 70% performance parity. Check out the numbers here: kilo.ai/auto-efficie...
011
Kilo @kilocode.ai · 24/06/2026
10 months ago we predicted AI coding bills would hit $100k/dev/yr. This week Ramp confirmed ~$90k. The fix isn't capping usage or downgrading everyone. It's routing each task to the model that fits it. Kilo's Auto Model does it by default. kilo.codes/0c9ftEx
kilo.codes
Auto Model - Kilo Chooses the Right AI Model for Each Task
Auto Model routes coding tasks to the right AI model based on complexity, speed, and cost, reducing manual model switching and helping Kilo Gateway credits go further.
010
Kilo @kilocode.ai · 23/06/2026
Stop paying frontier prices to rename a variable! Auto Efficient routes each request to the cheapest model that can handle it, picked on a public benchmark you can check. Easy tasks run lean, hard ones stay reliable. Live now: kilo.codes/0c9ftEx
010
Kilo @kilocode.ai · 22/06/2026
Augment is sunsetting its JetBrains IDE extensions, w/ ~a month of notice. The alternatives: a CLI or an enterprise platform. Kilo's plugin already covers what PyCharm and IntelliJ devs want: native v7, Apache-2.0, 500+ models, no $100 floor. blog.kilo.ai/p/is-augment-sunsetting-its-ide-extensions
Promotional graphic comparing requested features to existing Kilo capabilities. On a black background, the Kilo logo appears above the headline: “What Augment users asked for. vs. What Kilo already ships.” A checklist on the right highlights five features: inline autocomplete and Next Edit mode, an agent panel inside the IDE, a checkpoint tracker for review and rollback, a native v7 build with no Node detection, and Apache-2.0 licensing with access to 500+ models and no $100 minimum spend. Footer text notes compatibility with PyCharm, IntelliJ, GoLand, and all JetBrains IDEs.
010
Kilo @kilocode.ai · 19/06/2026
These aren't public Terminal Bench 2.0 scores from optimized scaffolds. They're how each model runs inside Kilo's harness, with its tool pipeline and retry logic. Coverage is still growing, so some models show no data yet. Write-up: blog.kilo.ai/p/terminal-bench-scores-are-now-in
blog.kilo.ai
Terminal Bench Scores Are Now in Your Editor
Real benchmark data where you actually make model decisions
000
Kilo @kilocode.ai · 19/06/2026
Why both numbers: a model that needs three attempts to pass a task effectively costs 3x its sticker price. Score without cost is half the picture. The picker now shows value-per-dollar in the same panel where you already check price and context limits.
100
Kilo @kilocode.ai · 19/06/2026
Kilo now shows benchmark data right in the model picker, CLI and VS Code. Each model lists its Terminal Bench completion score and average cost per attempt, measured in Kilo's own harness. GPT-5.5 completes 74.1% of tasks, Kimi K2.6 hits 54.4%. Full table: kilo.ai/leaderboard
kilo.ai
Kilo - Best AI Coding Models 2026 | Live AI Leaderboard
Compare the best AI coding models by real Kilo usage, industry benchmarks, pricing, speed, and context window. See live rankings for coding and agent workflows.
100
Kilo @kilocode.ai · 18/06/2026
We ran GLM-5.2 and Kimi K2.7 Code through the same test: plan a feature flag service, then build it. GLM's plan scored 9.0 to Kimi's 8.1. But once both built from GLM's plan, the services were near identical. The planner matters more than the builder now. blog.kilo.ai/p/glm-52-vs-...
blog.kilo.ai
GLM-5.2 vs Kimi K2.7 Code: Which Model Is Better at Planning vs Building?
We tested both models on the same backend task and found the biggest difference was not in writing code, but in deciding what code should be written.
000
Kilo @kilocode.ai · 17/06/2026
Inceptron is live in the Kilo Gateway. Sovereign, EU-hosted inference without giving up model choice. Kimi K2.6, GLM 5.1, MiniMax M2.5. GDPR and ISO 27001 compliant. From $0.15/1M input tokens on MiniMax M2.5. blog.kilo.ai/p/kilo-partn...
blog.kilo.ai
Kilo Partners with Inceptron for High-Performance EU Inference
Access fast and secure open-weight models in the Kilo Gateway
010
Kilo @kilocode.ai · 17/06/2026
$15 on Opus. $1.70 on the same task with a cheaper model. Most of what an agent does all day doesn't need a frontier model, you're just paying like it does. Here's how to cut your AI bill: blog.kilo.ai/p/4-spend-le...
000
Kilo @kilocode.ai · 16/06/2026
SpaceX is buying Cursor for $60 billion. SpaceX has the compute, Cursor has the distribution into half the Fortune 500, and the base models are converging. Once a tool gets acquired, its model choices serve the acquirer, not you. blog.kilo.ai/p/spacex-jus...
blog.kilo.ai
SpaceX Just Bought Cursor for $60 Billion. Why the Deal Matters.
When a rocket company needs an AI coding tool badly enough to spend $60B, the strategic center has moved from model quality to compute access.
020
Kilo @kilocode.ai · 16/06/2026
Fable 5 and Mythos 5 got pulled 3 days after launch over a US export directive. The frontier didn't go with them. GPT-5.5 tops KiloBench, Nemotron 3 Ultra is free and self-hostable, MiniMax M3 runs ~1/40th Fable's cost. The frontier is wider than one lab. blog.kilo.ai/p/you-dont-h...
blog.kilo.ai
You Don't Have to Use Fable and Mythos to Work on the Frontier
In a complex regulatory environment, model freedom ensures that your workflows don't stop
020
Kilo @kilocode.ai · 15/06/2026
Don't take our word for it. We put Fable 5 head to head with GPT-5.5 and wrote down what actually happened. blog.kilo.ai/p/claude-fab...
blog.kilo.ai
Claude Fable 5 vs GPT-5.5: better planning, similar execution
Update: We wrote this post on June 11 and published it on June 13.
000
Kilo @kilocode.ai · 15/06/2026
Better than Fable 5, better than Le Chaton Fat, and better than whatever you're switching to tomorrow. Multiple models beats one single model. Every time.
210
Kilo @kilocode.ai · 15/06/2026
The group chat will no longer be the one telling you your team lost. World Cup ClawByte, one-click install in KiloClaw, daily summaries + get yesterday's scores in your time zone, all piped to Telegram.
000
Kilo @kilocode.ai · 15/06/2026
Good question. The headline number is KiloBench, our internal eval on Terminal Bench 2.0, scoring real completion rate and cost per attempt rather than a spec sheet. We also rank models by real usage across Code, Plan, Debug, Ask, and Review. All live here: kilo.ai/leaderboard
kilo.ai
Kilo - Best AI Coding Models 2026 | Live AI Leaderboard
Compare the best AI coding models by real Kilo usage, industry benchmarks, pricing, speed, and context window. See live rankings for coding and agent workflows.
010
Kilo @kilocode.ai · 15/06/2026
Dad what was it like when we had access to Fable?
100
Kilo @kilocode.ai · 13/06/2026
✨ Code Reviews, REVIEWS.md and Memory, so the agent learns your standards: blog.kilo.ai/p/code-revie...
100
Kilo @kilocode.ai · 13/06/2026
✨ Kilo Console, a browser-based UI for the Kilo CLI (beta): blog.kilo.ai/p/kilo-conso...
100
Kilo @kilocode.ai · 13/06/2026
✨ Coding Plans, MiniMax M3 bought with the balance you already have: blog.kilo.ai/p/coding-pla...
100
Kilo @kilocode.ai · 13/06/2026
🦞 KiloClaw, a hosted OpenClaw that sets itself up and runs your morning: blog.kilo.ai/p/kilo-claw-...
100
Kilo @kilocode.ai · 13/06/2026
✨ Agent Manager, run multiple agents in parallel, each in its own isolated git worktree: blog.kilo.ai/p/agent-mana...
100
Kilo @kilocode.ai · 13/06/2026
ICYMI: Product Week shipped five things this week. All of them, in one place 🧵
Kilo Product Week recap graphic, headed "That's a wrap." Headline reads "Five for five," with the note "Five days. Five ships. One week. Everything shipped." A checklist on the right shows all five launches marked shipped: Monday Agent Manager, Tuesday KiloClaw, Wednesday Coding Plans, Thursday Kilo Console, and Friday Code Reviews. Black background, yellow accent, monospace type.
200
Kilo @kilocode.ai · 13/06/2026
And locally, Code mode now reviews your uncommitted changes before you push. All three are live: blog.kilo.ai/p/code-revie...
000
Kilo @kilocode.ai · 13/06/2026
The better part? You don't have to write those standards down. Toggle on Code Review Memory and it watches how you respond to PRs. Dismiss line-length nitpicks, act on error-handling comments, and it proposes updating your REVIEWS.md to match. It even opens the PR.
Kilo Code Reviews graphic about the Memory feature, headed "You don't write the rules. It learns them." Headline reads "It reviews the way you already do." A three-step flow: 1) Watch, you respond to PRs, dismissing line-length nitpicks and acting on error-handling notes; 2) Learn, it analyzes your feedback and spots the pattern; 3) Propose, it opens a PR updating your REVIEWS.md to match. A caption notes the standards stay version-controlled, built from how your team actually works. Black background, yellow accent, monospace type.
100
Kilo @kilocode.ai · 13/06/2026
Most AI review tools run the same generic rubric on every repo and miss what your team actually cares about. Code Reviews now adapt: drop a REVIEWS.md in your repo and the agent enforces your conventions, not someone's default. Live now.
Kilo Product Week graphic, day 5. Headline reads "Reviews that fit your repo." Subhead: "Drop a REVIEWS.md in your project. The agent reviews by your conventions, not a generic rubric." A code card on the right shows a sample REVIEWS.md file with review standards: an error-handling section flagging unhandled promises and requiring typed error returns, and a style section that skips line-length nitpicks but enforces named exports. Black background, yellow accent, monospace type.
110
Kilo @kilocode.ai · 12/06/2026
Two new coding models and a sold-out token plan this week in Kilo. Kimi K2.7 Code from Moonshot, Claude Fable 5 topping our coding benchmarks, and the MiniMax token plans selling out fast enough to need a new batch already.
Kilo's weekly New in Kilo graphic for the week of June 8, 2026. Three new arrivals listed: 01 Kimi K2.7 Code, a remarkable new coding model from Moonshot AI. 02 MiniMax Token Plans, first batch sold out, new batch now live. 03 Claude Fable 5, top of our coding benchmarks. A yellow callout box highlights Claude Fable 5 as Top of Benchmarks for planning and large-context coding, crushing our planning and large-context coding benchmarks. Kilo logo in the top right corner, kilo.ai and @kilocode in the footer.
110
Kilo @kilocode.ai · 12/06/2026
It's beta, so feedback actually shapes it. Full walkthrough and the Discord to send notes: blog.kilo.ai/p/kilo-conso...
blog.kilo.ai
Kilo Console (Beta) is live!
Manage git worktrees, sessions and settings with browser-based UI.
010
Kilo @kilocode.ai · 12/06/2026
Open a project, see its git worktrees, launch a CLI session in any of them. Every setting shows where its value comes from, global config or the project itself, so you stop guessing which one's winning.
Screenshot of the Kilo Console running in a desktop web browser with a dark-themed interface. The center panel displays a stylized yellow Kilo logo above a prompt input containing the text “let’s improve UI.” The selected model is shown as “GPT-5.5 OpenAI,” with shortcuts for agents and commands displayed alongside. A tip below reads, “Press Escape to stop the AI mid-response.” The left sidebar lists a project called “Kilocode Landing” and numerous worktrees with whimsical names such as Radical Crabapple, Vivid Blinker, Pickle Guide, and Cosmic Orchid. The currently selected worktree is “Radical Crabapple.” On the right, a context panel shows details for the selected worktree, along with Reset and Remove buttons and a section labeled Changes. The interface emphasizes managing multiple isolated work environments, AI models, and project contexts from a single dashboard.
110
Kilo @kilocode.ai · 12/06/2026
The Kilo CLI now has a face. Kilo Console is a local, browser-based UI for managing your projects, git worktrees, sessions, and settings. No more hand-editing JSON. Now in beta.
110
Kilo @kilocode.ai · 11/06/2026
"Open doesn't just mean weights." Chris Alexiuk of NVIDIA on everything the Nemotron family opens up: the data, the recipes, the technical report. The whole point is an open science community around models.
010
Kilo @kilocode.ai · 11/06/2026
A secret got committed to Git, then "erased" with a branch rewind. The repo looked clean. We ran Grok Build 0.1 on this Terminal-Bench task. It found the orphaned commit, saved the secret, and actually scrubbed the history. 27 steps, 41 seconds, $0.09. Writeup: blog.kilo.ai/p/we-asked-g...
Black graphic with the Kilo logo. Headline reads "A secret was buried in Git. Grok dug it out in 41s." Three stat boxes show 27 steps, 41 seconds of agent time, and $0.09 total cost. Below: "Human expert estimate: about 30 minutes."Graphic titled "Step one: find what Git remembers." Headline reads "The repo looked clean. Git doesn't forget." A terminal window shows git log with two ordinary init commits, then git fsck --unreachable revealing an unreachable commit, then git show exposing a hidden commit named "feat: add scratch notes" that added secret.txt. Caption: "The secret was sitting in an orphaned commit the whole time."Graphic titled "Step two: actually erase it." Headline reads "Deleting isn't done until you verify." A five-step list: save the secret before cleaning, expire the reflog, re-check with git fsck (highlighted, noting objects were still there), run git gc --prune=now, then check again to confirm the repo is clean. Caption: "Grok caught its own incomplete cleanup mid-task, then fixed it."Graphic titled "KiloBench: the scorecard." Headline reads "Every check came back clean." Four checked items: secret recovered to /app/secret.txt, history scrubbed with fsck clean, README and commits untouched, full checksum integrity pass. Stat line: 27 steps, 41 seconds, $0.09, with a human expert estimate of about 30 minutes.
010
Kilo @kilocode.ai · 11/06/2026
"So it's just another subscription?" The opposite. One balance, 500+ models, and now coding plans too. Starting with MiniMax M3: blog.kilo.ai/p/coding-pla...
blog.kilo.ai
Coding Plans are live in Kilo
Coding Plans have landed in Kilo. The first one is MiniMax - and there's a lot more coming.
110
Kilo @kilocode.ai · 11/06/2026
Roo Code shut down. Copilot moved everyone to usage-based billing on June 1. Different decisions, same lesson: when your whole setup hangs on one vendor, their next change is your problem. Coding Plans buy external token plans on your existing balance, no BYOK config maze.
Kilo Coding Plans graphic titled "Why build on a balance." Headline reads "Your stack keeps changing without you." Two cards below: one labeled "a product decision," noting Roo Code shut down and took its workflows with it; the other labeled "a pricing decision, June 1," noting Copilot went usage-based, keeping the flat fee but dropping the unlimited ceiling. A highlighted bar at the bottom reads "Same lesson, either way: build on one vendor, and their next change is your problem." Black background, yellow accent, monospace type.
221
Kilo @kilocode.ai · 11/06/2026
MiniMax M3 benches near Claude Opus 4.8 at a tenth of the price. Coding Plans are live in Kilo. Buy them with the balance you already have, no separate subscription.
Kilo Product Week graphic, day 3. Headline reads "Opus-class. 1/10th the price." Subhead: "Coding Plans are live. The first is MiniMax M3, bought with the balance you already have." A plan card on the right lists MiniMax M3 Token Plan Plus at $20/month: ~1.7B tokens per month, 1M context window, 3–4 concurrent agents, top 10 on the Kilo leaderboard, and multimodal image and video input. Black background, yellow accent, monospace type.
220
Kilo @kilocode.ai · 10/06/2026
Switching from Copilot? The first 500 new users get $2 in credits to try any model in Kilo, across VS Code, JetBrains, and CLI. Sign up through this link to claim it. And yes, $2 goes a long way when the models are open weight. That's the whole point. kilo.codes/CK5daxL
Promotional graphic on a black background in Kilo's brand style. A grey label at the top reads "Switching from Copilot?" with the Kilo logo in the top right corner. A large headline reads "$2. Any model." with $2 in yellow. Below: "The first 500 new users get $2 in credits to try any model in Kilo." Three dark cards list the platforms: VS Code, JetBrains, and CLI. The footer shows kilo.ai in yellow and "While credits last" in grey.
000
Kilo @kilocode.ai · 10/06/2026
This is what a sustainable workflow looks like: open weight models, transparent costs, and no surprise bill at the end of the month. Run them hosted in Kilo, locally, or with your own keys.
110
Kilo @kilocode.ai · 10/06/2026
Sometimes that's the whole review.
Reddit comment by user turkert, posted 6 hours ago, with 1 upvote. The comment reads: "Kilo Code + MiMo v2.5 rocks."
100
Kilo @kilocode.ai · 10/06/2026
The use cases keep coming.
Reddit comment by user Regenfeld, posted 3 days ago, with 2 upvotes. The comment reads: "Same combination. Kilocode + Deepseek V4 Pro API (Official). I've been using it for developing several mods for old games. It works really well for me."
100
Kilo @kilocode.ai · 10/06/2026
Token efficiency matters a lot more when every token is metered.
Reddit comment by user hackmebr0, posted 3 days ago, with 2 upvotes. The comment reads: "I switched from CC + GLM to kilo + GLM and DeepSeek and it's significantly better. You have to hand hold a bit more but the token efficiency is noticeable."
100