Cerebras Tab Autocomplete Extension

VSCode tab-autocomplete extension using GPT-OSS-120B for code completion

Browser extension Rated 7 Oct 2026Dormant

Link: github.com/0xmiki/cerebras-tab-model
Built by: will to code

Bloat

65/500

Where it landed on the scale

Bloat
0

Float
300

Moat
400

500

Summary

This is a one-day experiment: a GitHub repo with a working-sounding extension, no releases, no docs, and no activity since the day it was made. The idea of using a fast general-purpose model for completions is interesting, but nothing shows a way to charge for it or keep anyone from rebuilding it.

Bloat · Not much here works or looks likely to last, yet.

Breakdown

Five measures, 100 points each. Open the receipts under any of them to see the posts and pages behind the points.

Viability

Is there a real problem, someone who would pay, and a market this builder can actually reach?

It's framed as an experiment, not a product. Nothing points to a payer, a price, or a market fit for local developers, and it depends on a third-party inference provider.

14/100

Rubric 0–20No clear problem or user. A toy, a practice clone, or a solution looking for a problem.

Receipts (2)

The creator calls it 'an interesting experiment' and talks about future possibilities, not users or pricing.
“Today's project was an interesting experiment When Copilot first launched, LLMs were relatively weak and slow. We were forced to train custom, small-parameter code completion models, and that became the industry standard. But why are we still doing that? General purpose llms have become powerful and we now have inference providers that deliver 2000+ tok/sec So, I decided to test a new approac”
Post · @willtocode · 21 Nov 2025 · open on Telegram (opens in a new tab)
The repo has no releases, no license, and no homepage, so there's no distribution or business shape yet.
Checked 7 Oct 2026 · github.com/0xmiki/cerebras-tab-model (opens in a new tab)

Moat

What stops someone copying it?

It's a thin layer over a hosted model API, and a developer could rebuild it in a weekend. The comment-as-prompt behavior is a prompt-level trick, not something hard to copy.

8/100

Rubric 0–20A thin wrapper or a clone that anyone here could rebuild in a weekend.

Receipts (2)

The repo describes itself as an extension using a general-purpose LLM hosted on Cerebras, which is a wrapper around someone else's inference.
Checked 7 Oct 2026 · github.com/0xmiki/cerebras-tab-model (opens in a new tab)
The claimed advantage is flexibility from prompting, which anyone can replicate.
“Today's project was an interesting experiment When Copilot first launched, LLMs were relatively weak and slow. We were forced to train custom, small-parameter code completion models, and that became the industry standard. But why are we still doing that? General purpose llms have become powerful and we now have inference providers that deliver 2000+ tok/sec So, I decided to test a new approac”
Post · @willtocode · 21 Nov 2025 · open on Telegram (opens in a new tab)

Momentum

Did the updates keep coming?

Work happened on a single day in November 2025 and nothing has shipped in the 320 days since, so there's no sustained cadence to speak of.

0/100

Sustained shipping0/60
Worked on across 1 days.
Consistency0/40
Updates in 1 of 1 month.
Quiet
No update for 320 days, so scaled to 25%.

Receipts (1)

Earliest post about it.
“Today's project was an interesting experiment When Copilot first launched, LLMs were relatively weak and slow. We were forced to train custom, small-parameter code completion models, and that became the industry standard. But why are we still doing that? General purpose llms have become powerful and we now have inference providers that deliver 2000+ tok/sec So, I decided to test a new approac”
Post · @willtocode · 21 Nov 2025 · open on Telegram (opens in a new tab)

Infrastructure

Does a real, working product exist?

A public repo exists and the creator says the extension works, but with no release, docs, or install path, it's closer to a prototype than a usable product.

41/100

It's live20/20
https://github.com/0xmiki/cerebras-tab-model responded when checked.
HTTPS5/5
Hosted on a platform (store, GitHub or Telegram), so HTTPS is the platform's.
Real product15/35
The post says it works and compares its speed to Copilot, and the code is public. But there's no release, install path, or docs, so it reads more like a demo than something people can use.
Own home0/10
No home found.
Maintained0/15
Last sign of shipping 2025-11-21.
Operations1/15
There's only a GitHub repo. No backend, accounts, docs, API, status page, or second platform shows up.

Receipts (4)

https://github.com/0xmiki/cerebras-tab-model responded when checked.
Checked 7 Oct 2026 · github.com/0xmiki/cerebras-tab-model (opens in a new tab)
The creator says the extension is as fast as Copilot and handles comments as instructions.
“Today's project was an interesting experiment When Copilot first launched, LLMs were relatively weak and slow. We were forced to train custom, small-parameter code completion models, and that became the industry standard. But why are we still doing that? General purpose llms have become powerful and we now have inference providers that deliver 2000+ tok/sec So, I decided to test a new approac”
Post · @willtocode · 21 Nov 2025 · open on Telegram (opens in a new tab)
Zero releases and a single commit day suggest a prototype.
Checked 7 Oct 2026 · github.com/0xmiki/cerebras-tab-model (opens in a new tab)
The repo has no homepage, no releases, and no sign of supporting infrastructure.
Checked 7 Oct 2026 · github.com/0xmiki/cerebras-tab-model (opens in a new tab)

Sustainability

Is it being set up to last?

There's no distribution channel, pricing, payments, privacy policy, or support contact, and the project has no sign of continued commitment. It also relies entirely on a third-party inference provider.

2/100

Distribution0/30
No working way to get it was found.
Payments0/15
No payment provider found.
Pricing0/10
No pricing found on the site.
Revenue0/5
No post mentions customers, payments, or revenue, and the crawler found no payment provider.
Privacy policy0/7
No privacy policy found.
Terms0/5
No terms of service found.
Support0/8
No support contact found.
Commitment2/20
The post gestures at a future direction like an auto-complete.md, but there's no roadmap kept, team, channel, or funding. The repo was last touched the day it was created.

Receipts (3)

The post is about the technical approach and never mentions users or money.
“Today's project was an interesting experiment When Copilot first launched, LLMs were relatively weak and slow. We were forced to train custom, small-parameter code completion models, and that became the industry standard. But why are we still doing that? General purpose llms have become powerful and we now have inference providers that deliver 2000+ tok/sec So, I decided to test a new approac”
Post · @willtocode · 21 Nov 2025 · open on Telegram (opens in a new tab)
'Much more to explore in the future' is intent only, with no concrete plan.
“Today's project was an interesting experiment When Copilot first launched, LLMs were relatively weak and slow. We were forced to train custom, small-parameter code completion models, and that became the industry standard. But why are we still doing that? General purpose llms have become powerful and we now have inference providers that deliver 2000+ tok/sec So, I decided to test a new approac”
Post · @willtocode · 21 Nov 2025 · open on Telegram (opens in a new tab)
Created and last pushed on the same day, 2025-11-21.
Checked 7 Oct 2026 · github.com/0xmiki/cerebras-tab-model (opens in a new tab)