Weekly GitHub Stars Ranking: ComfyUI Adds 4,342 Stars as the Top 10 Holds Steady
116 tracked AI repositories added a net 23,181 stars in a week, up 0.43% and about 20% faster than last week. ComfyUI alone gained 4,342 stars, 18.7% of the pool's growth, while Claude Code and n8n slowed. The top ten is unchanged for a second week and the entry threshold keeps rising.
💡 What You Will Learn
116 tracked AI repositories added a net 23,181 stars in a week, up 0.43% and about 20% faster than last week. ComfyUI alone gained 4,342 stars, 18.7% of the pool's growth, while Claude Code and n8n sl
📜 Table of Contents
First, the methodology, otherwise the numbers below are easy to misread.
Data methodology
The data comes from the official GitHub API. This snapshot was taken on October 25, 2026 (UTC) and compared with a snapshot from seven days earlier under the same methodology. The tracking pool holds 116 LLM and AI Agent repositories; 110 returned star counts normally, and 6 returned empty values due to renames or migrations and are excluded from gain and loss calculations. Weekly growth is the current star count minus the count seven days ago, with no smoothing and no holiday adjustment.
Over the week the pool gained a net 23,181 stars, rising from about 5.41 million to 5.43 million, a gain of 0.43%. The same measure last week was 18,964 stars and 0.35%, so this week ran about 20% faster.
This week's TOP10
| Rank | Project | Stars | Weekly gain |
|---|---|---|---|
| 1 | n8n | 208,587 | +663 |
| 2 | Ollama | 184,752 | +1,005 |
| 3 | Awesome Prompts | 174,545 | +1,143 |
| 4 | Transformers | 168,157 | +334 |
| 5 | Dify | 159,524 | +714 |
| 6 | Open WebUI | 156,748 | +671 |
| 7 | Langflow | 156,018 | +213 |
| 8 | Claude Code | 153,352 | +1,100 |
| 9 | LangChain | 148,445 | +405 |
| 10 | ComfyUI | 142,925 | +4,342 |
Nearly a fifth of all growth came from one image tool
ComfyUI added 4,342 stars in a week, reaching 142,925, a gain of 3.13%. That is three times last week's 1,447 and accounts for 18.7% of the pool's net growth; in the final 24 hours before the snapshot it alone picked up 325 stars, the largest single-day figure of any repository. Workflow tools on the image generation side are clearly drawing attention faster than other segments right now.
vLLM came second with 1,409 stars, up 1.50%, at 95,485. It was not in the top ten for growth last week and jumped straight to second, suggesting renewed interest in the inference serving layer.
Behind them: Awesome Prompts with 1,143, Codex CLI with 1,113, Claude Code with 1,100 and Ollama with 1,005. The top ten for growth together account for 12,924 stars, 55.8% of the pool's net gain; the other 106 repositories split the rest, with weekly gains mostly in the single digits to tens of stars.
Two headline projects downshift at once
Claude Code went from 1,850 stars last week to 1,100, a drop of 750; n8n fell from 794 to 663 and slipped out of the growth top ten to eleventh; Anthropic Quickstarts went from 699 to 369. Three different directions slowing in the same week looks more like a natural cooling after a burst of recommendations, and one week is not enough to conclude anything, so next week is worth watching.
No seat changes for two weeks, but the middle of the list moved
The top ten is identical in membership and order to last week, the second straight week without a seat change. Fourth and fifth are tightening: Transformers added only 334 stars while Dify added 714, narrowing their gap from 9,013 to 8,633. At the current pace a crossover is more than twenty weeks away, so no swap is imminent, though the direction is consistent.
The movement was in the middle of the list: OpenHands rose from 19 to 18 and MCP Servers fell from 18 to 19; Langfuse gained a seat to 52 while Continue slipped to 53; Kestra and FastGPT swapped 58 and 59; Cover Agent and MCP Playwright traded 102 and 103. The gaps at those positions are within a few hundred stars, so a single round of curation can flip them, which limits how much they tell us.
No new faces at the top
No new projects entered the top ten this week. The closest challengers are eleventh-ranked llama.cpp at 132,348 stars, up 716, and twelfth-ranked Codex CLI at 130,873, up 1,113. To enter the top ten they must pass ComfyUI at 142,925, which leaves llama.cpp 10,577 stars behind and Codex 12,052. The problem is that ComfyUI outgrew llama.cpp by 3,626 stars this week, so the threshold itself is rising and the gap should keep widening at the current pace.
Further down, some projects are building momentum. Langfuse added 463 stars, up 1.29%, third among repositories above 20,000 stars; LangGraph added 424, up 0.98%. One handles agent observability and evals, the other multi-step orchestration, both tools you reach for after an agent is already running. Their growth outpacing the model layer suggests attention is moving toward the later stages of deployment.
The decliners are the usual names
Eleven repositories lost stars this week, for a combined loss of 62. GPT Engineer fell the most at 20; Grok-1 lost 11, TextGen WebUI 9, ChatGLM 6, LocalStack 5, and six others lost between 1 and 3 each. That scale looks like routine unstarring rather than users leaving.
The slower growers matter more. Langflow added 213 stars, up 0.14%; Transformers added 334, up 0.20%. Both are heavily used, but infrastructure-layer repositories are long past the stage of being repeatedly recommended and mass-starred, and new attention has moved to applications and supporting tools.
How to read this list
Stars measure attention only, not quality or performance. This week's numbers point in a fairly clear direction: image workflow tools are gathering stars quickly, the inference serving layer is warming up, coding agents are steady, and model repositories plus early chat interfaces keep slowing. When choosing a stack, treat total stars as a proxy for ecosystem maturity and weekly growth as a signal of where attention is going, look at both, and use neither as an acceptance test. Anything you actually deploy still depends on issue response times, release cadence, and a run in your own environment.
Same time next week
The ranking updates every Monday with the same source and methodology, so week-over-week comparison stays clean. If you want to keep this table for later, bookmark this page; subscribe to site updates and next week's growth table and rank changes will reach you first. If a project you use moved unusually this week, leave a note in the comments and we will pull it out for a closer look next time.
Written by our editorial team; tools listed here are tested or verified against public sources. Links point to official sites or GitHub repos for reference only โ no paid placements.
