Indras-Mirror/Xolotl

Xolotl (formerly Strata-DSV4.1): DeepSeek-V4.1-Flash, all 552B, on one gaming PC (RTX 4090 + 90 GB RAM) at ~27 tok/s, plus GLM-5.3-Flash, Qwen3.8-Flash-Next and Qwen3.8-27B. A heavily extended fork of Strata: MoE experts tiered across VRAM, pinned RAM and NVMe; OpenAI + Anthropic API.

#LLM inference

Visit GitHub

C++

361

266 contributors

MIT License

website

FeaturedAdvertise to 180K+ monthly visitors

Bifrost

Enterprise AI Gateway Built For Scale: Route, Govern, and Secure All Your AI Traffic

Momentum

How fast Indras-Mirror/Xolotl is gaining stars, and where it lands on Trendshift and GitHub Trending

Activity and trending

Stars, forks and merged pull requests each day or month, lined up with the repository's rank on Trendshift Trending

Oct 16 – Oct 26, 2026

Stars +273 in 11 days

Forks +42 in 11 days

Merged PRs 33 in 11 days

Trendshift Trendingwhere the repository ranked each day it made the top 25

Top 25 in C++

915

Oct 16Oct 18Oct 20Oct 22Oct 24Oct 26

Stars, forks, merged pull requests and Trendshift Trending rank, by day
Day	Stars gained	Forks gained	Merged PRs	C++
Fri, Oct 16, 2026	+0	+1	0	Not trending
Sat, Oct 17, 2026	+0	+5	0	Not trending
Sun, Oct 18, 2026	+13	+6	0	Not trending
Mon, Oct 19, 2026	+81	+7	5	#9
Tue, Oct 20, 2026	+55	+6	1	#15
Wed, Oct 21, 2026	+36	+8	9	Not trending
Thu, Oct 22, 2026	+18	+3	3	Not trending
Fri, Oct 23, 2026	+19	+1	1	Not trending
Sat, Oct 24, 2026	+28	+2	7	Not trending
Sun, Oct 25, 2026	+18	+2	5	Not trending
Mon, Oct 26, 2026	+5	+1	2	Not trending
Total	+273	+42	33

#1 on the list
Rank that day
Not trending

Achievements

Trendshift snapshots GitHub's daily Trending list and runs its own trending list

Trendshift Trending

Indras-Mirror/Xolotl has trended on Trendshift's daily and weekly list

Daily

First achieved on 19th, Oct 2026 for C++

lxe/dsx

Rust

DeepSeek-V4.1-Flash-1017 (552B MoE) on 2x RTX 3090 + 128 GB RAM: Rust + CUDA engine and 2-bit quant. 30-37 tok/s decode.

7013

tpurtell/ds41rt

Rust

Native DeepSeek V4.1 Flash inference across one RTX PRO 6000 and four DGX Spark expert workers

375

antirez/ds4

DeepSeek 4 Flash and PRO local inference engine for Metal, CUDA and ROCm

24.1k2.4k

Entrpi/ds4

CUDA Blackwell performance fork of antirez/ds4 (DeepSeek-V4-Flash): ~4x prefill, ~1.5x decode, continuous batching, weight server, DSpark

7617

eddoursul/StrataFast

C++

A fork of https://github.com/Niko1221/Strata optimized for my 3090+5070 Ti setup, see docs/COMPARISON.md. Renamed to StrataFast.

535

Anemll/ds4-ssd

DeepSeek V4 Flash specific inference engine. SSD MoE expert paging (slot-bank) + disk KV cache for long agent sessions. Metal-first, narrow & high-performance.

336

Niko1221/Strata

C++

Qwen3.8-Flash-Next on any consumer hardware: one-click install for Windows / Linux. Strata inference engine, OpenAI/Anthropic API on localhost, optional image input.

33.2k3.6k

shawnshekari/strixite

C++

LLM inference engine written from scratch in HIP for AMD Strix Halo (gfx1151) and Qwen3.8-Flash-Next - ~50 tokens/s out to 492k context

839

iamwavecut/ninfer-all

C++

One line of NInfer for the RTX 3090, RTX 4090 and RTX 5090, consolidated from the forks that carry it: GGUF block formats, ternary Bonsai 2, device route profiles, MTP and DFlash2.

15632