apex-inference-chip

An inference chip design that runs a real LLM (Qwen2.5-0.5B) on FPGA — one transformer decoder layer in RTL, every silicon value bit-exact against a golden model. 0.56 tok/s measured, a 140× climb, full evidence trail.

SigmanticAI/apex-inference-chip

View on GitHub ↗

An inference chip design that runs a real LLM (Qwen2.5-0.5B) on FPGA — one transformer decoder layer in RTL, every silicon value bit-exact against a golden model. 0.56 tok/s measured, a 140× climb, full evidence trail.

apex-inference-chip grew -21.9% in the last 30 days, ranks 76th of 177 in Web / Edge / On-Device Runtime, last commit 95 days ago.

Star trajectory

82 days tracked
2026-08-31438 stars today2026-11-22
Momentum

-21.9% in 30d, below the Web / Edge / On-Device Runtime median of +1.8%.

Maintenance

Last commit 95 days ago, staler than the median of 6d.

How it compares in Web / Edge / On-Device Runtime

apex-inference-chip
438
stars
-21.9% 30d
commit 95 days ago
Web / Edge / On-Device Runtime median
311
stars
+1.8% 30d
commit ~6d ago
Leader · mlc-llm
23.4k
stars
+0.8% 30d
commit today

Ranked 76th of 177 by stars.

Ranked by shared tags, then keywords, then stars
  1. Run Mixtral-8x7B models in Colab or consumer desktops

    -3 (-0.1%) in 30d · Committed 957d ago · Python · MIT

    Shares Inference Optimization, Model Optimization, Web / Edge / On-Device Runtime

  2. On-device LLM Inference Powered by X-Bit Quantization

    +0 (+0.0%) in 30d · Committed 9d ago · Python · Apache-2.0

    Shares Web / Edge / On-Device Runtime, Inference Optimization, Model Optimization

  3. 2bit quants, kernels and runtime for consumer devices.

    Committed 16d ago · C++ · Other license

    Shares Web / Edge / On-Device Runtime, Inference Optimization, Model Optimization

  4. Run a 1-billion parameter LLM on a $10 board with 256MB RAM

    +6 (+0.3%) in 30d · Committed 272d ago · C · MIT

    Shares Web / Edge / On-Device Runtime, Inference Optimization, Model Optimization

  5. Run larger LLMs with longer contexts on Apple Silicon by using differentiated precision for KV cache…

    +2 (+0.6%) in 30d · Committed 549d ago · Python · Other license

    Shares Inference Optimization, Model Optimization, Web / Edge / On-Device Runtime

  6. OrbitKV is a High Performance KV Cache System for LLM Inference.

    -52 (-15%) in 30d · Committed yesterday · Rust · Apache-2.0

    Shares Inference Optimization, Model Optimization, Web / Edge / On-Device Runtime

  7. Flash weight streaming for MLX: run massive models larger than your RAM on Apple Silicon.

    +5 (+3.6%) in 30d · Committed 162d ago · Python · MIT

    Shares Inference Optimization, Web / Edge / On-Device Runtime, Model Optimization

  8. Fastest MoE/LLM inference runtime for consumer and edge Blackwell GPUs. SN74 on Gittensor.

    +5 (+5.7%) in 30d · Committed 41d ago · C++ · MIT

    Shares Inference Optimization, Web / Edge / On-Device Runtime, Model Optimization

  9. Gemma 4 26B-A4B inference in ~2 GB of RAM on any M-series MacBook

    +442 (+6.3%) in 30d · Committed 7d ago · Swift · Apache-2.0

    Shares Web / Edge / On-Device Runtime, Inference Optimization

  10. Serve large Qwen models fast on the GPUs you actually own. Qwen3.8-27B on a single 24 GB card with vLLM: 127…

    +592 (+27%) in 30d · Committed 12d ago · Python · Apache-2.0

    Shares Inference Optimization, Model Optimization

Language
Python
License
Apache License 2.0
Contributors
1
Forks
10
Watchers
0
Open issues
0
Releases
0
Created
2026-08-17
Stars this week
-35

You've seen where apex-inference-chip stands today.

Get the moves before everyone else: the 5 Web / Edge / On-Device Runtime repos climbing fastest, every Tuesday.

Free weekly digest. Double opt-in, unsubscribe anytime.

Maintain apex-inference-chip?

Claim this page to add your own call-to-action.

Claim this page