Python · MIT
ARahim3/kaggle-tpu-lab
Frontier-class open models on a free Kaggle TPU v5e-8: GLM-5.3-Flash 320B MoE (~165 tok/s, our own JAX engine) and Qwen3.8-27B bf16 (~130 tok/s), 262k context, prefix caching. Works with Claude Code, Codex, opencode and pi.
★ 763
Stars
101 forks · 20 open issues
+2.7/day
Growth · measured
Stars counted between two of our own snapshots roughly a week apart. The repository page gives the exact span.
31.8
Momentum / 100
see the breakdown below
2mo ago
Created
last push 32d ago
Star history
How this score was built
| Component | Weight | Earned | Why |
|---|---|---|---|
| Velocity | 55 | 10.9 | +2.7/day, log-scaled against a 750/day ceiling |
| Acceleration | 15 | 0.0 | last week ran 0.38× the monthly pace |
| Freshness | 15 | 12.5 | last push 32d ago; decays to zero at 180 days |
| Scale | 15 | 8.4 | 763 stars, log-scaled against 150k — stops a 3-star repo topping the board |
| Momentum | 100 | 31.8 | sum of the above |
Detail
| Topics | anthropic-apiclaude-codecoding-agentsfree-gpuglmglm-5-3-flashjaxkagglellm-inferencellm-servinglong-contextmixture-of-expertsopenai-apipallasqwenqwen3speculative-decodingtputpu-inferencevllm |
| Matched on | LLM & Inference (llm-inference, llm-serving) |
| Short window | +19 stars over 7 days = 2.7/day |
| Long window | +186 stars over 26 days = 7.2/day |
| Lifetime average | 11.60 stars/day over 66 days |
| First seen here | 2026-10-11 |
| Last read | 2d ago |