Bittensor-trained 9B model beats Qwen's flagship 9B at coding using competitively sourced anonymous compute
const_reborn · x · 2026-11-06
constreborn highlights an under-the-radar Bittensor release: a full top-to-bottom model, pre-trained on Teutonic (3) and post-trained in a competitive environment, reaching coding performance above Qwen's flagship 9B model.
Notably, the compute was competitively sourced from anonymous network participants: pre-training on Bittensor compute via Lium, with post-training sandboxes from Targon Compute. The models are small but the metrics aren't—the author says the alignment story, aiming "a laser cannon at intelligence," is the thing to watch.
Related event: 9B Coding Model Trained on Bittensor Beats Qwen Flagship on SWE-bench(2 posts)→
More from Infra
- Veteran investor: memory stocks aren't cheap even at P/E of 3 as LTAs are meaningless — bookwormengr · 2026-11-06
- Enterprise AI's future is local: tactical workloads favor Dell, HPE, Lenovo, Supermicro over frontier labs — DavidLinthicum · 2026-11-06
- Texas Governor-Elect: Approved Data Centers Can Be Completed, New Projects to Pay Taxes — GavinSBaker · 2026-11-06
- Marin's open 535B-A23B MoE finishes main pretraining on 18T tokens, checkpoints soon — dlwh · 2026-11-06
- Buoyant Open-Sources Kinference, an Inference Control Plane for vLLM on Kubernetes — vllm_project · 2026-11-06
- vLLM ranks #2 most-engaged open-source repo in GitHub Octoverse, #8 by contributor growth — vllm_project · 2026-11-06