GYRO quant of Qwen3.8-Flash-Next tested: near Q3 quality at much smaller size
PlasticRevenue4601 · reddit · 2026-10-16
A Redditor spent three days testing Agantai's GYRO low-bit quant of Qwen3.8-Flash-Next, which claims better KLD, top-1 accuracy and smaller size than ISTA-lab quants. Setup: GYRO Strata fork, GYRO-M quant, 200k context, k8v4 KV cache, xhigh thinking. Results: zero looping over three days (longest 2-3h autonomous runs), stable tool calls except Cyrillic text edits, solid context retention, and ISTA-2bit-level speed. Verdict: roughly ISTA Q3 quality at far smaller size — impressive work from a single contributor.
More from Infra
- NVIDIA ships FP4-quantized Qwen3.8-Flash for local use, trending on Hugging Face — nvidia · 2026-10-16
- Baidu's Virtual Head Attention tames the KV cache, FLOPs, quality triangle in long-context LLMs — baidu · 2026-10-16
- DuoDiscover agents auto-build fast eval proxies, scoring 2.68x TPU kernel speedup — berkeley_ai · 2026-10-16
- Google Raises 2026 Capital Spending to Up to $205 Billion — emmanuelvivier · 2026-10-16
- Google open-sources AX, a Kubernetes-style orchestrator for billions of AI agent tasks — thisdudelikesAI · 2026-10-16
- Valtz v0.1.0.11 adds FLUX.3 support, local image gen up to 4K on Mac — TgoAI · 2026-10-16