Kimi K3 DSpark Upgrade: 1M Context Without Performance Degradation, 140k Downloads
BanghuaZ · x · 2026-08-01
Addressing feedback on performance degradation for long contexts, the team released an upgraded DSpark checkpoint for the Kimi K3 model.
Key Upgrades & Benchmarks:
- Long Context: Achieved an accept length of 4.2 in the RULER v2 benchmark at 1M context length.
- Agentic Use Cases: Reached an accept length of 4.9 in the Inference AgenticX benchmark by SemiAnalysis.
- Overall Suite: Mean accept length of 4.55 across the benchmark suite, with improvements on most benchmarks (noting a minor regression on GSM8K).
Additionally, DSpark crossed 140,000 downloads within the first few days of its launch.
Related event: Kimi K3 Upgrades Long-Context and Agent Capabilities(2 posts)→
More from coding & agent
- Connected steps are not a workflow: nobody designs the handoff — usehive · 2026-10-16
- Making Blender animations with Opus, Astra and Argon: how you prompt still matters — chongdashu · 2026-10-16
- Running PolyStrata + Cline coding agent on a $0-fancy laptop at ~90 tok/s — rpwoerk · 2026-10-16
- How do you attribute agent gains: model training or infrastructure? — jrxna · 2026-10-16
- Feed Claude your books, blogs and top engineers' code for claude.md — 'it now writes better code than me' — blaizedsouza · 2026-10-16
- Open-source /label plugin auto-names and colors Claude Code sessions in one step — hermanschutte · 2026-10-16