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.
More from coding & agent
- Month of AI Bugs Returns: Over 20 AI System Vulnerabilities to Be Disclosed — wunderwuzzi23 · 2026-08-01
- Meta Engineer Shares MLSys Keynote: Using AI to Liberate Systems Researchers — salykova_ · 2026-08-01
- React Aria launches TokenField component for building AI prompt inputs — pacocoursey · 2026-08-01
- Supabase Launches Evals to Benchmark AI Coding Agents on Real Tasks — tristanbob · 2026-08-01
- Building Software While Sleeping: 3 Lessons from Managing Autonomous AI Employees — leebase65 · 2026-08-01
- Kanbots: Run 11 AI Coding Agents in Parallel on One Kanban Board — tom_doerr · 2026-08-01