Open Weights vs Frontier: Just a 3-Point Gap but 1/3 the Cost
MicahBerkley · x · 2026-08-23
As of August 2026, the best open-weight model, Moonshot's Kimi K3, scores 60 on the Artificial Analysis Intelligence Index, trailing Claude Opus 5 (63) and Fable 5 (62) by just three points. However, K3 costs $3 in / $15 out per million tokens, one-third of Fable 5's price ($10/$50). The article argues open weights are now the truth for value, with closed models retaining a moat only in long-horizon agentic coding tasks like DeepSWE.
More from Venture
- GovAI launches Entrepreneur-in-Residence program with $150k funding — cephaloform · 2026-08-23
- GovAI Launches Entrepreneur-in-Residence: $150k Seed Funding for AI Governance Startups — Manderljung · 2026-08-23
- GovAI hires Entrepreneurs-in-Residence with salary + $150k funding — Manderljung · 2026-08-23
- AI compute differs from gold: GPU depreciation and physical limits reshape hedging — AccBalanced · 2026-08-23
- Harvard dropouts' AI trading startup Spectre Intelligence joins Y Combinator — ycombinator · 2026-08-23
- How Does the Market Value an AI Harness? ARR Multiples Examined — le_james94 · 2026-08-23