BREAKING
Decagon CEO on Open vs Closed AI
0
%
Decagon open-source workloads
Open source share of LLM spend falls
Year ago
19
Now
11
When each model type wins
Closed source
New tasks
●
General intelligence
●
Fast prototyping
●
No infra to manage
Open source
Mature tasks
●
Fine-tuned for speed
●
High volume
●
Lower latency
The maturity curve for AI use cases
1
Discovery on frontier
↓
2
Map failure modes
↓
3
Fine-tune open model
↓
4
Stable production
Choice depends on maturity
AI NEWS BLITZ
Decagon's CEO says closed-source models still win most enterprise AI work.