Agentic search at the Pareto frontier
Context-1, an open-weights model from Chroma, is trained with SFT and RL, and is competitive with frontier models at search, while being 10x faster and 25x cheaper.
Context-1
Executed Context-1 Search Subagent
• Search for bullcase on NY Times business
• Searching
searched bullcase NY Times business positive investment thesis media company profitability growth prospects
searched New York Times business model revenue streams digital transformation subscription strategy competitive advantages market positioning
searched NY Times stock analysis bullish case positive outlook investor report
searched New York Times strong growth trajectory profitability prospects subscription model digital revenue business case for investors
searched The New York Times compelling investment thesis market leadership journalism quality subscription growth long-term value creation
grepped bullcase.*NY Times|bullish.*New York Times|NY Times.*positive.*case
searched NY Times shares investment opportunity long-term upside market share advantage journalism quality subscription growth potential
searched The New York Times company strategic vision 15 million subscribers target 2027 valuation case for investors
selected 6 documents
• Concluded search with 6 documents
Inference speed is the bottleneck
Agents spend a large percentage of their time gathering information.
| Task | Time |
|---|---|
| LLM inference | 10.0s |
| search() | 0.1s |
| LLM inference | 18.4s |
| search() | 0.1s |
| LLM inference | 15.2s |
| search() | 0.1s |
| LLM inference | 48.8s |
| search() | 0.1s |
| LLM inference | 13.1s |
| search() | 0.1s |
| LLM inference | 31.7s |
| search() | 0.1s |
| LLM inference | 15.2s |
| search() | 0.1s |
| LLM inference | 22.4s |
| search() | 0.1s |
| LLM inference | 13.2s |
| search() | 0.1s |
| LLM inference | 48.9s |
| search() | 0.1s |
| LLM inference | 16.3s |
| search() | 0.1s |
| LLM inference | 24.3s |
| search() | - |
| Total | 278.5s |
| Task | Time |
|---|---|
| Opus 4.5 | 400 tok/s |
| LLM inference | 1.0s |
| search() | 0.1s |
| LLM inference | 1.8s |
| search() | 0.1s |
| LLM inference | 1.5s |
| search() | 0.1s |
| LLM inference | 4.9s |
| search() | 0.1s |
| LLM inference | 1.3s |
| search() | 0.1s |
| LLM inference | 3.2s |
| search() | 0.1s |
| LLM inference | 1.5s |
| search() | 0.1s |
| LLM inference | 2.2s |
| search() | 0.1s |
| LLM inference | 1.3s |
| search() | 0.1s |
| LLM inference | 4.9s |
| search() | 0.1s |
| LLM inference | 1.6s |
| search() | 0.1s |
| LLM inference | 2.4s |
| Total | 28.8s |