AI News
Google announces Gemini 4 Carbon, with cyber defenders first
Carbon leads Artificial Analysis. Google’s rollout starts with selected cyber defenders.
Latest news
Lumina Journal
AI News
OpenAI launches GPT-6.1 Terra at half Sol’s API price
GPT-6.1 Terra is live in ChatGPT Work, Codex and the API. OpenAI is pitching faster computer use and lower costs, with mixed results against its larger models.
AI News
Kimi K3.1 Pro is out, with a million-token context window
Kimi’s new Pro and Highspeed models are live across its apps and API. The release brings strong coding results, a million-token context window and model weights promised by 4 November.
Showing 3 of 25 stories
LuminaBench / Capabilities
AI capabilities, compared.
One index across our creation tests. Explore the quality of the results and the resources it takes to make them.
Lumina Capabilities Index
Capability across tested configurations.
Complete suite · all active categories · 20 of 36 configurations
Capabilities Index vs Cost
Capability and total suite cost
Complete suite · all active categories · 8 of 9 families
Capabilities Index vs Generation Time
Capability and total suite generation time
Complete suite · all active categories · 8 of 9 families
Capabilities Index vs Tokens
Capability and total suite tokens
Complete suite · all active categories · 8 of 9 families
Capabilities Index vs Tool Calls
Capability and total suite tool calls
Complete suite · all active categories · 8 of 9 families
Scores average the category indexes, rounded to a whole number. Resource totals add the completed tests.