Kevin Schaul
Kevin Schaul

Visual journalist/hacker covering AI

← All evals

Article tracking: Trump

Determine whether an article is describing a new action/policy by the Trump administration.

Read full blog post

View on GitHub

Model Pass rate Tests Avg tokens
claude-opus-4-0 81% 200 –
claude-haiku-4-5 76% 200 31
kimi-k2-thinking (moonshotai) 76% 200 330
gpt-4o-mini 75% 200 2
qwen3-235b-a22b-2507 75% 200 –
qwen3.6-35b-a3b (venice/fp8) 75% 200 683
deepseek-chat-v3-0324 74% 200 –
qwen3-14b 74% 200 –
claude-3-5-sonnet-20241022 73% 200 23
qwen3-30b-a3b-2507 73% 200 2
qwen3.8-27b (venice/fp8) 73% 200 174
claude-sonnet-4-0 73% 200 –
claude-sonnet-5 72% 200 5
gpt-oss-20b 72% 199 –
claude-fable-5-1 72% 200 48
claude-opus-5 71% 200 74
gemma-3-12b 71% 200 –
gpt-oss-120b 71% 200 119
gpt-5.5 – 21 –
gpt-5.6-luna – 19 –
gpt-6-astra – 23 –

Full results

Loading results…