#34
GPT-5.5
gpt-5.5Helpfulness
Instruction
Following
Comprehension
Empathy
Creative
Writing
Helpfulness
0.0
Instruction Following
0.0
Comprehension
0.0
Empathy
0.0
Creative Writing
0.0
Speed
Avg 72 tok/s
Overall Assistant Score
An average score combining the 5 main categories.
89.90 pts
Rank #34
68th Percentile
0
Novice
33
Capable
66
Proficient
100
Expert
GPT-5.5 is a strong practical assistant for complex, multi-step help—planning, research, and sticking to detailed instructions. Our tests rate it highly on helpfulness, instruction following, and comprehension. Empathy is okay but not warm. Creative writing is where the generation shows its tradeoff: internal reasoning runs in terse, broken-English scratchpads that are efficient for agents and coding but flatten metaphor, rhythm, and voice in the final draft. A reliable workhorse when getting things done beats sounding literary.
Overall Score vs Price
Price axis is logarithmic (USD per 1M input tokens) · Higher score and lower price is better
Ringed point is this model
Anthropic
Baidu
Bytedance
DeepSeek
Google
Meta
Minimax
Mistral
Moonshot
OpenAI
Qwen
SpaceX AI
Tencent
Xiaomi
Zhipu AI
Compare this model
Intelligence
Overall Score · Higher is better
Speed
Output tokens per second · Higher is better
Price
USD per 1M input tokens · Lower is better
OpenAI Models
Overall Score · Same provider
GPT Family
Overall Score · Same model family
Closest Rivals
Overall Score · Nearest by overall rank