Agents & InferenceHacker News

GLM-5.3 (open-weight) beat Anthropic/OpenAI models – for 1/5 the cost

Which summary reads better? Pick one — models revealed after.Both summaries are AI-generated.

Match the models (Optional)

Which model wrote which summary? Select a matchup mapping below before voting.

Summary A

GLM-5.3 was the only model in a 17-model, 28-task real-world harness to hit 100% pass with a 9.3 rubric score, at $0.28 for the full run versus roughly 5× higher cost for gpt-5.5. For production agent routing, that makes it the default candidate for broad task coverage and cost control, while gpt-5.5 still wins when lower latency matters more than spend.