perf: five cuts to the per-message allocation count - #1131
Merged
CodSpeed HQ / CodSpeed Performance Analysis
succeeded
Jul 26, 2026 in 0s
Performance Gate Passed
⚠️ Different runtime environments detected
Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.
⚡ 4 improved benchmarks
✅ 194 untouched benchmarks
Performance Changes
| Mode | Benchmark | BASE |
HEAD |
Efficiency | |
|---|---|---|---|---|---|
| ⚡ | Memory | bench_content_extraction_simple |
173 B | 30 B | ×5.8 |
| ⚡ | Memory | bench_full_token_generation_simple |
173 B | 30 B | ×5.8 |
| ⚡ | Memory | bench_content_extraction_extended |
471 B | 129 B | ×3.7 |
| ⚡ | Memory | bench_full_token_generation_extended |
471 B | 129 B | ×3.7 |
Tip
Curious why this is faster? Comment @codspeedbot explain why this is faster on this PR, or directly use the CodSpeed MCP with your agent.
Comparing perf/alloc-round3 (647d57d) with main (f6a20a3)
Loading