The two are evenly matched.
Scores updated · 9 tests both models report · How we compare
Where each one wins
Tests won in each of the one area where both have results. Each piece is one test, so a longer bar means more evidence; grey means the two scored within a point of each other.
What it costs
Prices per million tokens, roughly 750,000 words. The bars show the cost of a million tokens read plus a million written.
ERNIE 5.5 Flash Preview costs 72% less for the same work.
The biggest differences
The tests each model wins by the widest margin, up to three each. Scores are out of 100.
Where ERNIE 5.5 Flash Preview pulls ahead
- Long professional tasks in banking, consulting and lawAA ApexAgents+22.9points ahead
Where Kimi K3 pulls ahead
- Finds hard-to-locate facts by browsing the webBrowseComp+3.8points ahead
Every test, side by side
All 9 tests both models report. The winning score is in its model's colour; marks a score checked independently.
AgentsEven
- AA ApexAgentsERNIE 5.5 Flash Preview by 22.964.241.3+22.9
- BrowseCompKimi K3 by 3.887.491.2+3.8
- MCP Atlastie84.684.2tie
Other results6 tests, not counted
Tests outside the eight areas. They are not counted above: several are summary scores built from other tests, or the same test under another name.
- Terminal-Bench 4.0ERNIE 5.5 Flash Preview by 33.946.512.6+33.9
- NL2RepoERNIE 5.5 Flash Preview by 6.864.858+6.8
- OfficeQA ProERNIE 5.5 Flash Preview by 4.667.963.3+4.6
- DeepSWE 1.1ERNIE 5.5 Flash Preview by 2.271.269+2.2
- SpreadsheetBench 2ERNIE 5.5 Flash Preview by 1.636.434.8+1.6
- Agents' Last Examtie2928.3tie
Questions people ask
Which is better, ERNIE 5.5 Flash Preview or Kimi K3?
Neither wins more tests than the other in any of the one area where both have results. ERNIE 5.5 Flash Preview costs 72% less.
Which is cheaper?
ERNIE 5.5 Flash Preview costs $1.00 per million input tokens and $4.00 per million output tokens; Kimi K3 costs $3.00 and $15.00. That makes ERNIE 5.5 Flash Preview about 72% cheaper for the same work.
How do you compare the two?
We use the 9 benchmark tests both models have published scores on. The verdict counts the 3 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 6 are listed but not counted, because several are summary scores or repeat a test. Each score is the one shown on the model's own page.