Kimi K3
Not listed in any tool yet.
VerdictPal · editorial desk · 2026VerdictPal
Compare · Two models, head-to-head
Pick two models. See the diff.
Composite scores, benchmark ranks, capability grids, and pricing per million tokens. All sourced from the same data that powers the model atlas.
Kimi K3 vs Kimi K2.5
Kimi K33stats won
Tied0stats
Kimi K2.54stats won
Spec showdown
Capacity, price, freshness.
Kimi K3vsKimi K2.5
Kimi K3 dominates on composite score (65 vs 53), driven by broader benchmark coverage and stronger per-family performance. Kimi K3 has the higher Vals/AA intelligence score (75 vs 35). Kimi K2.5 is significantly cheaper at $0.60/1M input vs $3.0/1M input. Kimi K3's 1M context window dwarfs Kimi K2.5's 262K — a meaningful difference for long-document or codebase work.
- Composite scoreKimi K365Winner on this rowKimi K2.553
- Intelligence (Vals/AA)Kimi K375Winner on this rowKimi K2.535
- Context windowKimi K31MWinner on this rowKimi K2.5262K
- Max outputKimi K3—Kimi K2.566KWinner on this row
- Input price / 1M (lower wins)Kimi K3$3.0Kimi K2.5$0.60Winner on this row
- Output speed (AA default API)Kimi K338 t/sKimi K2.550 t/sWinner on this row
- Time to first token (AA default API)Kimi K36.19sKimi K2.51.02sWinner on this row
Composite score
vs
Intelligence (Vals/AA)
vs
Context window
vs
Max output
vs
Input price / 1M (lower wins)
vs
Output speed (AA default API)
vs
Time to first token (AA default API)
vs
Capability grid
What each model can do.
Modalities
Access
Family radar
Top score per benchmark family.
Reasoning
vs
Coding
vs
Agentic
vs
Long context
vs
Benchmark ranks
Every shared benchmark, side by side.
Reasoning
Artificial Analysis Intelligence Index#3/67 · #44/67
vs
GPQA Diamond#4/69 · #32/69
vs
Humanity's Last Exam#7/65 · #37/65
vs
Coding
SciCode#3/65 · #31/65
vs
Agentic
Terminal-Bench#3/69 · #48/69
vs
τ³-Bench Banking#20/64 · #48/64
vs
Long context
AA-LCR#2/64 · #39/64
vs
Performance
AA output speed#57/69 · #55/69
vs
AA time to first token#24/69 · #43/69
vs
Dossier facts
Dates, sources, fineprint.
Ecosystem
Which tools wrap each model.
Kimi K2.5
Not listed in any tool yet.
Verdict
Which model to actually pick.
Kimi K3 dominates on composite score (65 vs 53), driven by broader benchmark coverage and stronger per-family performance. Kimi K3 has the higher Vals/AA intelligence score (75 vs 35). Kimi K2.5 is significantly cheaper at $0.60/1M input vs $3.0/1M input. Kimi K3's 1M context window dwarfs Kimi K2.5's 262K — a meaningful difference for long-document or codebase work.