Which is better: Kimi K3 or SWE-2?

AskClash compares Kimi K3 and SWE-2 side by side across SWE-bench, GPQA, HLE, Terminal-Bench, coding agent scores, token pricing, AskClash ACB, and context window so you can see which model wins on each benchmark.

How do Kimi K3 and SWE-2 compare on coding benchmarks?

The comparison table shows SWE-bench, SWE-Pro, SWE-Atlas, Terminal-Bench, DeepSWE, and coding agent index scores for both Kimi K3 and SWE-2 when publicly disclosed.

LLM Comparison

Kimi K3 vs SWE-2: benchmark scores, pricing & comparison.

Side-by-side Kimi K3 vs SWE-2 comparison across SWE-bench, GPQA, HLE, Terminal-Bench, coding agent scores, token pricing, context window, and AskClash RWT. Green marks the winner on each benchmark.

Open live leaderboard Kimi K3 model page SWE-2 model page

Rank #17 vs #24AskClash overall scores 63.9 vs 59.7.

Pricing $3.00/$15.0 vs $0/$0Input and output token prices per 1M tokens when published.

Open Weight vs ProprietaryMoonshot AI vs Cognition.

Kimi K3 vs SWE-2 benchmark comparison

Green cells highlight the winning model for each metric. Scores are cached from the AskClash LLM leaderboard snapshot.

Metric	Kimi K3	SWE-2
Overall Score	63.9	59.7
Leaderboard Rank	#17	#24
ACB	57.7	60.0
RWT	8.0	8.5
Coding Agent Index	51.9	—
HLE	43.5	—
GPQA	93.5	—
SWE-Atlas	66.1	—
Terminal-Bench	88.3	92.8
DeepSWE	68.5	73.0
GDPval-AA	1668.0	—
MCP Atlas	84.2	—
Finance Agent	54.4	—
CharXiv	84.8	—
MMMU-Pro	81.6	—
Input Price (per 1M tokens)	$3.00	$0
Output Price (per 1M tokens)	$15.0	$0
Context Window	1M	256K
Benchmarks Published	13	3

Kimi K3 vs SWE-2 head-to-head charts

Kimi K3 leads 1 and SWE-2 leads 4 of 5 shared benchmarks. SWE-2 is cheaper on a defined 1M-input / 200K-output workload. Charts show only benchmarks both models publish.

Kimi K3SWE-2

Overall

63.9Kimi K3

59.7SWE-2

ACB

57.7Kimi K3

60.0SWE-2

RWT

8.0Kimi K3

8.5SWE-2

Terminal-Bench

88.3Kimi K3

92.8SWE-2

DeepSWE

68.5Kimi K3

73.0SWE-2

Kimi K3

Input$3.00

Output$15.0

Workload$6.00

Context1M

SWE-2

Input$0

Output$0

Workload$0.00

Context256K

Workload = published cost of 1M input + 200K output tokens. Open the live leaderboard for interactive compare charts.

More Kimi K3 and SWE-2 comparisons

Explore how Kimi K3 and SWE-2 stack up against other top-ranked LLMs.

Claude Fable 5.1 vs Kimi K3 Claude Fable 5.1 vs SWE-2 Claude Fable 5 vs Kimi K3 Claude Fable 5 vs SWE-2 Claude Opus 5 vs Kimi K3 Claude Opus 5 vs SWE-2 GPT-5.6 Sol vs Kimi K3 GPT-5.6 Sol vs SWE-2 Qwen3.8 Max vs Kimi K3 Qwen3.8 Max vs SWE-2

How to read this comparison

Benchmark scores

Higher is better for all benchmark scores (SWE-bench, GPQA, HLE, Terminal-Bench, etc.). Green marks the model with the higher score.

Token pricing

Lower is better for input and output prices. Green marks the cheaper model per 1M tokens.

Coverage matters

Models with fewer disclosed benchmark cells may have inflated percentile scores. Check the benchmark cell count for context.

This comparison page is generated from the AskClash LLM leaderboard cache. Open the live leaderboard for real-time scores and interactive filtering.