Skip to main content
BenchLM
Advertise

Model comparison

Kimi K2.6 vs Step 3.7 Flash

Data verified

Head-to-head evidence from 28 shared benchmark results across 6 categories. Overall scores shown here use the public BenchAlign v5 ranking lane.

Kimi K2.6
Moonshot AI
57.04/100
Margin
6.3pts
← winning
50.76/100
2 category wins0 category wins

Verified leaderboard positions: Kimi K2.6 #11; Step 3.7 Flash unranked

BenchAlign evidence: Kimi K2.6 estimated; Step 3.7 Flash estimated. Intervals and evidence labels describe ranking uncertainty, not a guarantee for a specific workload.

Evidence parity. Kimi K2.6 and Step 3.7 Flash share 28 comparable benchmark results. 2 of 8 categories are comparable. 30 results are unique to Kimi K2.6; 2 to Step 3.7 Flash.

Updated July 16, 2026
Shared results
28
Kimi K2.6 only
30
Step 3.7 Flash only
2
Comparable categories
2 / 8

Pick Kimi K2.6 if you want the stronger benchmark profile. Step 3.7 Flash only becomes the better choice if you want the cheaper token bill.

Confidence note. This is a partial-evidence comparison with 28 shared benchmark results across 6 evidence categories; 2 of 8 categories currently have scoreable aggregates for both models. Treat the verdict as directional until coverage is more balanced.

Why this result

Kimi K2.6 is clearly ahead on the provisional aggregate, 71 to 57. The gap is large enough that you do not need to squint at the spreadsheet to see the difference.

Kimi K2.6's sharpest advantage is in coding, where it averages 64.4 against 56.3. The single biggest benchmark swing on the page is BrowseComp, 83.2% to 75.8%.

Kimi K2.6 is also the more expensive model on tokens at $0.95 input / $4.00 output per 1M tokens, versus $0.20 input / $1.15 output per 1M tokens for Step 3.7 Flash. That is roughly 3.5x on output cost alone.

Category breakdown

Exact category averages are shown below. Not measured means BenchLM does not have enough sourced public coverage for that model and category.

Category scores and score margins for Kimi K2.6 and Step 3.7 Flash
CategoryKimi K2.6ΔStep 3.7 Flash
CodingKimi K2.664.4Margin← 8.1Step 3.7 Flash56.3
AgenticKimi K2.673.5Margin← 7.1Step 3.7 Flash66.4
KnowledgeKimi K2.642.2MarginNo overlapStep 3.7 FlashNot measured
MathKimi K2.667.1MarginNo overlapStep 3.7 FlashNot measured
MultimodalKimi K2.679.8MarginNo overlapStep 3.7 FlashNot measured

Decisive benchmark drivers

The largest measured benchmark gaps in this matchup, with exact reported values.

More
A · Kimi K2.6B · Step 3.7 Flash

  1. BrowseComp

    Agentic
    Source ↗
    A 83.2%B 75.8%
    Winner: Kimi K2.6Δ 7.4
    BrowseComp: Kimi K2.6 scored 83.2%; Step 3.7 Flash scored 75.8%. Kimi K2.6 wins this benchmark.

  2. Terminal-Bench 2.0

    Agentic
    Source ↗
    A 66.7%B 59.5%
    Winner: Kimi K2.6Δ 7.2
    Terminal-Bench 2.0: Kimi K2.6 scored 66.7%; Step 3.7 Flash scored 59.5%. Kimi K2.6 wins this benchmark.

  3. SWE-bench Pro

    Coding
    Source ↗
    A 58.6%B 56.3%
    Winner: Kimi K2.6Δ 2.3
    SWE-bench Pro: Kimi K2.6 scored 58.6%; Step 3.7 Flash scored 56.3%. Kimi K2.6 wins this benchmark.

Operational comparison

Runtime and commercial metrics are compared only when both models have a complete sourced value.

MetricKimi K2.6Step 3.7 FlashComparison
Input / output priceUSD per 1M tokensKimi K2.6$0.95 input / $4 outputStep 3.7 Flash$0.2 input / $1.15 outputStep 3.7 Flash has the lower combined listed price.
Generation speedtokens per secondKimi K2.6Not availableStep 3.7 FlashNot availableA complete speed comparison is not available.
First-answer latencyseconds to first tokenKimi K2.6Not availableStep 3.7 FlashNot availableA complete latency comparison is not available.
Context windowmaximum listed tokensKimi K2.6256KStep 3.7 Flash256KListed context windows are equal.

Benchmark Deep Dive

AgenticKimi K2.6 wins
BenchmarkKimi K2.6Step 3.7 FlashResult
Terminal-Bench 2.0Source 66.7%59.5%Kimi K2.6 leads
BrowseCompSource 83.2%75.8%Kimi K2.6 leads
OSWorld-VerifiedSource 73.1%—Not comparable
ToolathlonSource 50%49.5%Kimi K2.6 leads
MCP AtlasSource 55.9%—Not comparable
Claw-EvalSource 62.3%67.1%Step 3.7 Flash leads
DeepSearchQASource 92.5%92.8%Step 3.7 Flash leads
WideResearchSource 80.8%—Not comparable
AA Agentic IndexSource 30.3%21.5%Kimi K2.6 leads
τ²-bench resultsSource 95.9%98.5%Step 3.7 Flash leads
GDPval-AASource 34.5%25.9%Kimi K2.6 leads
GDPval-AASource 11911017Kimi K2.6 leads
APEX-Agents-AASource 28.5%14.8%Kimi K2.6 leads
Gert LabsSource 56.82%51.57%Kimi K2.6 leads
ResearchClawBenchSource 18.0%—Not comparable
OSWorld 2.0Source 4.6%—Not comparable
AA BriefcaseSource 815—Not comparable
AA EnterpriseOps-GymSource 38.5%—Not comparable
AA Harvey LABSource 0.0%—Not comparable
AA ITBenchSource 31.2%—Not comparable
AA Tau3 BankingSource 20.6%—Not comparable
HLE w/ toolsSource —47.2%Not comparable
CodingKimi K2.6 wins
BenchmarkKimi K2.6Step 3.7 FlashResult
SWE-bench VerifiedSource 80.2%—Not comparable
LiveCodeBench v6Source 89.6%—Not comparable
SWE-bench ProSource 58.6%56.3%Kimi K2.6 leads
SWE MultilingualSource 76.7%—Not comparable
SciCodeSource 52.2%—Not comparable
Terminal-Bench 2.0Source 66.7%59.5%Kimi K2.6 leads
Vibe Code BenchSource 37.89%—Not comparable
cursorBench31Source 47.6%—Not comparable
AA Coding IndexSource 61.8%39.6%Kimi K2.6 leads
Terminal-Bench HardSource 43.9%35.6%Kimi K2.6 leads
AA-SciCodeSource 53.5%40.0%Kimi K2.6 leads
AA Terminal-Bench 2.1Source 65.9%—Not comparable
Reasoning
BenchmarkKimi K2.6Step 3.7 FlashResult
AA-LCRSource 69.7%63.7%Kimi K2.6 leads
CritPtSource 8.0%2.3%Kimi K2.6 leads
Knowledge
BenchmarkKimi K2.6Step 3.7 FlashResult
GPQASource 90.5%—Not comparable
GPQA-DSource 90.5%—Not comparable
HLESource 34.7%—Not comparable
Artificial Analysis Intelligence IndexSource 44.2%30.3%Kimi K2.6 leads
AA-GPQA DiamondSource 91.1%80.9%Kimi K2.6 leads
AA-HLESource 35.9%19.9%Kimi K2.6 leads
AA-Omniscience IndexSource 6.4%-37.5%Kimi K2.6 leads
AA-Omniscience AccuracySource 32.8%25.4%Kimi K2.6 leads
AA-Omniscience Hallucination RateSource 39.3%84.4%Kimi K2.6 leads
AA Openness IndexSource 33.3%—Not comparable
Math
BenchmarkKimi K2.6Step 3.7 FlashResult
AIME26Source 96.4%—Not comparable
HMMT Feb 2026Source 92.7%—Not comparable
MMAnswerBenchSource 86.0%—Not comparable
FrontierMath v2 (Tiers 1-3)Source 38.966%—Not comparable
FrontierMath v2 (Tier 4)Source 14.580%—Not comparable
Multimodal
BenchmarkKimi K2.6Step 3.7 FlashResult
MMMU-ProSource 79.4%—Not comparable
MMMU-Pro w/ PythonSource 80.1%—Not comparable
CharXivSource 80.4%—Not comparable
MathVisionSource 87.4%—Not comparable
V*Source 96.9%95.3%Kimi K2.6 leads
AA-MMMU-ProSource 79.4%75.3%Kimi K2.6 leads
Design Arena WebsiteSource 13101218Kimi K2.6 leads
SimpleVQASource —79.2%Not comparable
Inst. Following
BenchmarkKimi K2.6Step 3.7 FlashResult
AA-IFBenchSource 76.0%67.3%Kimi K2.6 leads
Frequently Asked Questions (3)

Which is better, Kimi K2.6 or Step 3.7 Flash?

Kimi K2.6 is ahead on BenchLM's provisional leaderboard, 71 to 57. The biggest single separator in this matchup is BrowseComp, where the scores are 83.2% and 75.8%.

Which is better for coding, Kimi K2.6 or Step 3.7 Flash?

Kimi K2.6 has the edge for coding in this comparison, averaging 64.4 versus 56.3. Inside this category, AA Coding Index is the benchmark that creates the most daylight between them.

Which is better for agentic tasks, Kimi K2.6 or Step 3.7 Flash?

Kimi K2.6 has the edge for agentic tasks in this comparison, averaging 73.5 versus 66.4. Inside this category, GDPval-AA is the benchmark that creates the most daylight between them.

Self-host vs API cost

Estimates at 50,000 req/day · 1000 tokens/req average.

Kimi K2.6
API / mo$3,713
Self-host / mo$18,221
Break-even326M/day
Step 3.7 Flash
API / mo$1,012
Self-host / moNot listed
Break-even—
Proprietary model — self-hosting not applicable.
Model the full break-even

Related Comparisons

Explore More

Moonshot AI Compare Pricing Methodology Find Your Best LLM Overall Rankings
Last updated: July 16, 2026

The AI models change fast. We track them for you.

A weekly brief for engineers and researchers covering new models, ranking shifts, and pricing changes.

Free. No spam. Unsubscribe anytime.