Models / Kimi K2.6

Moonshot AI

Kimi K2.6

Kimi K2.6 is best for writing; a strong pick at a modest price. The cheapest model of similar intelligence is DeepSeek-V4-Flash (0423), at $0.15 per million tokens against $1.71.

Superseded. Kimi K3 is Moonshot AI's newer model in this line, at $6.00 blended: see Kimi K3 or compare the two.

Results 16 Nov 2026 · 50 results · 9 sources

Test Kimi K2.6 on your prompt → It runs beside the models you could use instead, with every answer checked and costed.

Price per million tokens
$0.95 in · $4.00 out
Moonshot AI's list price, 10 Nov 2026 · compare prices$0.16 cached input
Speed
Not measured
Intelligence Index
27.0
Artificial Analysis; 23.6–27.0 across 2 settings
Spring Prompt overall
Not ranked yet
needs our benchmarks and two groups of results
Developer
Moonshot AI, based in China
Weights published
Released
20 Apr 2026
262,144 tokens of context

Where it stands among the models you could choose

Available models grouped by price, intelligence, speed and developer location, plus models with published weights. Retired models are excluded.

AmongIntelligence IndexSpring Prompt overallPrice, blendedSpeed
Models available today 61st of 185Claude Opus 5.5 – $1.71 blended, 64th percentilecheapest: Mistral Nemo –
Models priced $1–3 per million tokens 14th of 34Muse Spark 1.3 – sets this group –
Models with similar intelligence sets this group – $1.71 blended, 53rd percentilecheapest: DeepSeek-V4-Flash (0423) –
Models from developers based in China 20th of 75MiMo-V2.6-Pro – $1.71 blended, 84th percentilecheapest: Qwen3.5-9B –
Open-weights models 21st of 104MiMo-V2.6-Pro – $1.71 blended, 88th percentilecheapest: Mistral Nemo –

The name under each rank is the leader of that group; green is the top quarter of the group and red the bottom quarter, by rank. Price is shown as a percentile: the share of the group that costs less, so lower is cheaper. "Sets this group" marks the measure the group is defined by. Spring Prompt overall is our score out of 100 across every source (how it works). Price is the developer's list price (or the typical OpenRouter host where we haven't read one) for three input tokens to one output; intelligence and speed are from Artificial Analysis (data sourced from Artificial Analysis); "similar intelligence" means within 4 points on its Intelligence Index. Developer locations are where each company is based, not where a model is served. Groups under 3 models are not ranked.

How good is it, and for what?

Each use case ranks the available models its benchmarks measured, then those priced $1–3 per million tokens. The bar shows its percentile in that field, best to the right. Open a row for the results behind it.

Product listingsTurning a sparse product feed and photos into listings that can go live Not measured
  • CatalogBench: has not measured this model.
Decks from an analysisTurning a finished analysis into a deck you could present as it is Not measured
  • DeckBench: has not measured this model.
User surveysPlanning a user survey and reading its results without being misled Not measured
  • SurveyBench: has not measured this model.
Marketing planningPlanning a year of ad spend without overspending Not measured
  • ROASBench: has not measured this model.
Agents and tool useMulti-step tasks with tools: support desks, coding agents, function calls 67th of 159 available · 14th of 31 at its price
BenchmarkKimi K2.6Rank among availableBest available
OpenHands IndexOpenHands Index: average · reported 57.1% 14th of 31 Claude Fable 5 81.0%
Microsoft STATE-BenchSTATE-Bench v0.7: task success (pass@1) · reported 48.3% 3rd of 4 GPT-5.4 55.7%
Artificial AnalysisArtificial Analysis: Terminal-Bench 4.0 · reported 0.5% 62nd of 107 Kimi K3.1 Pro 67.2%
Artificial AnalysisArtificial Analysis: Terminal-Bench 2.1 · reported 65.9% 37th of 107 Claude Fable 5.1 91.4%
Artificial AnalysisArtificial Analysis: τ-bench banking · reported 23.3% 39th of 104 Grok 4.6 50.7%
Vending-Bench 2Vending-Bench 2: money after a year · reported $6,205 15th of 53 GPT-6 Astra $15,515
  • tau2-bench: has not measured this model.
  • Berkeley Function Calling Leaderboard (BFCL) V4: pending. Its latest results are dated 16 Dec 2025, before this model was released.
Professional workReal tasks from banking, consulting and law, business documents and freelance projects 43rd of 51 available · 8th of 12 at its price
BenchmarkKimi K2.6Rank among availableBest available
GDP.pdfGDP.pdf: professional document tasks · reported 12.0% 28th of 31 GPT-5.6 Sol 30.7%
  • APEX-Agents: has not measured this model.
  • Remote Labor Index: has not measured this model.
Reasoning and knowledgeHard questions across science, maths and general knowledge 56th of 187 available · 12th of 35 at its price
BenchmarkKimi K2.6Rank among availableBest available
Artificial AnalysisArtificial Analysis: Artificial Analysis Intelligence Index · reported 27.0 61st of 185 Claude Opus 5.5 57.6
Artificial AnalysisArtificial Analysis: Humanity's Last Exam · reported 37.5% 53rd of 186 Claude Opus 5.5 61.4%
Artificial AnalysisArtificial Analysis: GPQA Diamond · reported 91.1% 29th of 168 GPT-6 Astra 96.3%
WritingWhat people prefer in blind comparisons, and judged writing quality 36th of 169 available · 8th of 28 at its price
BenchmarkKimi K2.6Rank among availableBest available
Arena (formerly LMArena)Arena Text: overall · reported 1,461 39th of 153 Gemini 4 Argon 1,525
UGI LeaderboardUGI: writing score · reported 64.4 25th of 124 Gemini 3.8 Flash 78.6
Sticking to the factsSummarising without inventing things, and factual answers 78th of 139 available · 17th of 26 at its price
BenchmarkKimi K2.6Rank among availableBest available
Vectara Hallucination LeaderboardVectara: hallucination rate · reported 10.8% 48th of 71 GPT-5.4 nano 3.1%
Arena (formerly LMArena)Arena Text factuality: overall · reported 1,456 40th of 109 Gemini 4 Argon 1,511
SimpleQA Verified (Epoch AI)SimpleQA Verified: correct answers · reported 34.9% 45th of 68 GPT-6 Astra 75.6%
Speed under pressureGood decisions against a real clock (fast chess) Not measured
  • BulletBench: has not measured this model.

Nearest alternatives

Against Kimi K2.6 at $1.71 blended, Intelligence Index 27.0. "Similar intelligence" means within 4 points or better.

Where to run it

16 hosts on OpenRouter at the standard service tier, with a median of $1.44 blended per million tokens; quantisation where the host states it: INT4, FP8, FP4, BF16. Cheapest first.

HostQuantisationInputOutputBlended
ChutesINT4$0.39$1.80$0.74
DigitalOceannot stated$0.57$2.40$1.03
StreamLakeFP8$0.60$2.52$1.08
CoreWeaveFP4$0.65$3.41$1.34
CrusoeBF16$0.70$3.50$1.40
SiliconFlowFP8$0.77$3.40$1.43
DeepInfraFP4$0.75$3.50$1.44
ParasailINT4$0.75$3.50$1.44

Prices per million tokens from OpenRouter's public endpoint list, checked 10 Nov 2026; the 8 dearer hosts are left out.

Against the alternatives

The models you would most likely weigh it against: the leaders of the groups above.

Scroll sideways to see every alternative.

Kimi K2.6Muse Spark 1.3
most intelligent at $1–3 per million tokens
DeepSeek-V4-Flash (0423)
cheapest with similar intelligence
Claude Opus 5.5
most intelligent available today
MiMo-V2.6-Pro
most intelligent from China
Price per million tokens $1.71 $2.00$0.15$8.00$0.54
Tokens a second – 254–9643
Intelligence Index 27.0 48.124.457.646.3
Spring Prompt overall – 59–82–
Rank among available models, by use case
Product listings – 17th–13th–
Decks from an analysis – 25th–7th–
User surveys – ––4th–
Marketing planning – 25th–9th–
Agents and tool use 67th of 159 13th59th3rd23rd
Professional work 43rd of 51 8th–3rd–
Reasoning and knowledge 56th of 187 11th69th1st13th
Writing 36th of 169 7th63rd5th23rd
Sticking to the facts 78th of 139 11th81st3rd10th
Speed under pressure – 33rd–––

Shaded figures are better than Kimi K2.6's. A dash means no result.

Where it has been measured

14th most intelligent of the 34 models priced $1–3 per million tokens. Among models available today, in the top quarter (by percentile) for writing; in the bottom quarter for professional work.

Artificial Analysis lists it but has not published a speed for it yet

Every result

50 published results from 9 sources, each in the source's own units, with the configuration that produced it.

Arena (formerly LMArena) · reported by Arena (formerly LMArena) · 8 results
MeasureValueRankConfigurationDated
Arena Document: overallArena rating, higher is better 1,450
1,442–1,458
32 of 44 Kimi K2.6 13 Sep 2026
Arena Text factuality: overallArena rating, higher is better 1,456
1,452–1,460
53 of 182 Kimi K2.6 2 Oct 2026
Arena Text: business, management and financeArena rating, higher is better 1,455
1,448–1,463
68 of 406 Kimi K2.6 2 Oct 2026
Arena Text: creative writingArena rating, higher is better 1,433
1,425–1,441
70 of 411 Kimi K2.6 2 Oct 2026
Arena Text: expert promptsArena rating, higher is better 1,503
1,493–1,513
37 of 363 Kimi K2.6 2 Oct 2026
Arena Text: instruction followingArena rating, higher is better 1,456
1,449–1,462
56 of 413 Kimi K2.6 2 Oct 2026
Arena Text: overallArena rating, higher is better 1,461
1,457–1,465
58 of 413 Kimi K2.6 2 Oct 2026
Arena Text: writing, literature and languageArena rating, higher is better 1,441
1,434–1,448
70 of 412 Kimi K2.6 2 Oct 2026

Not shown: Not accuracy or correctness: it ranks which answer voters preferred, with answer length and formatting controlled for.

Artificial Analysis · reported by Artificial Analysis · 19 results
MeasureValueRankConfigurationDated
Artificial Analysis: Artificial Analysis Coding Indexindex score, higher is better 61.8 71 of 187 Kimi K2.6 20 Oct 2026
Artificial Analysis: Artificial Analysis Intelligence Indexindex score, higher is better 27.0 136 of 413 Kimi K2.6 20 Oct 2026
Artificial Analysis: Artificial Analysis Intelligence Indexindex score, higher is better 23.6 171 of 413 Kimi K2.6 (no reasoning) 20 Oct 2026
Artificial Analysis: GPQA Diamond% of questions, higher is better 91.1% 51 of 360 Kimi K2.6 20 Oct 2026
Artificial Analysis: GPQA Diamond% of questions, higher is better 78.8% 185 of 360 Kimi K2.6 (no reasoning) 20 Oct 2026
Artificial Analysis: Humanity's Last Exam% of questions, higher is better 37.5% 104 of 419 Kimi K2.6 20 Oct 2026
Artificial Analysis: Humanity's Last Exam% of questions, higher is better 19.6% 204 of 419 Kimi K2.6 (no reasoning) 20 Oct 2026
Artificial Analysis: IFBench% of instructions, higher is better 76.0% 20 of 288 Kimi K2.6 20 Oct 2026
Artificial Analysis: IFBench% of instructions, higher is better 44.3% 173 of 288 Kimi K2.6 (no reasoning) 20 Oct 2026
Artificial Analysis: Long-context reasoning (AA-LCR)% of questions, higher is better 81.0% 58 of 407 Kimi K2.6 20 Oct 2026
Artificial Analysis: Long-context reasoning (AA-LCR)% of questions, higher is better 69.7% 204 of 407 Kimi K2.6 (no reasoning) 20 Oct 2026
Artificial Analysis: SciCode% of problems, higher is better 51.5% 98 of 205 Kimi K2.6 20 Oct 2026
Artificial Analysis: Terminal-Bench 2.1% of tasks, higher is better 65.9% 74 of 188 Kimi K2.6 20 Oct 2026
Artificial Analysis: Terminal-Bench 4.0% of tasks, higher is better 0.5% 143 of 204 Kimi K2.6 20 Oct 2026
Artificial Analysis: Terminal-Bench Hard% of tasks, higher is better 43.9% 36 of 282 Kimi K2.6 20 Oct 2026
Artificial Analysis: Terminal-Bench Hard% of tasks, higher is better 37.9% 60 of 282 Kimi K2.6 (no reasoning) 20 Oct 2026
Artificial Analysis: τ-bench banking% of tasks, higher is better 23.3% 79 of 175 Kimi K2.6 20 Oct 2026
Artificial Analysis: τ²-bench telecom% of tasks, higher is better 95.9% 14 of 286 Kimi K2.6 20 Oct 2026
Artificial Analysis: τ²-bench telecom% of tasks, higher is better 93.9% 35 of 286 Kimi K2.6 (no reasoning) 20 Oct 2026

Not shown: Not your codebase or tools.

GDP.pdf · reported by GDP.pdf · 1 result
MeasureValueRankConfigurationDated
GDP.pdf: professional document tasks% of rubric, higher is better 12.0% 41 of 49 Kimi K2.6 6 Jun 2026

Not shown: Not your documents; 100 tasks, so small differences are noise.

Microsoft STATE-Bench · reported by Microsoft STATE-Bench · 6 results
MeasureValueRankConfigurationDated
STATE-Bench v0.7: consistency (pass^5)% of tasks, higher is better 29.3% 3 of 5 Kimi K2.6 (statebench) 25 May 2026
STATE-Bench v0.7: customer support task success (pass@1)% of tasks, higher is better 45.1% 5 of 5 Kimi K2.6 (statebench) 25 May 2026
STATE-Bench v0.7: shopping assistant task success (pass@1)% of tasks, higher is better 47.9% 5 of 5 Kimi K2.6 (statebench) 25 May 2026
STATE-Bench v0.7: task success (pass@1)% of tasks, higher is better 48.3% 3 of 5 Kimi K2.6 (statebench) 25 May 2026
STATE-Bench v0.7: travel task success (pass@1)% of tasks, higher is better 51.9% 3 of 5 Kimi K2.6 (statebench) 25 May 2026
STATE-Bench v0.7: user experiencescore from 1 to 5, higher is better 3.37 4 of 5 Kimi K2.6 (statebench) 25 May 2026

Not shown: Not the model alone: it runs inside STATE-Bench's agent loop against simulated users, and versions of the benchmark are not comparable.

OpenHands Index · reported by OpenHands Index · 6 results
MeasureValueRankConfigurationDated
OpenHands Index: average% resolved, higher is better 57.1% 14 of 34 Kimi K2.6 (openhands) 30 Jun 2026
OpenHands Index: front end (SWE-Bench Multimodal)% resolved, higher is better 41.2% 6 of 34 Kimi K2.6 (openhands) 30 Jun 2026
OpenHands Index: greenfield (Commit0)% resolved, higher is better 25.0% 14 of 34 Kimi K2.6 (openhands) 30 Jun 2026
OpenHands Index: information gathering (GAIA)% resolved, higher is better 74.5% 9 of 34 Kimi K2.6 (openhands) 30 Jun 2026
OpenHands Index: issue resolution (SWE-Bench)% resolved, higher is better 74.6% 13 of 34 Kimi K2.6 (openhands) 30 Jun 2026
OpenHands Index: testing (SWT-Bench)% resolved, higher is better 70.4% 12 of 34 Kimi K2.6 (openhands) 30 Jun 2026

Not shown: Not the model alone: the score is for the model inside the OpenHands agent, at the SDK version shown.

SimpleQA Verified (Epoch AI) · reported by SimpleQA Verified (Epoch AI) · 1 result
MeasureValueRankConfigurationDated
SimpleQA Verified: correct answers% of questions, higher is better 34.9% 50 of 85 Kimi K2.6 10 Aug 2026

Not shown: Not answers grounded in your documents; tests what the model remembers.

UGI Leaderboard · reported by UGI Leaderboard · 6 results
MeasureValueRankConfigurationDated
UGI: requested-length error% off the requested word count, lower is better 34.0% 278 of 379 Kimi K2.6 (no reasoning) 20 Apr 2026
UGI: requested-length error% off the requested word count, lower is better 18.0% 158 of 379 Kimi K2.6 (thinking reasoning) 20 Apr 2026
UGI: style adherencescore from 0 to 1, higher is better 0.34 224 of 379 Kimi K2.6 (no reasoning) 20 Apr 2026
UGI: style adherencescore from 0 to 1, higher is better 0.35 199 of 379 Kimi K2.6 (thinking reasoning) 20 Apr 2026
UGI: writing scorescore out of 100, higher is better 64.2 74 of 379 Kimi K2.6 (no reasoning) 20 Apr 2026
UGI: writing scorescore out of 100, higher is better 64.4 72 of 379 Kimi K2.6 (thinking reasoning) 20 Apr 2026

Not shown: Not other format limits such as character counts or bullet counts.

Vectara Hallucination Leaderboard · reported by Vectara Hallucination Leaderboard · 2 results
MeasureValueRankConfigurationDated
Vectara: answer rate% of documents, higher is better 99.7% 44 of 108 Kimi K2.6 22 Sep 2026
Vectara: hallucination rate% of summaries, lower is better 10.8% 68 of 108 Kimi K2.6 22 Sep 2026

Not shown: Not a quality score: a low rate usually means content filters were triggered, and hallucination rates are measured on answered documents only.

Vending-Bench 2 · reported by Vending-Bench 2 · 1 result
MeasureValueRankConfigurationDated
Vending-Bench 2: money after a yearUS dollars, higher is better $6,205 15 of 63 Kimi K2.6 1 Oct 2026

Not shown: Not a real business; one simulated market with set rules.

Will Kimi K2.6 work for your product?

Benchmarks aren't your task. Paste the prompt your product uses and run it beside the models you could use instead, with every answer checked and costed.

New here? Your email starts a free trial. No credit card needed.