Models · Moonshot AI (Kimi) · Out sinceReleased 20 Apr 2026

Kimi K2.6

Available on OpenRouterOn OpenRouterAvailable nowGenerally availableCan run on your own serversOpen weights

Kimi K2.6 is made by Moonshot AI (Kimi). We don't have enough test results yet to rank it. It's mid-priced to use. Its makers have published it, so you can run it on your own servers.

1T/32B active MoE, modified-MIT. Thinking and Instant (non-thinking) modes.

Writing & creativity: —
Research & analysis: —
Coding: —
Price: Mid-priced$0.95 / $4
Price per 1M (blended)Blended / 1M: $1.71
MemoryContext: 262K
Longest answerMax output: —
Test resultsResults: 29 (7 independent7 indep.)
Out sinceReleased: 20 Apr 2026
Made byVendor: Moonshot AI (Kimi)
UnderstandsInputs: text, image, video

Scores are out of 100 for each area, compared with every model we track; “#” is its rank. Context is how much text it can read at once (1M is roughly 700,000 words). Blended price mixes the cost of what you send and what it writes back.

Thinking levels

Reasoning settings

No reasoningDefault

Links

Using it through OpenRouter

On OpenRouter

OpenRouter is a service that gives access to many AI models in one place. This model is listed there as moonshotai/kimi-k2.6.

Route it as moonshotai/kimi-k2.6 at $0.65 in / $3.41 out per 1M tokens, 262K context. Listed since 20 Apr 2026.

frequency_penaltyinclude_reasoninglogit_biaslogprobsmax_tokensmin_pparallel_tool_callspresence_penaltyreasoningrepetition_penaltyresponse_formatseedstopstructured_outputstemperaturetool_choicetoolstop_ktop_logprobstop_p

Run it yourself

Self-hosting

On your own servers,Licence, sizeunder your own control.and hardware.

Compare all open models →All open-weights models →

Details comingselfHost data pending

This model can be downloaded and run on your own servers. We're collecting its licence, size and hardware needs now.Open-weights model; licence, parameter count, hardware tier and base-model availability will appear after the next data run.

Every test result

Every result

All the numbers,with where they came from.

Every result we've found for this model, with who measured it and a link to where we read it.

All raw rows (vendor and independent kept separate), with setting label, source, date, harness and notes.

All results for this model
						NotesNotes
AIME 2026	96.4%	Defaultthinking	Maker's own figureVendor-reportedMoonshot AI ↗	20 Apr 2026	—
BrowseComp	83.2%	Defaultthinking	Maker's own figureVendor-reportedMoonshot AI ↗	20 Apr 2026	—	Discard-all context management; 86.3 with Agent Swarm
EQ-Bench Creative Writing v3 (Elo)	1725	Defaultmoonshotai/Kimi-K2.6	Independent testIndependentEQ-Bench ↗	1 Oct 2026	—	leaderboard rank 31; rubric score 16.67/20; slop 13.30; avg length 8333 chars; Elo judged by Claude Sonnet 4.6; effort not stated by EQ-Bench
GPQA Diamond	90.5%	Defaultthinking	Maker's own figureVendor-reportedMoonshot AI ↗	20 Apr 2026	—
HMMT February 2026	92.7%	Defaultthinking	Maker's own figureVendor-reportedMoonshot AI ↗	20 Apr 2026	—	HMMT 2026 (Feb)
Humanity's Last Exam	34.7%	Defaultthinking	Maker's own figureVendor-reportedMoonshot AI ↗	20 Apr 2026	—	HLE-Full no tools (36.4 text-only)
Humanity's Last Exam (with tools)	54.0%	Defaultthinking	Maker's own figureVendor-reportedMoonshot AI ↗	20 Apr 2026	—	HLE-Full with search/code/browse tools
IMO-AnswerBench	86.0%	Defaultthinking	Maker's own figureVendor-reportedMoonshot AI ↗	20 Apr 2026	—
Kimi Code Bench v2 (internal)	50.9%	Defaultthinking	Maker's own figureVendor-reportedMoonshot AI ↗	12 Jun 2026	Kimi Code CLI	Kimi K2.6 column in the K2.7 Code model card
LiveCodeBench	86.8%	Default	Independent testIndependentVals.ai ↗	1 Sep 2026	—	±0.972 stderr; $0.065/test
LiveCodeBench	89.6%	Defaultthinking	Maker's own figureVendor-reportedMoonshot AI ↗	20 Apr 2026	—	LiveCodeBench v6
LLM Creative Story-Writing Benchmark (Lech Mazur)	0.1	DefaultKimi K2.6	Independent testIndependentLech Mazur (LLM Creative Story-Writing Benchmark) ↗	28 Sep 2026	—	rank 28/56; Thurstone comparison score (centered at 0); est. win chance 52%; 95% bootstrap 0.006 to 0.200
MCP Atlas	69.4%	Defaultthinking	Maker's own figureVendor-reportedMoonshot AI ↗	12 Jun 2026	Kimi Code CLI	Kimi K2.6 column in the K2.7 Code model card
MCPMark	55.9%	Defaultthinking	Maker's own figureVendor-reportedMoonshot AI ↗	20 Apr 2026	—
MCPMark Verified	72.8%	Defaultthinking	Maker's own figureVendor-reportedMoonshot AI ↗	12 Jun 2026	Kimi Code CLI	Kimi K2.6 column in the K2.7 Code model card
MLS-Bench-Lite	26.7%	Defaultthinking	Maker's own figureVendor-reportedMoonshot AI ↗	12 Jun 2026	Kimi Code CLI	Kimi K2.6 column in the K2.7 Code model card
MMMU-Pro (no tools)	79.4%	Defaultthinking	Maker's own figureVendor-reportedMoonshot AI ↗	20 Apr 2026	—	Without tools; 80.1 with Python
OSWorld 2.0	4.6%	Defaultkimi-k2.6	Independent testIndependentOSWorld 2.0 via Epoch AI ↗	1 Oct 2026	—	read from Epoch AI benchmark_data.zip (osworld_2_external.csv); original leaderboard https://osworld-v2.xlang.ai/; partial score 0.22100000000000003; tool setting standard; step budget 500
OSWorld-Verified	73.1%	Defaultthinking	Maker's own figureVendor-reportedMoonshot AI ↗	20 Apr 2026	—
ProgramBench (fully resolved)	48.3%	Defaultthinking	Maker's own figureVendor-reportedMoonshot AI ↗	12 Jun 2026	Kimi Code CLI	Kimi K2.6 column in the K2.7 Code model card
SciCode	52.2%	Defaultthinking	Maker's own figureVendor-reportedMoonshot AI ↗	20 Apr 2026	—
SWE-bench Multilingual	76.7%	Defaultthinking	Maker's own figureVendor-reportedMoonshot AI ↗	20 Apr 2026	In-house SWE-agent-derived framework (bash/createfile/insert/view/strreplace/submit)	avg of 10 runs
SWE-Bench Pro (public, v1)	58.6%	Defaultthinking	Maker's own figureVendor-reportedMoonshot AI ↗	20 Apr 2026	In-house SWE-agent-derived framework (bash/createfile/insert/view/strreplace/submit)	avg of 10 runs
SWE-bench Verified	76.7%	Defaultkimi-k2.6	Independent testIndependentEpoch AI ↗	8 May 2026	—	Epoch-run SWE-bench Verified (Epoch scaffold); no runs after 2026-06; stderr 1.9pt; data https://epoch.ai/data/benchmark_data.zip (swe_bench_verified.csv)
SWE-bench Verified	80.2%	Defaultthinking	Maker's own figureVendor-reportedMoonshot AI ↗	20 Apr 2026	In-house SWE-agent-derived framework (bash/createfile/insert/view/strreplace/submit)	avg of 10 runs
Terminal-Bench 2.0	66.7%	Defaultthinking	Maker's own figureVendor-reportedMoonshot AI ↗	20 Apr 2026	Terminus 2	Preserve-thinking mode, avg of 10 runs
Toolathlon	50.0%	Defaultthinking	Maker's own figureVendor-reportedMoonshot AI ↗	20 Apr 2026	—
Vals CorpFin v2	66.7%	Defaultvals id kimi/kimi-k2.6	Independent testIndependentVals.ai ↗	12 Aug 2026	—	vals id kimi/kimi-k2.6; rank 17/134; ±0.928 stderr; $0.07328/test
Vectara Hallucination Leaderboard (HHEM)	10.8%	Defaultmoonshotai/kimi-k2.6	Independent testIndependentVectara ↗	22 Sep 2026	—	factual consistency 89.2 %; answer rate 99.7 %; avg summary 116.7 words; HHEM-2.3 judge; effort not stated (API default)