AI Tools Review

Kimi K3.1 Pro

By Kimi

Released: 2026-10-23

LLM
Kimi
Open Weights
Free
New

Kimi K3.1 Pro is Kimi's open-weight AI model, released 23/10/2026. Moonshot's model card reports Kimi K3.1 Pro at maximum reasoning effort (temperature 1.0, top-p 0.95) against Claude Fable 5.1, GPT-6 Astra, Claude Opus 5.5, DeepSeek-V4.1-Flash-1017 and GLM-5.3. It is close to Claude Opus 5.5 on Terminal-Bench 4.0 (65.7 against 66.4) and ahead on DeepSWE v1.1 and BrowseComp, but behind the closed models on several reasoning and Program Bench rows. Weights are published under a custom licence, and many rows are Moonshot internal benchmarks.

Visit Kimi K3.1 Pro
Read our full in-depth review: Kimi K3.1 Pro Review: Is It Worth Using?

AI-Powered

Leverages advanced AI technology to deliver cutting-edge capabilities and results.

Fast & Efficient

Optimized performance ensures quick results without compromising on quality.

Purpose-Built

Specifically designed for llms tasks and workflows.

How does Kimi K3.1 Pro score on independent benchmarks?

Kimi K3.1 Pro ranks #11 of 28 models on the Artificial Analysis Intelligence Index (v5.0.1) with a score of 46.5, as measured on 22 November 2026.

Intelligence Index
46.5
AA v5.0.1
Coding Agent Index
48.6
in Kimi Code CLI
Cost per task
$3.68
AA Intelligence Index task
Price per 1M tokens
$2 / $10
input / output

Where does Kimi K3.1 Pro rank against other models?

How good is Kimi K3.1 Pro at agentic coding?

How does Kimi K3.1 Pro do on individual tests?

Evaluation (run by Artificial Analysis)Kimi K3.1 ProBest of any model
Terminal-Bench 4.067.2%67.2%
FrontierCode 2.025.5%48.8%
Humanity's Last Exam52.5%74.3%
GDPval-AA (Elo)18611866

How has Kimi K3.1 Pro's score changed over time?

We have tracked Kimi K3.1 Pro on Intelligence Index v5.0.1 since 22/11/2026. A point is added each day the score, coding index or price changes, so this chart fills in over time.

What does the Kimi K3.1 Pro system card report?

Moonshot's model card reports Kimi K3.1 Pro at maximum reasoning effort (temperature 1.0, top-p 0.95) against Claude Fable 5.1, GPT-6 Astra, Claude Opus 5.5, DeepSeek-V4.1-Flash-1017 and GLM-5.3. It is close to Claude Opus 5.5 on Terminal-Bench 4.0 (65.7 against 66.4) and ahead on DeepSWE v1.1 and BrowseComp, but behind the closed models on several reasoning and Program Bench rows. Weights are published under a custom licence, and many rows are Moonshot internal benchmarks.

These are Moonshot AI's own published results (self-reported, with the vendor's harness and settings), so they are not directly comparable with other labs' numbers. Sources: Kimi K3.1 Pro (Hugging Face model card) (02/11/2026), Kimi K3.1 Pro License (02/11/2026). Last checked 22/11/2026.

Coding

BenchmarkKimi K3.1 ProClaude Fable 5.1GPT-6 AstraClaude Opus 5.5
Terminal-Bench 4.0Claude Code harness; comparison scores are the maximum reported values65.7%57.9%58.2%66.4%
Terminal-Bench-Science 0.138.1%40%68.1%63.3%
SWE Marathon v1.151.7%45.6%42.5%46.3%
BenchCAD w/o tools72.8%60.6%n/a73%
BenchCAD w/ tools92.7%92.6%95.9%96.2%
Program Bench31.5%53.5%50%65%
DeepSWE v1.175.5%67.4%74.1%74.2%
PostTrain Bench v1.236.5%44.6%41.9%43.8%
Kimi Code Bench 2.0 (internal)Moonshot internal benchmark77.1%n/an/an/a
Kimi Code Bench 3.0 Long-horizon (internal)Moonshot internal benchmark61.2%58.8%66.5%58.5%

Knowledge & science

BenchmarkKimi K3.1 ProClaude Fable 5.1GPT-6 AstraClaude Opus 5.5
HLE-Full w/ tools2,500-question set, with tools70%65.6%57.2%67.7%
BioMysteryBench80.6%81.6%78.9%81.9%
HLE-Full2,500-question set, no tools; GLM-5.3 figure is the text-only subset51.1%57.6%53.6%59.8%
HLE-Diamond1,000-question set; GLM-5.3 figure is the text-only subset39.4%55.1%65.8%60.9%
GPQA-Diamond94.2%93.7%96.1%93.9%

Agents & tool use

BenchmarkKimi K3.1 ProClaude Fable 5.1GPT-6 AstraClaude Opus 5.5
BrowseCompContext compaction at 200K tokens, 500-step maximum, temperature 1.0, top-p 1.092.1%n/a91.5%90.8%
Agents' Last Exam (ALL)Runs that bypass stated requirements score zero43.2%n/a41.8%42.6%
Agents' Last Exam (CLI)Runs that bypass stated requirements score zero44.2%n/a44.2%40.7%
Automation Bench-AA73.3%59.4%68.5%69.5%
GDPval-AA v2.11846 Elo175415731866
AA-Briefcase v1.11678 Elo167215691803
APEX-Agents69.4%68.6%64.7%73.5%
OfficeQA Pro v264.4%54.5%54.4%62.2%
OfficeQA Pro67.4%68.6%66.2%n/a
Frontier Finance58.6%60.8%45%56.7%
Harvey Legal Agent23.7%19.1%n/an/a
HANDBOOK.md44.6%38.8%n/a41.5%
OSWorld V2.179.6%80.7%75.8%81.8%
Business Operation (internal)Moonshot internal benchmark37.8%30.7%32.3%n/a
DECK-Arena (internal)Moonshot internal benchmark1456 Elon/a14231253
Design Bench (internal)Moonshot internal benchmark957 Elon/a1076927

Multimodal

BenchmarkKimi K3.1 ProClaude Fable 5.1GPT-6 AstraClaude Opus 5.5
CharXiv (RQ) w/ pythonTool-augmented, averaged over four runs93.4%92.7%92.8%93.8%
MMMU-Pro w/ pythonTool-augmented, averaged over four runs90%88.4%87.8%89.8%
GDP.pdf w/ pythonTool-augmented, averaged over five runs85.7%83.8%85.5%83.9%
ChartoGraphy w/ pythonTool-augmented, averaged over five runs90.2%88.4%91%89%
BabyVision w/ pythonTool-augmented, averaged over four runs95.4%95.1%96.6%97.7%
ZeroBench_main w/ pythonpass@5; tool-augmented, averaged over five runs63%59%58%61%
GDP.pdf34.6%27.6%34.2%30.6%
CharXiv (RQ)89%88.8%90%92.1%
MMMU-Pro84.1%86.3%86.9%87.7%
ZeroBench_mainavg@5 (pass@5 is 39.0); comparison columns show avg@529.8%23%43.2%35%
PerceptionBench68.3%63%77.6%72.1%

Safety & honesty

BenchmarkKimi K3.1 ProClaude Fable 5.1GPT-6 AstraClaude Opus 5.5
Kimi Hallu Bench Hard w/ tools (internal)Lower is better; Moonshot internal benchmark8.3%23.3%5.5%15%
Kimi Hallu Bench w/o tools (internal)Lower is better; Moonshot internal benchmark5.4%11.4%4.2%6.8%

Other

BenchmarkKimi K3.1 ProClaude Fable 5.1GPT-6 AstraClaude Opus 5.5
ExploitBenchpass@5 maximum-capability score, 300-turn budget74.7%83%100%91%
ExploitGym20.7%30.4%42.4%34.5%
xdev-bench-hard (internal)Moonshot internal benchmark81%n/an/an/a

What did safety testing find for Kimi K3.1 Pro?

Safety level: No vendor safety framework rating on the model card

  • The model card reports cyber capability rows (ExploitBench 74.7, ExploitGym 20.7, both at maximum effort) but no safety framework rating.
  • Moonshot reports internal hallucination benchmarks (Kimi Hallu Bench, lower is better): 5.4 without tools and 8.3 on the hard set with tools.

Independent figures: Artificial Analysis, refreshed daily on our AI model benchmarks page. Full analysis: our Kimi K3.1 Pro deep dive.

Kimi Model Timeline

Kimi K3.1 ProCurrent

1,048,576 tokens (about 1M) context

Specifications

pricing$2/$10 per 1M tokens (input/output)
context Window1,048,576 tokens (about 1M)