--- title: Kimi K3.1 Pro: Pricing, Specs, a Cyber Restriction, and... description: Moonshot AI released Kimi K3.1 Pro, a mixture-of-experts model with 2.8T total parameters, 104B active, and a 1M-token context window. Moonshot's own... canonical: https://daily.dev/posts/kimi-k3-1-pro-pricing-specs-a-cyber-restriction-and-the-verbosity-test-n3h66a1k1 twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Kimi K3.1 Pro: Pricing, Specs, a Cyber Restriction, and the Verbosity Test | daily.dev og:description: Moonshot AI released Kimi K3.1 Pro, a mixture-of-experts model with 2.8T total parameters, 104B active, and a 1M-token context window. Moonshot's own... og:url: https://daily.dev/posts/kimi-k3-1-pro-pricing-specs-a-cyber-restriction-and-the-verbosity-test-n3h66a1k1 og:image: https://api.daily.dev/og/posts/N3H66a1K1.png og:image:alt: Kimi K3.1 Pro: Pricing, Specs, a Cyber Restriction, and the Verbosity Test og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Kimi K3.1 Pro: Pricing, Specs, a Cyber Restriction, and the Verbosity Test **[Collections](https://daily.dev/sources/collections)** · 2 min read · 1 upvotes · 0 comments ## Summary Moonshot AI released Kimi K3.1 Pro, a mixture-of-experts model with 2.8T total parameters, 104B active, and a 1M-token context window. Moonshot's own evaluations claim it beats every listed US frontier model on 6 of 14 benchmarks and matches or outperforms Claude Opus 5.5 and GPT-6 Astra on several tasks. API pricing is $2 per million input tokens and $10 per million output, half the price of Opus 5.5 and 80% below GPT-6 Astra, though DeepSeek-V4.1-Flash-1017 is up to 17x cheaper. Moonshot deliberately removed cyber-related training data to limit the model's cyber capabilities. A 'neural yap benchmark' measuring verbosity ranks it chattier than GPT and Claude 5.5 models but less than Claude Opus 5. ## Content Moonshot AI has released Kimi K3.1 Pro. Moonshot's own evaluations say it beats every listed US frontier model on 6 of 14 benchmarks, and it matches Claude Opus 5.5 and GPT-6 Astra on several others. Arena found it outperforms all non-OpenAI/Anthropic models on agentic tasks, with the exceptions of Gemini 4 Argon and Muse Wonder Full. It's also strong at 3D modeling. ## Specs Kimi K3.1 Pro is a mixture-of-experts model with 2.8T total parameters and 104B active, and a 1M-token context window. It's natively multimodal. Moonshot says it can build complete game projects in Unreal, Unity and Godot, do parametric 3D and CAD work in Blender and CAD software, and cut hallucinations by 65% on financial modeling tasks. It's available now on Kimi.ai, Kimi Work and Kimi Code, plus a Highspeed version that runs 2-3x faster. The API is at platform.kimi.ai. ## Pricing The API costs $2 per million input tokens and $10 per million output tokens. That's half of Opus 5.5 ($4/$20) and 80% below GPT-6 Astra ($10/$50). It's still well above other Chinese models, though: DeepSeek-V4.1-Flash-1017 is up to about 17x cheaper per token. ## Open weights and the cyber carve-out Moonshot said full weights would be out by Wednesday, November 4, and @scaling01 now reports Kimi-K3.1-Pro is open-weights. There's a notable restriction behind the release. For the first time, Moonshot removed cyber-related data from mid-training and RL to deliberately weaken the model's cyber capabilities. The company committed "not to release open weights that would extend the absolute frontier of cyber capability." ## It doesn't yap much A custom "neural yap benchmark" measures verbosity rather than intelligence. It scores four things: stock phrases ("you're totally right"), typography like em dashes and emojis, formatting such as markdown essays, and response length. GPT models yap the least, then Claude 5.5, then Kimi K3.1 Pro. Claude Opus 5 yaps the most, writing long essay-style answers whatever the prompt. ## Questions this post answers ### How much does the Kimi K3.1 Pro API cost compared to Claude Opus 5.5 and GPT-6 Astra? Kimi K3.1 Pro costs $2 per million input tokens and $10 per million output tokens. That's half the price of Claude Opus 5.5 ($4/$20) and 80% below GPT-6 Astra ($10/$50). It's still pricier than some Chinese competitors, since DeepSeek-V4.1-Flash-1017 runs up to about 17 times cheaper per token. _Comparing model costs before committing to an API is easier when tracked on daily.dev._ ### What are the specs of Moonshot AI's Kimi K3.1 Pro model? Kimi K3.1 Pro is a mixture-of-experts model with 2.8 trillion total parameters, 104 billion of them active, and a 1 million-token context window. It is natively multimodal and can build full game projects in Unreal, Unity and Godot, do parametric 3D and CAD work in Blender, and reduce hallucinations by 65% in financial modeling tasks. _Developers picking between frontier models for agentic work can follow spec comparisons like this on daily.dev._ ### Why did Moonshot AI limit the cyber capabilities of Kimi K3.1 Pro? Moonshot deliberately removed cyber-related data from mid-training and reinforcement learning to limit the model's cyber capabilities, marking the first time it has done so. The company committed to not releasing open weights that would extend the absolute frontier of cyber capability. _Anyone weighing the safety trade-offs of open-weight models can follow these policy shifts on daily.dev._ ## Similar posts on daily.dev - [Everyone Read Kimi K3 as a Price War. It Is the Opposite.](https://daily.dev/posts/everyone-read-kimi-k3-as-a-price-war-it-is-the-opposite--xshhpdjyb) · Medium · 3 upvotes · 0 comments --- Tags: [#ai](https://daily.dev/tags/ai), [#llm](https://daily.dev/tags/llm), [#moonshot-ai](https://daily.dev/tags/moonshot-ai) [View this post on daily.dev](https://daily.dev/posts/kimi-k3-1-pro-pricing-specs-a-cyber-restriction-and-the-verbosity-test-n3h66a1k1) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Kimi K3.1 Pro: Pricing, Specs, a Cyber Restriction, and the Verbosity Test","url":"https://daily.dev/posts/kimi-k3-1-pro-pricing-specs-a-cyber-restriction-and-the-verbosity-test-n3h66a1k1","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/kimi-k3-1-pro-pricing-specs-a-cyber-restriction-and-the-verbosity-test-n3h66a1k1"},"datePublished":"2026-11-01T18:53:37.293Z","dateModified":"2026-11-04T15:34:14.539Z","description":"Moonshot AI released Kimi K3.1 Pro, a mixture-of-experts model with 2.8T total parameters, 104B active, and a 1M-token context window. Moonshot's own...","isAccessibleForFree":true,"articleSection":"Collections","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Collections","logo":"https://media.daily.dev/image/upload/s--fk_6ycEi--/f_auto,q_auto/v1780996001/logos/collections?_a=BAMAMiWQ0","url":"https://daily.dev/sources/collections"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/kimi-k3-1-pro-pricing-specs-a-cyber-restriction-and-the-verbosity-test-n3h66a1k1","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"ai,llm,moonshot-ai","timeRequired":"PT2M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Collections","item":"https://daily.dev/sources/collections"},{"@type":"ListItem","position":3,"name":"Kimi K3.1 Pro: Pricing, Specs, a Cyber Restriction, and the Verbosity Test"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/kimi-k3-1-pro-pricing-specs-a-cyber-restriction-and-the-verbosity-test-n3h66a1k1#faq","mainEntity":[{"@type":"Question","name":"How much does the Kimi K3.1 Pro API cost compared to Claude Opus 5.5 and GPT-6 Astra?","acceptedAnswer":{"@type":"Answer","text":"Kimi K3.1 Pro costs $2 per million input tokens and $10 per million output tokens. That's half the price of Claude Opus 5.5 ($4/$20) and 80% below GPT-6 Astra ($10/$50). It's still pricier than some Chinese competitors, since DeepSeek-V4.1-Flash-1017 runs up to about 17 times cheaper per token. Comparing model costs before committing to an API is easier when tracked on daily.dev."}},{"@type":"Question","name":"What are the specs of Moonshot AI's Kimi K3.1 Pro model?","acceptedAnswer":{"@type":"Answer","text":"Kimi K3.1 Pro is a mixture-of-experts model with 2.8 trillion total parameters, 104 billion of them active, and a 1 million-token context window. It is natively multimodal and can build full game projects in Unreal, Unity and Godot, do parametric 3D and CAD work in Blender, and reduce hallucinations by 65% in financial modeling tasks. Developers picking between frontier models for agentic work can follow spec comparisons like this on daily.dev."}},{"@type":"Question","name":"Why did Moonshot AI limit the cyber capabilities of Kimi K3.1 Pro?","acceptedAnswer":{"@type":"Answer","text":"Moonshot deliberately removed cyber-related data from mid-training and reinforcement learning to limit the model's cyber capabilities, marking the first time it has done so. The company committed to not releasing open weights that would extend the absolute frontier of cyber capability. Anyone weighing the safety trade-offs of open-weight models can follow these policy shifts on daily.dev."}}]} ```