--- title: Kimi K3-256k | daily.dev description: Kimi Code offers four model IDs across two model families: Kimi K3 (flagship, 2.8T parameters, up to 1M context) and Kimi K2.7 Code (optimized for code... canonical: https://daily.dev/posts/kimi-k3-256k-tk0yt6bbf twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Kimi K3-256k | daily.dev og:description: Kimi Code offers four model IDs across two model families: Kimi K3 (flagship, 2.8T parameters, up to 1M context) and Kimi K2.7 Code (optimized for code... og:url: https://daily.dev/posts/kimi-k3-256k-tk0yt6bbf og:image: https://api.daily.dev/og/posts/tK0YT6BBf.png og:image:alt: Kimi K3-256k 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 K3-256k **[Hacker News](https://daily.dev/sources/hn)** · 6 min read · 0 upvotes · 0 comments ## Summary Kimi Code offers four model IDs across two model families: Kimi K3 (flagship, 2.8T parameters, up to 1M context) and Kimi K2.7 Code (optimized for code completion). The newer k3-256k variant uses roughly half the quota of the full 1M k3 model. Key guidance covers how to switch models in the official CLI and VS Code extension, how to configure third-party tools via OpenAI or Anthropic-compatible APIs, and how to avoid common pitfalls like cache invalidation on model switches, 401 errors from plan limitations, and reasoning effort overhead. Practical tips include starting a new session when switching models, manually compacting context before switching from k3 to k3-256k, and setting the context window to 1048576 for full K3 capacity in third-party tools. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Community take How the wider developer community reacted, aggregated from 2 discussions and 96 comments across hackernews (as of 2026-07-29). **TL;DR:** The community is broadly positive about Kimi K3-256k as a cheaper, practical option for most coding workflows, with many noting 256k is sufficient for typical use. Side discussions touch on Anthropic reliability issues, Kimi's waitlist capacity crunch, and geopolitical concerns about Chinese AI models. **Sentiment:** 55% positive · 30% mixed · 15% skeptical **The case for** - Half the quota cost of the 1M context version, making it a no-brainer for users who stay under 256k anyway. - Cache is preserved when switching from 256k to 1M, so users can start cheap and upgrade seamlessly. - Many practitioners report 256k is sufficient for most agentic coding sessions, including multi-hour refactoring work. - Being open-weights, the model is available through multiple third-party providers like OpenRouter. **The pushback** - For large refactoring tasks on moderately-sized codebases, 256k can fill up quickly and cause frequent compaction. - Kimi's subscription is currently waitlisted due to capacity constraints from hardware export restrictions. - Some users report Kimi's models have become noticeably slower/dumber recently, with speculation about quantization. - Third-party providers may be secretly quantizing the model, though some providers report their precision level. **By community** - hackernews (positive): Developers largely welcome the 256k variant as a cost-effective default, while debating context management strategies, Anthropic reliability, and geopolitical implications of Chinese AI models. **Hottest debate:** Whether 256k context is genuinely sufficient for real-world agentic coding tasks or causes painful compaction on larger projects. **Open questions** - Is the k3-256k model quantized relative to the full k3-1M, or is it purely a context window restriction? - Will the US restrict Chinese open-weight models and pressure European countries to follow suit? - Does Kimi's efficient context usage stem from model tuning or from the Kimi Code harness design? **Highlights** > My current Claude Code session has been going on for like 35 hours and has used up around 400 million tokens, thankfully almost all of those being cached (95-98%) - pretty typical for long form agentic work. First you spend like 2-3 hours working on a plan, once you have that you just tell the model to go and implement it, do adversarial sub-agent review loops before each commit and also make sure that all tooling and tests pass (including coverage requirements). You do need to poke it in a slightly different direction every few hours, though. Not even any novel work, just some refactoring and SSE notification hardening, bug fixes, alongside environment tuning and getting rid of some bottlenecks (also migrated from Oracle to PostgreSQL but that's mostly done). That said, Kimi somehow manages to use less context in the main thread than Anthropic's models (even when you use sub-agents and also dynamic workflows in Claude Code), might have something to do with either how the model is tuned or their Kimi Code harness - because even in most of the longer form sessions it doesn't seem to fill up quite as quickly (note: because the kimi vis tool doesn't have a full summary view across all agents, these are the main long running agent stats across some sessions, not sub-agents): total tokens cache hit rate wall time peak context 283M 98% 3963m 466k 258M 97% 2724m 467k 98M 94% 1353m 393k 67M 97% 614m 434k 75M 98% 1447m 498k 53M 99% 191m 375k 6M 96% 139m 124k 7M 98% 86m 118k 11M 99% 61m 147k I could see 256k context being sufficient for all sorts of work, even if intermediate progress/plan tracking files and docs might have to be used along the way, in addition to whatever plan support the harness has (for example, if you document something that will be relevant for load testing you might need that in 10 turns but not during the ones before then). > — [KronisLV on hackernews · 3 comments](https://news.ycombinator.com/item?id=49102385) > The article explicitly says "The current version switching from 256k to 1M does not affect the cache." > — [longwave on hackernews](https://news.ycombinator.com/item?id=49103480) > No, the waiting list is true. Kimi had become that popular. I was a subscriber of Kimi back when latest version was Kimi K2. Later I unsubscribed because I jumped over to GLM subscription (they had amazing deal). Now when I wanted to try out Kimi K3 to find out what the fuzz was all about, I couldn’t subscribe to them. I remember reading a post from Moonshot team about this, they are doing this because they are almost at peak capacity and want to reserve it to keep the quality for their current customers. We are actually witnessing an open-weight model catching up at catching mainstream users attention. And instead of behaving like Anthropic, they actually care about their users experience. > — [Alifatisk on hackernews · 1 comments](https://news.ycombinator.com/item?id=49102495) > Not surprising it's a hard cutoff: they almost certainly have two infrastructure configurations for the two max sequence lengths Fewer nodes dedicated to prefill per instance, and fewer nodes in total since you don't need to support a higher KV cache. Disaggregated inference also means they can tune the balance of compute dedicated to prefill seperately from decode > — [BoorishBears on hackernews](https://news.ycombinator.com/item?id=49103105) > This seems functionally similar to OpenAI having a step in pricing once you exceed a certain context length (also at 272k aka 2^18 aka 256k). Having a lot of active context increases the per-token cost (flops issued and bytes read per token out) so it makes sense to pass that cost on to users. I'm actually surprised it's implemented as a hard cutoff instead of a smooth gradient. > — [wren6991 on hackernews · 1 comments](https://news.ycombinator.com/item?id=49102875) **Source threads** - [hackernews](https://news.ycombinator.com/item?id=48935541) · 19 points · 2 comments - [hackernews](https://news.ycombinator.com/item?id=49101852) · 57 points · 94 comments --- Tags: [#llm](https://daily.dev/tags/llm), [#ai-assisted-development](https://daily.dev/tags/ai-assisted-development) [View this post on daily.dev](https://daily.dev/posts/kimi-k3-256k-tk0yt6bbf) ```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 K3-256k","url":"https://daily.dev/posts/kimi-k3-256k-tk0yt6bbf","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/kimi-k3-256k-tk0yt6bbf"},"datePublished":"2026-07-29T19:50:16.996Z","dateModified":"2026-07-29T22:42:01.440Z","description":"Kimi Code offers four model IDs across two model families: Kimi K3 (flagship, 2.8T parameters, up to 1M context) and Kimi K2.7 Code (optimized for code...","image":"https://media.daily.dev/image/upload/s--CxzD6vbw--/f_auto/v1722860399/public/Placeholder%2005","thumbnailUrl":"https://media.daily.dev/image/upload/s--CxzD6vbw--/f_auto/v1722860399/public/Placeholder%2005","isAccessibleForFree":true,"articleSection":"Hacker News","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":"Hacker News","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/hn","url":"https://daily.dev/sources/hn"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/kimi-k3-256k-tk0yt6bbf","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"llm,ai-assisted-development","timeRequired":"PT6M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Hacker News","item":"https://daily.dev/sources/hn"},{"@type":"ListItem","position":3,"name":"Kimi K3-256k"}]} ```