--- title: OpenAI and Anthropic cut their LLM prices by up to 80%... description: OpenAI and Anthropic have cut LLM API prices by up to 80% since June 2026, and the piece explains why. GPT-5.6 Luna dropped 80% in July, and Opus 5.5 later cut... canonical: https://daily.dev/posts/openai-and-anthropic-cut-their-llm-prices-by-up-to-80-within-a-few-months-why--qwjt9m0rl twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: OpenAI and Anthropic cut their LLM prices by up to 80% within a few months. Why? | daily.dev og:description: OpenAI and Anthropic have cut LLM API prices by up to 80% since June 2026, and the piece explains why. GPT-5.6 Luna dropped 80% in July, and Opus 5.5 later cut... og:url: https://daily.dev/posts/openai-and-anthropic-cut-their-llm-prices-by-up-to-80-within-a-few-months-why--qwjt9m0rl og:image: https://api.daily.dev/og/posts/QwJT9M0rL.png og:image:alt: OpenAI and Anthropic cut their LLM prices by up to 80% within a few months. Why? 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. # OpenAI and Anthropic cut their LLM prices by up to 80% within a few months. Why? **[Max Woolf's Blog](https://daily.dev/sources/minimaxir)** · 12 min read · 2 upvotes · 0 comments ## Summary OpenAI and Anthropic have cut LLM API prices by up to 80% since June 2026, and the piece explains why. GPT-5.6 Luna dropped 80% in July, and Opus 5.5 later cut cache read costs so deeply that total cost fell 40-60% versus Opus 5. Since cache reads make up 96-99% of agentic coding token spend, cutting cache pricing matters far more than cutting input/output rates. Open-weight competitors like DeepSeek and Moonshot's Kimi K3.1 Pro, available from dozens of OpenRouter providers, keep pricing pressure high. Both companies are also nearing IPOs, which raises questions about their inference profit margins, reportedly as high as 80% in some reports. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### Why do cache read token costs matter more than input/output pricing for AI coding agents? Because cache reads make up 96% to 99% of total token expenditure in agentic coding, since the cache holds the entire prior conversation replayed every turn. A cache hit rate of 90-100% means a 75% cut in cache read price, like Claude Fable 5.1's cut from $1 to $0.25 per million tokens, translates into a 70-75% reduction in total usage cost, far more impactful than cutting input or output rates. _Anyone sizing an agentic coding budget should watch cache pricing moves like these on daily.dev._ ### How much did Claude Opus 5.5 pricing drop compared to Opus 5? Claude Opus 5.5 is priced at $4 input / $20 output / $0.20 cache read per million tokens, down from Opus 5's $5 / $25 / $0.50. Anthropic stated Opus 5.5 requires less compute to serve than Opus 5 and its pricing reflects that. Combined with a larger cache read cut, total cost fell 40-60% for a model rated better than Claude Fable 5.1. _Developers comparing Claude model tiers can track pricing shifts like this through daily.dev._ ### Why are open-weight models like DeepSeek and Kimi putting pricing pressure on OpenAI and Anthropic? Open-weight models like DeepSeek Flash V4.1 and Moonshot AI's Kimi K3.1 Pro are hosted by dozens of competing inference providers (29 for DeepSeek Flash V4.1 1017 on OpenRouter), which drives prices toward marginal cost. Kimi K3.1 Pro matches Opus 5.5's performance closely (55 vs 58 on the Artificial Analysis index) at half the input/output price, priced at $2 / $10 / $0.20. _Teams weighing open-weight versus closed models can keep tabs on these price wars via daily.dev._ ## Community take How the wider developer community reacted, aggregated from 1 discussion and 150 comments across hackernews (as of 2026-10-26). **TL;DR:** Discussion largely frames the price cuts as a competitive response to cheap Chinese open-weight models (Kimi, DeepSeek, GLM), with heavy debate over whether the labs are sustainable, whether regulation/bans on open models are coming, and whether 'unlimited' token plans are realistic. **Sentiment:** 15% positive · 40% mixed · 45% skeptical **The case for** - Cheaper open-weight alternatives like Kimi and GLM let some users avoid 'cybersecurity' refusals and get work done more cheaply. - Some argue token costs will keep falling due to efficiency and throughput gains, eventually enabling near-unlimited usage plans similar to mobile data. **The pushback** - Many doubt the price war is sustainable and expect it to end in financial trouble or a bubble pop for frontier labs. - Skepticism that frontier labs are profitable at all, with disagreement over whether third-party Chinese model hosts are genuinely profitable or just gimping capability. - Concern that incumbents are pushing for AI regulation partly to get Chinese open-weight competitors banned and protect margins. - Doubt that a US ban on Chinese open models would be enforceable or would meaningfully stop usage. **By community** - hackernews (heated): Sprawling, contentious thread covering competition dynamics, possible bans on open-weight Chinese models, lab profitability, and regulation, with no clear consensus. **Hottest debate:** Whether frontier labs' economics are sustainable (and genuinely competition-driven) versus whether this is unsustainable price-war behavior headed for a bubble burst or regulatory capture via banning cheaper Chinese open-weight models. **Open questions** - Are OpenAI and Anthropic actually profitable on inference, or just burning cash to gain share? - Could the US realistically ban or restrict open-weight Chinese models, and would it even be enforceable? - Will token pricing eventually converge to something like 'unlimited' consumer plans the way mobile data did? **Highlights** > That's why Anthropic are rushing the IPO. You can explain a single quarter slowdown, you can not explain a multi quarter one. > — [kshacker on hackernews · 1 comments](https://news.ycombinator.com/item?id=50237116) > This is why Anthropic and OpenAI are pushing for more “regulation”. You can’t show ROI to investors by cutting inference costs. Regulation might inconvenience them a little, but the upside is huge if it means open models out of China get banned. > — [redrix on hackernews · 5 comments](https://news.ycombinator.com/item?id=50236694) > You are woefully misinformed and objectively incorrect. Read less Zitron. The American companies are tens of billions in the hole, but there are many, many smaller labs who are not, who make a killing on profit that absolutely covers R&D. I work in one. Kimi and DS both have >1 trillion param models. > — [anonymous908213 on hackernews](https://news.ycombinator.com/item?id=50237304) > Not DeepSeek themselves, but their open models are run by third-party companies in a profitable way. The unsustainable interference theory has been debunked. > — [holoduke on hackernews · 2 comments](https://news.ycombinator.com/item?id=50236817) > It won’t be a ban, it will be some “certified safe” badge from a federal agency. Companies will be scared into not using anything that is not “certified safe” and business insurance will even require it. Whoever (openAI or Anthropic) gets to the politicians first will be the only certified safe model provider. Maybe the FDA is a good analogy, you wouldn’t want to buy meat at the store not FDA certified would you? OpenAI and Anthropic want you to feel the same about AI. > — [chasd00 on hackernews · 1 comments](https://news.ycombinator.com/item?id=50237073) **Source threads** - [hackernews](https://news.ycombinator.com/item?id=50236102) · 111 points · 150 comments --- Tags: [#llm](https://daily.dev/tags/llm), [#openai](https://daily.dev/tags/openai), [#ai-coding](https://daily.dev/tags/ai-coding), [#anthropic](https://daily.dev/tags/anthropic), [#deepseek](https://daily.dev/tags/deepseek) [View this post on daily.dev](https://daily.dev/posts/openai-and-anthropic-cut-their-llm-prices-by-up-to-80-within-a-few-months-why--qwjt9m0rl) ```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":"OpenAI and Anthropic cut their LLM prices by up to 80% within a few months. Why?","url":"https://daily.dev/posts/openai-and-anthropic-cut-their-llm-prices-by-up-to-80-within-a-few-months-why--qwjt9m0rl","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/openai-and-anthropic-cut-their-llm-prices-by-up-to-80-within-a-few-months-why--qwjt9m0rl"},"datePublished":"2026-10-25T21:08:34.985Z","dateModified":"2026-10-26T01:16:18.967Z","description":"OpenAI and Anthropic have cut LLM API prices by up to 80% since June 2026, and the piece explains why. GPT-5.6 Luna dropped 80% in July, and Opus 5.5 later cut...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/be1c8056a162c72c2033c4ca6b1697d2?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/be1c8056a162c72c2033c4ca6b1697d2?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Max Woolf's Blog","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":"Max Woolf's Blog","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/a8167781e8344149a1ff10c9cd469144","url":"https://daily.dev/sources/minimaxir"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/openai-and-anthropic-cut-their-llm-prices-by-up-to-80-within-a-few-months-why--qwjt9m0rl","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":2},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"llm,openai,ai-coding,anthropic,deepseek","timeRequired":"PT12M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Max Woolf's Blog","item":"https://daily.dev/sources/minimaxir"},{"@type":"ListItem","position":3,"name":"OpenAI and Anthropic cut their LLM prices by up to 80% within a few months. Why?"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/openai-and-anthropic-cut-their-llm-prices-by-up-to-80-within-a-few-months-why--qwjt9m0rl#faq","mainEntity":[{"@type":"Question","name":"Why do cache read token costs matter more than input/output pricing for AI coding agents?","acceptedAnswer":{"@type":"Answer","text":"Because cache reads make up 96% to 99% of total token expenditure in agentic coding, since the cache holds the entire prior conversation replayed every turn. A cache hit rate of 90-100% means a 75% cut in cache read price, like Claude Fable 5.1's cut from $1 to $0.25 per million tokens, translates into a 70-75% reduction in total usage cost, far more impactful than cutting input or output rates. Anyone sizing an agentic coding budget should watch cache pricing moves like these on daily.dev."}},{"@type":"Question","name":"How much did Claude Opus 5.5 pricing drop compared to Opus 5?","acceptedAnswer":{"@type":"Answer","text":"Claude Opus 5.5 is priced at $4 input / $20 output / $0.20 cache read per million tokens, down from Opus 5's $5 / $25 / $0.50. Anthropic stated Opus 5.5 requires less compute to serve than Opus 5 and its pricing reflects that. Combined with a larger cache read cut, total cost fell 40-60% for a model rated better than Claude Fable 5.1. Developers comparing Claude model tiers can track pricing shifts like this through daily.dev."}},{"@type":"Question","name":"Why are open-weight models like DeepSeek and Kimi putting pricing pressure on OpenAI and Anthropic?","acceptedAnswer":{"@type":"Answer","text":"Open-weight models like DeepSeek Flash V4.1 and Moonshot AI's Kimi K3.1 Pro are hosted by dozens of competing inference providers (29 for DeepSeek Flash V4.1 1017 on OpenRouter), which drives prices toward marginal cost. Kimi K3.1 Pro matches Opus 5.5's performance closely (55 vs 58 on the Artificial Analysis index) at half the input/output price, priced at $2 / $10 / $0.20. Teams weighing open-weight versus closed models can keep tabs on these price wars via daily.dev."}}]} ```