--- title: Frontier models get cheap fast, Claude Opus 5 splits the... description: DeepSeek V4 Pro and Grok 4.6 both hit general availability today, and both are forcing a conversation about price-to-performance that Opus 5's launch made... canonical: https://daily.dev/posts/frontier-models-get-cheap-fast-claude-opus-5-splits-the-room-nfcdjol2x twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Frontier models get cheap fast, Claude Opus 5 splits the room | daily.dev og:description: DeepSeek V4 Pro and Grok 4.6 both hit general availability today, and both are forcing a conversation about price-to-performance that Opus 5's launch made... og:url: https://daily.dev/posts/frontier-models-get-cheap-fast-claude-opus-5-splits-the-room-nfcdjol2x og:image: https://api.daily.dev/og/posts/nFCdjoL2x.png og:image:alt: Frontier models get cheap fast, Claude Opus 5 splits the room 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. # Frontier models get cheap fast, Claude Opus 5 splits the room **[Agentic Digest](https://daily.dev/sources/agents_digest)** · 5 min read · 7 upvotes · 0 comments ## Summary DeepSeek V4 Pro and Grok 4.6 both hit general availability today, and both are forcing a conversation about price-to-performance that Opus 5's launch made unavoidable. Opus 5 keeps last generation's pricing but the real story is that most people are prompting it wrong, carrying over habits from older Claude models. Lovable tripled revenue and raised at $13.3B, proof vibe-coding platforms are finding real usage, not just hype. Meanwhile Oracle just told OpenJDK contributors that AI-written code isn't trustworthy enough to submit, even as its own leadership brags about AI writing Oracle's code. ## Content **TLDR:** DeepSeek V4 Pro and Grok 4.6 both hit general availability today, and both are forcing a conversation about price-to-performance that Opus 5's launch made unavoidable. Opus 5 keeps last generation's pricing but the real story is that most people are prompting it wrong, carrying over habits from older Claude models. Lovable tripled revenue and raised at $13.3B, proof vibe-coding platforms are finding real usage, not just hype. Meanwhile Oracle just told OpenJDK contributors that AI-written code isn't trustworthy enough to submit, even as its own leadership brags about AI writing Oracle's code. --- ## DeepSeek V4 Pro and Grok 4.6 ship the same day, both undercutting Claude on price DeepSeek V4 Pro hit general availability at $0.435/$0.87 per million input/output tokens, a 1M context window, and near-Claude scores on Terminal-Bench 2.1 and Cyber Gym — roughly 57x cheaper than Claude's $10.50/$50. Grok 4.6 launched alongside it at $2/$6 per million tokens, ties GPT-5.6 Sol on the Artificial Analysis Intelligence Index, and rolled out same-day across Cursor, Devin, Amp, and half a dozen other coding platforms. DeepSeek's early traffic is skewing toward roleplay apps like SillyTavern rather than enterprise workloads, which is a strange look for a launch this significant. Box CEO Aaron Levie framed the whole thing as Jevons paradox: cheaper AI doesn't shrink spend, it expands what people are willing to ask agents to do. [Read more](https://daily.dev/feed-by-ids?id=bdIjyqRT6&id=QPe7mJfgy&id=EEgTXIwON&id=SlOYIFyKq&id=oMFKRX9kJ&id=tcnyTkoiZ) ## Claude Opus 5 splits opinion, and the real story is how people are prompting it Opus 5 kept Opus 4's pricing ($15/$75 per million tokens) but the reasoning and coding gains are apparently real enough that one benchmark watcher says it "destroys all other models" on dynamic-game tests like ARC-AGI-3. The more useful thread going around: people are getting mediocre results because they're dragging over old habits — deleting all your skills, MCPs, and CLAUDE.md files and just stating the outcome you want reportedly produces a dramatic jump in quality. A developer-focused breakdown adds a routing rule worth remembering: Opus earns its premium on security audits and multi-file refactoring, but Sonnet 4 gets you 90-95% of the quality for a fifth of the cost on routine generation and docs. Its 200K context window is also now a real liability against GPT-4.1 and Gemini 2.5 Pro's 1M. [Read more](https://daily.dev/posts/JN3aqNm0M) ## Lovable triples revenue, hits $13.3B valuation Lovable raised $400M at a $13.3B valuation, more than double its December number, with Menlo Ventures and an EU-backed Scaleup Europe Fund co-leading. Revenue is nearing a $600M run rate, up from $500M in June, and the platform has produced over 60 million projects with about 900 million monthly visits to apps built on it. The company is also running automatic security scanning and picked up AIUC-1 certification after past criticism of AI-generated code reliability — the unglamorous work that matters once you're powering internal tools at Fortune 500 companies rather than side projects. [Read more](https://daily.dev/posts/848uEbsL5) ## Oracle bans AI-written code from OpenJDK while its own leadership brags about using it internally Oracle told OpenJDK contributors they can't submit AI-generated code, citing safety, security, and unresolved IP questions — you can use AI privately to debug or research, but the moment a model writes the actual patch, it's off-limits. The tension: Larry Ellison has claimed AI models now write Oracle's own code, and Oracle is spending $70B on datacenter expansion this year, aggressive enough that S&P downgraded its credit rating to one notch above junk. It's not really hypocrisy — it's an honest admission that generating code has gotten cheap, but understanding whether it's secure enough to put your name on is still the hard, unsolved part. [Read more](https://daily.dev/posts/iQhm7lfl9) --- ## Also notable - **A 16-year-old SQLite bug found in 15 minutes by an AI-driven fuzzer:** Tailscale spent months chasing a WAL-Reset data race in SQLite (now fixed in 3.51.3) using custom tooling, while a separate developer reproduced the same bug in 15 minutes using Antithesis paired with Claude agent skills writing the test harness. [Read more](https://daily.dev/posts/qBIKVYoyY) - **Cognition reportedly in talks to raise at $40B, just three months after $26B:** Cognition, maker of Devin, is in talks for a new round tied to hitting a $1B annualized revenue run rate, up from the $492M run rate CEO Scott Wu cited in May with enterprise usage growing 50% month-over-month. [Read more](https://daily.dev/posts/CxpOuh2yW) - **CodeRabbit raises $143M at $1.5B, launches PR triage and blast-radius analysis:** CodeRabbit's new Agentic Change Management suite adds PR triage by risk/value, change-impact analysis beyond plain diffs, and continuous production security scanning, alongside a $143M raise at a $1.5B valuation. [Read more](https://daily.dev/posts/tjUknTxJF) - **Survey: 80% of orgs traced a production incident to AI-generated code, but still call it a win:** A Wakefield Research survey of 400 executives found 80% traced a production incident to AI code in the past year, yet 89% report positive ROI and only 6% regularly test AI-generated code with AI tools. [Read more](https://daily.dev/posts/M8KuTLAGx) - **Blacksmith's valuation jumps almost 10x to $550M in under a year:** The CI infrastructure and AI code-testing startup raised a $45M Series B at $550M, up from $60M less than a year ago, growing from 700 to over 5,000 customers including Mercury and Supabase. [Read more](https://daily.dev/posts/BBBkrwprm) --- Tags: [#llm](https://daily.dev/tags/llm), [#ai-coding](https://daily.dev/tags/ai-coding), [#grok](https://daily.dev/tags/grok), [#deepseek](https://daily.dev/tags/deepseek) [View this post on daily.dev](https://daily.dev/posts/frontier-models-get-cheap-fast-claude-opus-5-splits-the-room-nfcdjol2x) ```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":"DiscussionForumPosting","mainEntityOfPage":"https://daily.dev/posts/frontier-models-get-cheap-fast-claude-opus-5-splits-the-room-nfcdjol2x","headline":"Frontier models get cheap fast, Claude Opus 5 splits the room","text":"DeepSeek V4 Pro and Grok 4.6 both hit general availability today, and both are forcing a conversation about price-to-performance that Opus 5's launch made unavoidable. Opus 5 keeps last generation's pricing but the real story is that most people are prompting it wrong, carrying over habits from older Claude models. Lovable tripled revenue and raised at $13.3B, proof vibe-coding platforms are finding real usage, not just hype. Meanwhile Oracle just told OpenJDK contributors that AI-written code isn't trustworthy enough to submit, even as its own leadership brags about AI writing Oracle's code.","url":"https://daily.dev/posts/frontier-models-get-cheap-fast-claude-opus-5-splits-the-room-nfcdjol2x","datePublished":"2026-08-13T04:18:12.454Z","dateModified":"2026-09-13T21:16:16.684Z","author":{"@type":"Organization","name":"Agentic Digest","logo":"https://media.daily.dev/image/upload/s--V91DY4ls--/f_auto,q_auto/v1772617267/logos/agents_digest","url":"https://daily.dev/sources/agents_digest"},"interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":7},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"isPartOf":{"@type":"WebPage","url":"https://daily.dev/sources/agents_digest","name":"Agentic Digest"}} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Agentic Digest","item":"https://daily.dev/sources/agents_digest"},{"@type":"ListItem","position":3,"name":"Frontier models get cheap fast, Claude Opus 5 splits the room"}]} ```