--- title: Meta enters the coding model race, OpenAI merges Codex... description: Meta launched Muse Spark 1.1 via a new paid API at $1.25/$4.25 per million tokens — roughly a quarter of what OpenAI and Anthropic charge — positioning it as a... canonical: https://daily.dev/posts/meta-enters-the-coding-model-race-openai-merges-codex-into-chatgpt-ebrp6n0n1 twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Meta enters the coding model race, OpenAI merges Codex into ChatGPT | daily.dev og:description: Meta launched Muse Spark 1.1 via a new paid API at $1.25/$4.25 per million tokens — roughly a quarter of what OpenAI and Anthropic charge — positioning it as a... og:url: https://daily.dev/posts/meta-enters-the-coding-model-race-openai-merges-codex-into-chatgpt-ebrp6n0n1 og:image: https://api.daily.dev/og/posts/EBRP6n0N1.png og:image:alt: Meta enters the coding model race, OpenAI merges Codex into ChatGPT 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. # Meta enters the coding model race, OpenAI merges Codex into ChatGPT **[Agentic Digest](https://daily.dev/sources/agents_digest)** · 5 min read · 4 upvotes · 1 comments ## Summary Meta launched Muse Spark 1.1 via a new paid API at $1.25/$4.25 per million tokens — roughly a quarter of what OpenAI and Anthropic charge — positioning it as a direct shot at the pricing margins of both labs. OpenAI simultaneously folded Codex into the ChatGPT desktop app alongside a new ChatGPT Work agent targeting Claude Cowork, and shipped GPT-5.6 at the same price as GPT-5.5. Wiz's GhostApproval symlink vulnerability, which was flagged in the previous digest, got broader coverage today with a detailed technical breakdown of the attack chain and confirmation that Anthropic has now shipped a symlink warning after initially rejecting the report. Cursor shipped a native iOS app in public beta and a wave of platform updates including team MCP distribution and a new Customize page. ## Content **TLDR:** Meta launched Muse Spark 1.1 via a new paid API at $1.25/$4.25 per million tokens — roughly a quarter of what OpenAI and Anthropic charge — positioning it as a direct shot at the pricing margins of both labs. OpenAI simultaneously folded Codex into the ChatGPT desktop app alongside a new ChatGPT Work agent targeting Claude Cowork, and shipped GPT-5.6 at the same price as GPT-5.5. Wiz's GhostApproval symlink vulnerability, which was flagged in the previous digest, got broader coverage today with a detailed technical breakdown of the attack chain and confirmation that Anthropic has now shipped a symlink warning after initially rejecting the report. Cursor shipped a native iOS app in public beta and a wave of platform updates including team MCP distribution and a new Customize page. --- ## Meta launches Muse Spark 1.1 with paid API at $1.25/$4.25 per million tokens Meta Superintelligence Labs released Muse Spark 1.1, its first model with a public paid API, priced at $1.25 per million input tokens and $4.25 per million output tokens — with $20 in free credits for new accounts. The model targets agentic and coding tasks, supports a 1M token context window, and reportedly sits near Claude Opus 4.8 and GPT-5.5 on several benchmarks, including a 92.9% score on CyBench versus Claude Opus 4.6's 93%. Zuckerberg framed the aggressive pricing as a deliberate move to pressure what he called the 'extreme' margins of rival labs, and Meta plans an open-source version alongside the proprietary API. Unlike previous Meta AI releases, this is hosted on Meta's own infrastructure via the new Meta Model API — a clear pivot away from the open-weight Llama approach toward a managed inference business. [Read more](https://daily.dev/feed-by-ids?id=H3ju1gqtf&id=WtfJrLEyt&id=6ixVWQUQl&id=iw9cx90OG&id=3b0XBmP1k&id=Yg1xSSLBa&id=adixcIbBT&id=lJ9Drdziy&id=cdX2UuMKz) ## OpenAI merges Codex into ChatGPT, launches ChatGPT Work to compete with Claude Cowork OpenAI folded the Codex desktop app into the ChatGPT desktop app and launched ChatGPT Work, a Codex-powered agentic tool for knowledge workers handling documents, spreadsheets, and third-party integrations like Slack and Google Drive. ChatGPT Work launched on desktop, web, and mobile simultaneously — an advantage over Claude Cowork's initially desktop-only rollout. OpenAI also shipped GPT-5.6 at the same price as GPT-5.5, with Lovable reporting 25% fewer steps and 35–48% fewer tool calls versus the prior model. The unified app also includes a revamped in-app browser with support for authenticated sites, multiple tabs, and file downloads, plus a new Chrome extension that replaces the shuttered Atlas browser. [Read more](https://daily.dev/feed-by-ids?id=kforqo0bT&id=0UhKWeUW3&id=iPFa7Y1Zc&id=x2j6oHVWk&id=ci0ARNfFr&id=v1L8WqbTb&id=GIflvv8NK&id=2mJxv2Z9b&id=wUTzRGE2m) ## GhostApproval symlink attack: technical breakdown and vendor patch status Wiz Research published a detailed technical breakdown of GhostApproval, the symlink attack affecting six major AI coding agents — Claude Code, Amazon Q Developer, Cursor, Windsurf, Augment, and Google Antigravity. The attack chains three weaknesses: prompt injection via README instructions, symlink following without path resolution, and UI misrepresentation in approval dialogs that hides the real target path. The fix is well-understood — resolve every path to its canonical location before opening it, enforce workspace boundaries with something like Linux's openat2() with RESOLVE_BENEATH, and always show the resolved path in confirmation dialogs. AWS, Cursor, and Google have patched; Augment and Windsurf have fixes in progress; Anthropic initially rejected the report but has since shipped a symlink warning. Developers can audit a fresh clone with `find . -type l` before pointing any agent at it. [Read more](https://daily.dev/feed-by-ids?id=kihdn3zbN&id=LpPpux8Vz&id=P5K7J9cPU) ## Cursor ships iOS app, team MCP distribution, and Customize page in version 3.9 Cursor launched a public beta iOS app for all paid plan users, letting developers launch and manage cloud agents from their phones with voice input, push notifications, Live Activities, and the ability to review diffs and merge PRs directly from the app. Separately, Cursor 3.9 introduced a Customize page consolidating plugins, skills, MCPs, subagents, rules, commands, and hooks in one place, plus a marketplace leaderboard. Admins can now configure Team MCP servers once and distribute them across cloud agents, the IDE, and CLI — team members install approved integrations locally without manual server configuration. Faire reported doubling weekly PR throughput and compressing an 18-month migration into a single-engineer effort using Cursor Cloud Agents, while Coinbase reported a 90% reduction in time from idea to production across 2,400+ developers. [Read more](https://daily.dev/feed-by-ids?id=EnZit4J9X&id=1gpygk9kW&id=BJ6RRnQ6j&id=FaC3OrJ9L&id=b78siZe6u&id=pRoafQhd4&id=NkxVJDxsA) --- ## Also notable - **Ollama raises $65M Series B, now at 8.9M monthly developers and inside 85% of Fortune 500:** Theory Ventures led the round, bringing total funding to $88M; the founders previously built Kitematic, which became Docker Desktop, and investors are betting open-weight models will generate the majority of AI tokens within 18–24 months. [Read more](https://daily.dev/feed-by-ids?id=axKQ7nHYj&id=6CuUE7abu) - **Anthropic's Jacobian lens reveals Claude internally representing 'panic' and 'fake' before hallucinating:** The J-lens interpretability technique surfaces words the model is likely to produce in the near future rather than just the next token, and showed those specific words in Claude Opus 4.6's internal state at the exact moment it decided to fabricate a bug it couldn't find. [Read more](https://daily.dev/posts/1ol8Y8pOb) - **Cursor usage data: top 1% of developers generate 30–40K lines per week, AI-accepted commits without review jumped from 10% to ~40% in one month:** The jump correlated with the release of Opus 4.7 and GPT-5.5, and context caching keeps output tokens at just 0.6% of total token usage — without caching, costs would be roughly 10x higher. [Read more](https://daily.dev/posts/i3iJwv0qN) - **OpenAI audited SWE-Bench Pro and found ~30% of its 731 public tasks are broken:** Four failure categories were identified — overly strict tests, underspecified prompts, low-coverage tests, and misleading prompts — and OpenAI is retracting its earlier recommendation to adopt the benchmark for frontier coding evals. [Read more](https://daily.dev/posts/SWouD4dTk) - **Fidji Simo steps down from OpenAI's No. 2 role due to neuroimmune condition relapse:** Simo joined in May 2025 after leading Instacart through its 2023 IPO and will move to a part-time advisory role, leaving a leadership gap as OpenAI eyes a potential IPO; CPO Kevin Weil and CMO Kate Rouch have also recently departed. [Read more](https://daily.dev/posts/jqRsSQ47b) ## Community discussion Top comments from developers on daily.dev. **@bits\_and\_bytes** · 0 upvotes > The Muse Spark pricing is the story here. Quarter of OpenAI and Anthropic pricing means teams that shelved their coding agent ideas because of cost can revisit them now. That's real market pressure. ## Similar posts on daily.dev - [Meta enters the crowded AI coding battle with Muse Spark 1.1](https://daily.dev/posts/meta-enters-the-crowded-ai-coding-battle-with-muse-spark-1-1-wtfjrleyt) · TechCrunch · 1 upvotes · 0 comments --- Tags: [#security](https://daily.dev/tags/security), [#llm](https://daily.dev/tags/llm), [#openai](https://daily.dev/tags/openai), [#ai-coding](https://daily.dev/tags/ai-coding) [View this post on daily.dev](https://daily.dev/posts/meta-enters-the-coding-model-race-openai-merges-codex-into-chatgpt-ebrp6n0n1) ```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/meta-enters-the-coding-model-race-openai-merges-codex-into-chatgpt-ebrp6n0n1","headline":"Meta enters the coding model race, OpenAI merges Codex into ChatGPT","text":"Meta launched Muse Spark 1.1 via a new paid API at $1.25/$4.25 per million tokens — roughly a quarter of what OpenAI and Anthropic charge — positioning it as a direct shot at the pricing margins of both labs. OpenAI simultaneously folded Codex into the ChatGPT desktop app alongside a new ChatGPT Work agent targeting Claude Cowork, and shipped GPT-5.6 at the same price as GPT-5.5. Wiz's GhostApproval symlink vulnerability, which was flagged in the previous digest, got broader coverage today with a detailed technical breakdown of the attack chain and confirmation that Anthropic has now shipped a symlink warning after initially rejecting the report. Cursor shipped a native iOS app in public beta and a wave of platform updates including team MCP distribution and a new Customize page.","url":"https://daily.dev/posts/meta-enters-the-coding-model-race-openai-merges-codex-into-chatgpt-ebrp6n0n1","datePublished":"2026-07-10T04:18:10.777Z","dateModified":"2026-07-10T04:18:30.997Z","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":4},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":1}],"comment":[{"@type":"Comment","text":"The Muse Spark pricing is the story here. Quarter of OpenAI and Anthropic pricing means teams that shelved their coding agent ideas because of cost can revisit them now. That’s real market pressure.","datePublished":"2026-07-14T03:40:39.960Z","url":"https://daily.dev/posts/EBRP6n0N1#c-t48MvLQsO","author":{"@type":"Person","name":"bits_007","url":"https://daily.dev/bits_and_bytes","image":"https://avatars.githubusercontent.com/u/182549945?v=4"}}],"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":"Meta enters the coding model race, OpenAI merges Codex into ChatGPT"}]} ```