xAI ships Grok 4.5

SpaceXAI has shipped Grok 4.5, its latest flagship model, and is pitching it as a stronger option for coding, agentic tasks, and general knowledge work. The release landed on July 8 in the company’s developer docs and product pages, which say the model is available through the xAI API and now powers Grok Build, its terminal-based coding agent. (docs.x.ai) The launch adds another entry to a crowded week of model rollouts, but SpaceXAI’s own materials focus less on spectacle than on product placement. (docs.x.ai) On its homepage, the company labels Grok 4.5 as its “new model,” while its release notes describe it as a model for “coding, agentic tasks, and knowledge work.” (x.ai) ### So what, exactly, did SpaceXAI release? July 8 is the date SpaceXAI lists for Grok 4.5 in its release notes, where it says the model is now available on the xAI API. The same note says developers can tune “reasoning effort” to low, medium, or high, with high as the default. (docs.x.ai) SpaceXAI’s homepage presents Grok 4.5 as the newest addition to a broader product stack spanning text, code, voice, images, and video. The company says that stack is exposed through a unified API. ### Why is the company emphasizing coding and agents? (x.ai) Grok Build is one clue. SpaceXAI says Grok 4.5 now powers that terminal tool, which it describes as “a powerful coding agent and CLI for complex coding work.” Its build documentation adds that developers can also use the same model directly in their own “agent loop, IDE integration, or coding tool.” (docs.x.ai) TechCrunch reported that Elon Musk described the release as an “Opus-class model,” while Reuters said the company called it its most intelligent offering to date and designed it for coding and agentic tasks. (x.ai) Those descriptions match the way SpaceXAI is placing the model across its developer surfaces rather than limiting it to a consumer chatbot. ### What does SpaceXAI say about pricing? SpaceXAI’s release notes put Grok 4.5 at $2 per 1 million input tokens and $6 per 1 million output tokens. (x.ai) The same note says reasoning effort is configurable, which matters because developers using agent-style workflows often trade off latency, cost, and answer depth. Computerworld, citing benchmark estimates, reported that Grok 4.5 was being marketed as cheaper on some coding tasks than rivals including GPT-5.5 in Codex and Claude Code setups, while adding that enterprise buyers would still need real-world testing. (techcrunch.com) ### Where does the SpaceXAI name fit in? SpaceXAI branding now appears across the company’s website, docs, and news pages. The main site identifies the company as “SpaceXAI — Creators of Grok,” and the release notes for Grok 4.5 are published under the same branding. (docs.x.ai) Third-party coverage tied the launch to the company’s broader rebrand and post-listing product push, but the directly verifiable point from official materials is simpler: Grok 4.5 shipped under the SpaceXAI name, and the company is distributing it through both developer APIs and Grok Build. (computerworld.com) ### What should developers watch next? SpaceXAI’s documentation says Grok 4.5 is already live in the API and in Grok Build, which means the next evidence will come from usage, pricing comparisons, and benchmark results published by developers and independent evaluators. (x.ai) Artificial Analysis said Grok 4.5 placed fourth on its Intelligence Index and performed near GPT-5.5 in its coding-agent index at lower cost, though those are third-party measurements rather than company claims. (docs.x.ai) The company’s release-notes page is the clearest place to watch for follow-up changes. That page already lists model updates, pricing changes, and agent-tool compatibility updates, and now includes the July 8 Grok 4.5 entry. (docs.x.ai) (artificialanalysis.ai)