Grok Build 0.1

On May 29, 2026, xAI announced that grok-build-0.1, its fastest coding model, became available through the xAI API in public beta. The page explicitly identifies the version as "grok-build-0.1" and describes it as a coding model trained for agentic coding tasks including web development, debugging, and MCP support, ser The announcement states the model accepts text and image inputs and produces text output, is the same model that powers the Grok Build CLI, and is recommended for use inside agentic harnesses including Grok Build, Cursor, Hermes Agent, OpenClaw, Kilo Code, and OpenCode. It also notes that the model is available via Ope

The Grok Build CLI changelog for v1.0.13, dated August 28, 2026, lists fixes and improvements shipping to Grok Build. Recent changes include length-truncated responses continuing automatically instead of failing, hooks being able to ask users to confirm a tool call, live pixel previews for pasted images on iTerm2, auto Additional v1.0.13 fixes cover Windows users correctly opening /.grok and worktree sessions, more reliable session data saving, truncated tool calls executing when arguments are complete, and faster MCP server startup when auth is already configured. Earlier entries in the changelog (v1.0.12) address MCP retry behavior

An independent DevOps.com article (June 1, 2026) covers xAI opening Grok Build 0.1 to developers via the xAI API, confirming the exact version name "Grok Build 0.1" and reiterating that the model is now reachable without a SuperGrok or X Premium+ subscription, broadening access beyond the Grok Build CLI. The article de The piece reports that the model is served at 100+ tokens per second at $1 per million input tokens and $2 per million output tokens, and positions Grok Build 0.1 as a tool for building AI-powered applications rather than only driving a single CLI product. It frames the API release as a meaningful expansion of xAI's fo

Tech Jack Solutions reported on May 31, 2026 that xAI's Grok Build 0.1 API public beta became active on May 29, 2026, roughly two weeks after the earlier Grok Build CLI beta. The model targets agentic software engineering tasks and is reportedly accessible via xAI's own API, OpenRouter, and the Vercel AI Gateway, broad The brief flags that the xAI primary blog returned a 404 at time of reporting, leaving details such as the 256K context window, MCP server optimization, and pricing as vendor-reported and unverified from an accessible primary source. The CLI access path is reported to require a SuperGrok Heavy subscription around $299

The Kilo Code model page for SpaceXAI: Grok Build 0.1 documents a 256,000-token context window, a 230,400-token maximum output, and input modalities, with input cost of $1.00 per million tokens routed via OpenRouter. The page confirms the model as SpaceXAI's fast coding model trained for agentic software engineering wo Benchmark figures on the page include Kilo Bench Terminal Bench 2.0 at 50.6% completion with $30.70 average cost per attempt, and PinchBench OpenClaw runs at an average score of 88.9% across 50 official models, average runtime of 220m 40s over 5 runs, and $20.581 average cost per benchmark run. Category breakdowns show

OpenRouter's date-pinned snapshot dated May 20, 2026 for x-ai/grok-build-0.1 confirms the same core specs as the canonical listing: a fast coding model from SpaceXAI for agentic software engineering, with text and image inputs, text output, a 256K context window, and no text output limit. It is currently in early acces Listings match the $1/$2 per million token pricing with a $0.20 cache read rate, hosted across SpaceXAI, SpaceXAI (ZDR), SpaceXAI Priority, and SpaceXAI Priority (ZDR). Best provider P50 latency is 0.31s with up to 91 tok/s throughput, and weighted-average effective prices around $0.28 per million input tokens and $2.0

OpenRouter's canonical listing for x-ai/grok-build-0.1 describes the model as SpaceXAI's fast coding model trained for agentic software engineering workflows, supporting text and image inputs with text output and optimized for tool use and multi-step development tasks. It powers xAI's Grok Build CLI and features a 256K Listed pricing on OpenRouter is $1 per million input tokens and $2 per million output tokens, with a $0.20 per million cache read rate. Providers include SpaceXAI, SpaceXAI (ZDR), SpaceXAI Priority, and SpaceXAI Priority (ZDR), with SpaceXAI (ZDR) showing the best P50 latency at 0.31s and 84 tok/s throughput. Effective