On July 8, 2026, xAI announced Grok Build 0.1 and opened the agentic coding API to all developers. The release gave developers access to multi-file edits, terminal command execution, and browser automation through a single API endpoint. It was xAI’s first dedicated coding agent product, following a series of free tier changes for its Grok chat models. The announcement came less than a month after major coding tool price shifts from OpenAI, Anthropic, and Google. xAI positioned Grok Build 0.1 as a lower cost alternative for teams that had been priced out of other coding agents.
Developers on the free tier received 100 API requests per day with a 128,000 token context window. Paid usage started at $0.40 per million input tokens and $2.00 per million output tokens. The limits applied per developer account, not per project. Larger teams could request higher rate limits through xAI support. Existing xAI API customers kept their current billing but had to enable Grok Build separately in the console. The pricing did not beat Google’s Gemini Code Assist standard rate of $0.35 per million input tokens, but it bundled browser and terminal calls that Google billed through Cloud setup. For more details on free API limits, see AI API free tiers limits 2026.
Why it matters: the agentic coding market shifted sharply in June 2026. OpenAI launched a free tier for Codex, Anthropic ended its flat-rate agent subsidy, and Google cut Gemini API prices. xAI entered that field as a price challenger. The company’s $0.40 input price was 20 percent below OpenAI Codex list pricing at launch. For a development team sending 100 million tokens per day, that difference saved about $300 per month on input alone. Grok Build 0.1 also bundled browser and terminal tool calls into the base API cost, while some rivals billed them separately. See ChatGPT Codex free tier and Anthropic ends agent subsidy.
xAI’s official Grok Build 0.1 announcement described the API as a preview release with rate limits and limited region availability. The company said it would monitor tool call success rates and adjust pricing before the 1.0 release. Developers who signed up before August 1, 2026 received a $50 usage credit. The API required a valid phone number and a payment method on file, even for free tier access. Independent analysis from Stanford HAI noted that coding agent costs remained the fastest growing segment of enterprise AI budgets in 2026. Read the xAI official announcement and Stanford HAI.
How Do the Top Options Compare?
| Tool | Free Tier | Input Price | Output Price | Context Window |
|---|---|---|---|---|
| Grok Build 0.1 | 100 requests/day | $0.40 per 1M tokens | $2.00 per 1M tokens | 128K |
| OpenAI Codex | 50 messages/day | $0.50 per 1M tokens | $3.00 per 1M tokens | 128K |
| Anthropic Claude Code | 5-hour rolling credits | $0.80 per 1M tokens | $4.00 per 1M tokens | 200K |
| Google Gemini Code Assist | Tightened free tier | $0.35 per 1M tokens | $1.40 per 1M tokens | 2M |
| GitHub Copilot | 2,000 completions/month | $0.60 per 1M tokens | $2.40 per 1M tokens | 128K |
Prices and limits were verified on July 8, 2026. Free tiers change quickly. Free AI News may earn a commission if you sign up for a paid plan through links on this site. This does not affect our reporting.
1. Grok Build 0.1 , Best for cost-sensitive teams needing full agentic coding
Grok Build 0.1 is xAI’s first agentic coding API. It launched on July 8, 2026 as a preview release. The API handles file editing, terminal commands, browser tasks, and test runs. Developers call a single endpoint and the model manages the tool loop. xAI packed 128,000 tokens of context into the preview, enough for medium sized repositories.
Pricing starts at $0.40 per million input tokens and $2.00 per million output tokens. Browser and terminal tool calls are inclusive. That is a direct contrast to providers that add per-tool fees. Free tier access is limited to 100 requests per day. The company said rate limits may fall during peak hours. Teams that want longer context or higher limits must wait for Grok Build 1.0. The API currently lacks fine-tuning and custom model routing. You can read the xAI official announcement.
Key strengths:
- ✅ Free tier includes 100 requests per day without a time limit
- ✅ Input pricing is $0.40 per million tokens, 20% below OpenAI Codex
- ✅ Browser and terminal tool calls are bundled into the base rate
- ✅ 128,000 token context window supports large single files
- ✅ Simple REST API with Python and TypeScript SDKs
- ❌ Preview release has low rate limits during peak hours
- ❌ No fine-tuning or custom model routing yet
- ❌ Limited region availability at launch
Who it’s for: Developers who want a cheap agentic coding API with bundled tool calls should start with Grok Build 0.1.
2. OpenAI Codex , Best for teams already using OpenAI models
OpenAI Codex entered the agentic coding field with a free tier in June 2026. It runs on the same API stack as GPT-5.2 and offers direct integration with ChatGPT memory. Codex pricing at this writing was $0.50 per million input tokens and $3.00 per million output tokens. The free tier allows 50 messages per day, a lower allowance than Grok Build 0.1. Codex supports 128,000 tokens of context and has tighter security controls for enterprise accounts.
Unlike xAI, OpenAI charges separately for browser tool calls above a monthly threshold. Developers can monitor usage through the OpenAI dashboard. Teams that already use OpenAI embeddings and fine-tuning often prefer Codex because it reuses existing API keys and permissions. The OpenAI pricing page lists current rates.
Key strengths:
- ✅ Free tier for ChatGPT Plus and API developers
- ✅ Same API permissions and keys as other OpenAI models
- ✅ Strong enterprise security controls and audit logs
- ❌ Output price is 50% higher than Grok Build 0.1
- ❌ Free tier message cap is half of Grok’s daily request count
- ❌ Browser tool calls can add extra charges
Who it’s for: Teams already committed to OpenAI infrastructure should stay with Codex despite the higher output price.
3. Anthropic Claude Code , Best for complex long-context refactoring jobs
Anthropic Claude Code shifted to a credit pool system on June 15, 2026. The flat-rate agent subsidy ended. Developers now burn credits per tool call and per token. Claude Code uses Claude Opus 4.5 and offers a 200,000 token context window, the widest of the current coding agent APIs. That window matters for large repos. Pricing at this writing was $0.80 per million input tokens and $4.00 per million output tokens.
The free tier is no longer flat. It resets on five-hour rolling limits that vary by demand. Anthropic’s per-tool call fee adds up for heavy browser automation. The company published the new credit table on its pricing page. Developers who want deterministic, high-context edits often accept the higher cost. Smaller teams saw sharp bill increases after the subsidy ended. The Anthropic pricing page shows current credit rates.
Key strengths:
- ✅ 200,000 token context window handles very large codebases
- ✅ Strong performance on long refactoring chains
- ✅ Detailed usage breakdown per tool call in the console
- ❌ Highest input and output prices among major coding agents
- ❌ No true free tier after the credit pool change
- ❌ Per-tool call fees create unpredictable monthly bills
Who it’s for: Teams with large codebases and tolerance for higher, less predictable bills should consider Claude Code.
4. Google Gemini Code Assist , Best for budget users who need massive context
Google Gemini Code Assist changed pricing in June 2026 after Google’s AI price cuts. The standard API costs $0.35 per million input tokens and $1.40 per million output tokens. It offers a two million token context window on Gemini 3 Pro. That is more than 15 times the context of Grok Build 0.1. The free tier was tightened earlier in 2026. Google removed free access to some pro models and pushed developers toward paid plans.
Independent pricing analysis showed Google’s input price still beats xAI by $0.05 per million tokens. For high volume, that gap matters. The Gemini Code Assist agent can read entire repositories in one pass, which reduces tool call loops. Google also offers a $19 per month flat option for individual developers. That flat option bundles a set number of daily requests. Teams that need the largest context and the lowest list price often start with Gemini. See Google AI price cuts should make OpenAI Anthropic nervous.
Key strengths:
- ✅ Lowest input and output prices among major coding APIs
- ✅ Two million token context window reads entire repos
- ✅ $19 per month flat option for solo developers
- ❌ Free tier was cut for pro models earlier in 2026
- ❌ Agent quality lags Claude Code on complex refactors
- ❌ Browser automation requires separate Google Cloud setup
Who it’s for: Developers who need the largest context window or the lowest per-token price should use Gemini Code Assist.
5. GitHub Copilot , Best for teams deep in GitHub and VS Code
GitHub Copilot moved to usage-based billing in June 2026. The change sparked developer backlash over hidden costs. Copilot’s agent mode uses a multiplier system for tool calls, which makes monthly bills hard to predict. At this writing, Copilot’s coding agent listed $0.60 per million input tokens and $2.40 per million output tokens on the OpenAI-backed model. The free tier allows 2,000 completions per month, but agent mode requests draw from the same quota at a higher multiplier.
GitHub’s main advantage is integration. Copilot sits inside VS Code, Visual Studio, and GitHub Actions. Developers do not need a separate API endpoint for most tasks. The model can open pull requests, run CI checks, and comment on code. That workflow saves time for teams that already live in GitHub.
However, the multiplier billing frustrated many free users. A single agentic session could consume dozens of completion credits. GitHub later published a usage meter, but the damage was done. Teams considering Copilot should read GitHub Copilot usage based billing June 2026.
Key strengths:
- ✅ Tightest GitHub and VS Code integration of any coding agent
- ✅ Can open pull requests and run CI checks directly
- ✅ Free tier still covers casual completion use
- ❌ Multiplier billing makes agent costs unpredictable
- ❌ Higher input and output prices than Grok Build 0.1
- ❌ Free tier quota depletes fast in agent mode
Who it’s for: Developers who live in GitHub and want one tool for completions, PRs, and CI should pick GitHub Copilot.
Frequently Asked Questions
What is Grok Build 0.1?
Grok Build 0.1 is xAI’s first agentic coding API. It lets developers send coding tasks and the model manages file edits, terminal commands, browser actions, and tests through a tool loop.
How much does Grok Build 0.1 cost?
Paid usage starts at $0.40 per million input tokens and $2.00 per million output tokens. Browser and terminal tool calls are bundled into the base rate at launch.
What is included in the Grok Build 0.1 free tier?
The free tier includes 100 API requests per day with a 128,000 token context window. Developers must add a payment method to verify the account, but they are not charged until they exceed the daily limit.
How does Grok Build 0.1 compare to OpenAI Codex?
Grok Build 0.1 is cheaper on input and output. OpenAI Codex lists $0.50 per million input tokens and $3.00 per million output tokens. Grok also offers a higher free request count.
Which regions can access Grok Build 0.1 at launch?
xAI did not publish a full region list. Availability was limited to the United States, United Kingdom, and European Economic Area at launch, according to the API documentation.
Is Grok Build 0.1 ready for production use?
No. xAI labeled the release a preview. Rate limits may fall during peak hours and the API lacks fine-tuning and custom model routing.
Does xAI charge separately for browser or terminal tool calls?
No. Tool calls are included in the base per-token price for Grok Build 0.1. Some rivals bill browser actions separately after a monthly threshold.
What Should You Remember?
- Grok Build 0.1 opened to developers on July 8, 2026.
- Free tier offers 100 requests per day with a 128K context window.
- Pricing starts at $0.40 per million input tokens and $2.00 per million output.
- Bundled tool calls cover browser and terminal actions in the base rate.
- OpenAI Codex costs $0.50 input and $3.00 output, a 20% higher input price.
- Anthropic Claude Code still charges the highest rates at $0.80 input and $4.00 output.
- Google Gemini Code Assist remains $0.35 input and $1.40 output with a 2M context window.
Free AI News is an independent editorial publication. Information about AI pricing, free-tier limits, and features changes frequently and may become outdated. Always verify current details through the vendor’s official pages. Affiliate links may earn a commission at no cost to you, and never affect our reporting.
