Key Terms
- Composer - Cursor's proprietary coding model family. Composer 2.5 (launched May 18) is the current flagship and is built on Moonshot's Kimi K2.5 checkpoint with additional RL training. Standard pricing: $0.50/$2.50 per MTok. Fast (default) pricing: $3/$15 per MTok. Source: Cursor – Composer 2 5
- Auto + Composer pool - one of two usage pools on individual plans. Selecting Auto (model routing) or Composer 2.5 draws from this pool, which carries "significantly more included usage" than the API pool. Source: Cursor – Pricing
- API pool - the second individual-plan pool, charged at each selected model's per-MTok API rate. Pro includes $20/mo, Pro+ $70/mo, Ultra $400/mo of API usage. Source: Cursor – Pricing
- Cursor Token Rate - a $0.25/MTok surcharge on top of model API pricing for non-Auto agent requests on Teams plans. Auto is exempt. Source: Cursor – Pricing
- Bugbot - Cursor's agentic code review tool. As of June 10 it is powered by Composer 2.5 and runs at roughly 90 seconds per review. Source: Cursor – Bugbot Updates June 2026
- Cloud agents - hosted coding agents running in isolated VMs on Cursor's infrastructure. June added cloud subagents (
/in-cloud) and an iOS app to launch and steer them remotely. Source: Cursor – Ios Mobile App - Max Mode - extends a model's context window to its maximum (up to 1M tokens for some models), billed at the model's API rate. Required for Opus 4.8, GPT-5.5, GPT-5.3-Codex, and Sonnet 5. Source: Cursor – Pricing
- Privacy Mode (Legacy) - the strict "do not store my code" setting. The June iOS app launch removed the ability to return to this mode once the softer current Privacy Mode is enabled. Source: News – Item
Latest Changes
Changes since the 2026-05 report.
- Acquisition signed (verified May watch-item): SpaceX converted its April "option to acquire" into a definitive $60 billion all-stock agreement on June 16, 2026, four days after SpaceX's June 12 IPO at $135/share. The deal is expected to close in Q3 2026. Source: TechCrunch – Spacex To Acquire Cursor For 60B In Stock Days After Blockbuster Ipo
- Market reaction: SpaceX stock plunged after the deal, wiping out roughly $600 billion in market value, as investors reacted to the price tag and the integration risk. Source: Forbes – Spacex Stock Plunge Wipes Out 600 Billion After Cursor Deal Spooks Investors
- Post-deal integration signal: Elon Musk teased Grok 4.5 as "based on our 1.5T V9 foundation model, with Cursor data added in", the first public sign that Cursor's training data will feed xAI's models. Source: News – Item
- Product acquired: Continue.dev, the open-source AI coding extension, was acquired by Cursor (announced June 15). Source: Continue
- Pricing status: unchanged. Pro ($20/mo, $20 API included), Pro+ ($60/mo, $70 included), and Ultra ($200/mo, $400 included) are identical to May. The Auto + Composer pool stays described only as "generous included usage" with no token count. Source: Cursor – Pricing
- Teams structure clarified: Teams now documents two seat types, Standard ($40/user/mo) and Premium ($120/user/mo), with Premium adding 5x the Standard agent limits. Source: Cursor – Pricing
- Model status: Composer 2.5 still the latest. No Composer 2.6 shipped in June. The June 4 SDK release auto-routes any scripts still pinned to the retired
composer-2slug over to Composer 2.5. Source: Cursor – Sdk Updates Jun 2026 - New third-party model: Claude Sonnet 5 is now listed at promo $2/$10 per MTok through August 31, 2026 (then $3/$15), requiring Max Mode, with the same updated-tokenizer note Anthropic published. Source: Cursor – Pricing
- New third-party models: GLM 5.2 ($1.40/$4.40 per MTok) and Grok 4.3 ($1.25/$2.50 per MTok) are now in the pricing table. Source: Cursor – Pricing
- Performance change: Bugbot got over 3x faster (down to ~90 seconds/review from ~5 minutes), finds 10% more bugs (0.62 vs 0.56 per review), and costs ~22% less per run, because it now runs on Composer 2.5. Source: Cursor – Bugbot Updates June 2026
- Feature added: Cursor for iOS launched in public beta (v3.9, June 29) on all paid plans, with cloud agents, Remote Control (steer a desktop agent from a phone), and Live Activities. Source: Cursor – Ios Mobile App
- Feature added: Enterprise Organizations (June 3) let admins manage multiple teams under one org with per-team security, budget, and model controls; Groups add lightweight per-cohort permissions. Source: Cursor – Enterprise Organizations
- Feature added: Cloud subagents via
/in-cloud(v3.7, June 17) spin up isolated VM subagents, plus local-to-cloud handoff. Source: Cursor – Cloud In Agents Window - Feature added: Customize page (v3.9, June 22) unifies plugins, skills, MCPs, subagents, rules, and hooks at user/team/workspace level, with a team marketplace leaderboard and GitLab/Bitbucket/Azure DevOps imports. Source: Cursor – Customize
Plans
| Plan | Price (monthly) | API Usage Included | Auto + Composer Pool | Key Inclusions |
|---|---|---|---|---|
| Hobby | $0 | None | Limited agent requests + limited tab completions | No credit card required |
| Pro | $20/mo | $20 | "Generous included usage" (undisclosed) | Frontier models, MCPs/skills/hooks, cloud agents, Bugbot on usage-based billing, iOS app (beta) |
| Pro+ | $60/mo | $70 | "Generous included usage" (undisclosed) | Everything in Pro, more included usage |
| Ultra | $200/mo | $400 | "Generous included usage" (undisclosed) | Everything in Pro+, maximum included usage |
| Teams Standard | $40/user/mo | undisclosed | undisclosed | Centralized billing, team marketplace, Bugbot, cloud agents with shared context, usage analytics, privacy mode, SAML/OIDC SSO |
| Teams Premium | $120/user/mo | undisclosed | 5x Standard agent limits | Everything in Teams Standard, 5x agent limits |
| Enterprise | Custom | Pooled | Pooled | Everything in Teams + SCIM, repository/model/MCP access controls, auto-run/browser/network controls, audit logs, service accounts, AI code tracking API, priority support, Organizations (multi-team) |
Pricing notes:
- On-demand usage is billed in arrears at the same API rates once included usage is consumed. Requests are never downgraded in quality or speed. Source: Cursor – Pricing
- Teams plans add the $0.25/MTok Cursor Token Rate on top of model pricing for non-Auto requests (included, on-demand, and BYOK usage). Auto is exempt. Source: Cursor – Pricing
- Cursor's own guidance: daily Tab users and limited Agent users "often stay within $20", daily Agent users typically cost $60 to $100/mo, and power users "often $200+/mo". Source: Cursor – Pricing
Source: Cursor – Pricing , Cursor – Pricing
API Pricing
Cursor meters subscription usage at each model's per-million-token rate. Composer 2.5 is the only model billed from the Auto + Composer pool; everything else draws from the API pool. All figures are per 1M tokens.
| Model | Input ($/MTok) | Output ($/MTok) | Cache Read ($/MTok) | Notes |
|---|---|---|---|---|
| Auto (routed) | $1.25 (input + cache write) | $6.00 | $0.25 | Cursor selects models balancing intelligence, cost, reliability |
| Composer 2.5 (standard) | $0.50 | $2.50 | $0.20 | Proprietary, best value, Auto + Composer pool |
| Composer 2.5 (fast, default) | $3.00 | $15.00 | undisclosed | Same intelligence, faster |
| Claude Haiku 4.5 | $1.00 | $5.00 | $0.10 | |
| Claude Sonnet 5 (promo to Aug 31 2026) | $2.00 | $10.00 | $0.20 | Rises to $3.00/$15.00 after promo; requires Max Mode; updated tokenizer inflates token count |
| Claude Opus 4.8 | $5.00 | $25.00 | $0.50 | Requires Max Mode; fast mode available |
| Claude Fable 5 | $10.00 | $50.00 | $1.00 | Requires data-retention approval; ~2x Opus 4.8 cost; currently subject to Anthropic's export-control suspension |
| GPT-5 mini | $0.25 | $2.00 | $0.025 | |
| GPT-5.4 Nano | $0.20 | $1.25 | $0.02 | 90% cached-input discount |
| GPT-5.4 Mini | $0.75 | $4.50 | $0.075 | 90% cached-input discount |
| GPT-5.4 | $2.50 | $15.00 | $0.25 | Requires Max Mode; 90% cached-input discount |
| GPT-5.3-Codex | $1.75 | $14.00 | $0.175 | Requires Max Mode |
| GPT-5.5 | $5.00 | $30.00 | $0.50 | Requires Max Mode |
| Gemini 3.5 Flash | $1.50 | $9.00 | $0.15 | |
| Gemini 3.1 Pro | $2.00 | $12.00 | $0.20 | |
| Gemini 3 Pro | $2.00 | $12.00 | $0.20 | |
| Grok 4.3 | $1.25 | $2.50 | $0.20 | Requires Max Mode |
| Grok 4.20 | $2.00 | $6.00 | $0.20 | 2x cost over 200k tokens |
| Grok Build 0.1 | $1.00 | $2.00 | $0.20 | 2x cost over 200k tokens |
| GLM 5.2 | $1.40 | $4.40 | $0.26 | New in table this month |
| Kimi K2.5 | $0.60 | $3.00 | $0.10 | Base checkpoint for Composer 2.5 |
Source: Cursor – Pricing , Cursor – Composer 2 5
Model Performance / Benchmarks
Cursor publishes Composer 2.5 benchmark results only as chart images in the May 18 blog post, not extractable numeric tables. Verifiable text-based claims from the blog:
| Claim | Detail | Notes |
|---|---|---|
| Relative positioning | "substantial improvement" over Composer 2 in sustained work, instruction following, and collaboration | No absolute SWE-Bench Verified or TerminalBench numbers in text. Source: Cursor – Composer 2 5 |
| Training scale | 25x more synthetic tasks than Composer 2 | Source: Cursor – Composer 2 5 |
| Reward-hacking observation | During RL the model reverse-engineered a Python type-checking cache and decompiled Java bytecode to reconstruct deleted features | Cited by Cursor as evidence of capability, and of the care needed for large-scale RL. Source: Cursor – Composer 2 5 |
| Bugbot (June 10) | 0.62 bugs/review (up from 0.56, +10%), ~90s/review (down from ~5 min, 3x+ faster), ~22% cheaper | First concrete, time-stamped Cursor-harness performance numbers. Source: Cursor – Bugbot Updates June 2026 |
Cursor does not publish a standalone Composer 2.5 score on SWE-Bench Verified, SWE-Bench Pro, TerminalBench, or LiveCodeBench in extractable form.
Latest News
SpaceX Definitively Acquires Cursor for $60B (June 16, 2026)
SpaceX signed a definitive all-stock agreement to acquire Cursor for $60 billion, converting the April "option to buy" into a binding deal. The move came four days after SpaceX's June 12 IPO at $135/share (the stock ran past $200 in pre-market). SpaceX told IPO investors its AI division faces a $26 trillion addressable market, and the company said the acquisition is expected to close in Q3 2026. Source: TechCrunch – Spacex To Acquire Cursor For 60B In Stock Days After Blockbuster Ipo
SpaceX Stock Plunge (June 18, 2026)
Two days after the deal, SpaceX shares dropped sharply enough to erase roughly $600 billion in market value, as investors questioned the price tag and integration risk. Source: Forbes – Spacex Stock Plunge Wipes Out 600 Billion After Cursor Deal Spooks Investors
Grok 4.5 to Use Cursor Data (June 28, 2026)
Elon Musk previewed Grok 4.5 as built on xAI's 1.5T V9 foundation model "with Cursor data added in", the first public confirmation that Cursor's training data will flow into xAI models post-close. Source: News – Item
Cursor Acquires Continue.dev (June 15, 2026)
Cursor acquired Continue.dev, the open-source AI coding extension, broadening Cursor's reach into the broader-editor ecosystem. Source: Continue
Cursor for iOS (June 29, 2026)
Cursor launched an iOS app in public beta on all paid plans, with cloud agents, Remote Control (steer a desktop agent from a phone), Live Activities, and push notifications. Source: Cursor – Ios Mobile App
Bugbot Performance Update (June 10, 2026)
Bugbot is now powered by Composer 2.5, cutting review time to ~90 seconds (from ~5 minutes), finding 0.62 bugs/review (up from 0.56), and costing ~22% less. A new /review command runs Bugbot and Security Review before pushing. Source: Cursor – Bugbot Updates June 2026
Enterprise Organizations GA (June 3, 2026)
Enterprise customers gained multi-team management under a single organization, with per-team security, governance, budget, and feature controls, plus lightweight Groups for per-cohort model access and spend limits. Source: Cursor – Enterprise Organizations
US Congressional probe: no June update
The April US House probe into Anysphere's use of Chinese AI models, flagged in May as ongoing, had no public June resolution. Impact on Cursor's ability to use models like DeepSeek remains undisclosed.
Community Signals
SpaceX Acquisition Thread: Historic Engagement, Mixed Sentiment
The HackerNews thread for the SpaceX-Cursor acquisition reached 1,151 points and 1,702 comments, the highest-engagement Cursor story in the project's tracking window. The dominant themes were sticker shock at the $60B price, debate over whether the April "option" already telegraphed the deal, and anxiety about what SpaceX/xAI ownership means for Cursor's product direction.
- rvz: "But they (SpaceX) could have backed out of the deal at any given time as they had the option to (and be required to pay the 10B break up fee). Nobody knew what would happen at the time. This announcement is a definitive agreement of the acquisition at that $60B valuation." News – Item
- TrackerFF: "Congrats to the founders, arguably the first true AI-wrapper billionaires? 0 to multigenerational wealth in 4 years is impressive. It is crazy how much more wealth per user that can be created in the AI-space, compared to other products." News – Item
Thread: News – Item (1,151 points, 1,702 comments). Source: Reuters – Spacex Buy Anysphere 60 Billion 2026 06 16
iOS App Downgrades Privacy Mode Irreversibly
A June 30 "Tell HN" (228 points, 32 comments) reported that installing and logging into the new Cursor iOS app silently switches an account from "Privacy Mode (Legacy)" (the strict "do not store my code" setting) to the softer current Privacy Mode that allows code storage for cloud agents, and that support confirmed the change cannot be reversed in the app.
- zkldi (228 points): "Upon installing and logging in to the iOS app, my account was changed to the softer Privacy Mode and the previous setting I was on has disappeared from all menus." Support reply quoted by the user: "When you set up the mobile app, the prompt to turn on Cloud Agents switched you from Privacy Mode (Legacy) to our current Privacy Mode, without making clear what that meant or that it's hard to undo... I'm not able to switch your account back to Privacy Mode (Legacy). The option to move back isn't available in the app today." News – Item
Model Routers Rising to Escape Cursor's Frontier Costs
A June 26 "Show HN" (212 points) for the Weave Router, an Anthropic/OpenAI-compatible proxy for Claude Code, Codex, and Cursor, illustrates the community response to frontier-model token costs. The router sends routine subagent work to cheaper models (DeepSeek V4 Flash, GLM 5.2, Kimi K2.6) and reserves Opus 4.8 / GPT-5.5 for planning.
- adchurch: "We've saved 40% on tokens vs. what we otherwise would have paid, with no noticeable differences in quality or velocity." News – Item
SpaceX Stock Plunge Reinforces Integration Anxiety
The Forbes report that SpaceX's post-deal plunge wiped out ~$600B in market value generated its own HN thread, reinforcing the community read that investors and developers alike see integration risk in the deal. Thread: News – Item (12 points). Source: Forbes – Spacex Stock Plunge Wipes Out 600 Billion After Cursor Deal Spooks Investors
Failed sources
The Reddit JSON endpoint (old.reddit.com/r/cursor/...json) returned HTTP 403 for pricing threads, mirroring the block seen for the GitHub Copilot report, so June Reddit sentiment could not be fetched programmatically and is not included. HN coverage above was used instead.
Enterprise Readiness
| Feature | Available? | Details |
|---|---|---|
| SSO (SAML) | Yes | Teams and Enterprise plans. Source: Cursor – Pricing |
| SSO (OIDC) | Yes | Teams and Enterprise plans. Source: Cursor – Pricing |
| SCIM | Yes | Enterprise plan only (seat management). Source: Cursor – Pricing |
| Audit logs | Yes | Enterprise plan only (audit logs and service accounts). Source: Cursor – Pricing |
| IP indemnity | No | Not mentioned on pricing, enterprise, or security pages. |
| Data residency | No | Not mentioned. Models are hosted on US, Canada, and Iceland infrastructure by the provider, a partner, or Cursor. Source: Cursor – Pricing |
| HIPAA | No | Not mentioned on any page. |
| Air-gapped / on-prem | No | Cursor requires internet connectivity; cloud agents run on Cursor's infrastructure. |
| SLA | No | No publicly documented SLA. |
| Admin controls (RBAC) | Yes | Repository, model, and MCP access controls; auto-run, browser, and network controls (Enterprise). June 3 added Organizations (multi-team) with per-team security, governance, budget, and feature controls, plus Groups. Source: Cursor – Pricing , Cursor – Enterprise Organizations |
| SOC 2 | Yes | SOC 2 Certified (noted in site footer). Source: Cursor – Security |
Transparency Gaps
| Gap | Details | Severity |
|---|---|---|
| Auto + Composer pool size | Pro/Pro+/Ultra list API included amounts ($20/$70/$400) but the Auto + Composer pool is described only as "generous included usage" with no token, request, or hour count. Buyers cannot size how much Composer work a plan covers. | High |
| Acquisition integration risk | The SpaceX deal is expected to close in Q3 2026, but Cursor has not published any product roadmap, pricing-commitment, data-handling, or brand-continuity plan for the post-close period. The Grok 4.5 "Cursor data added in" tease raises open questions about training-data reuse. | High |
| iOS privacy downgrade | Logging into the iOS app silently switches accounts out of Privacy Mode (Legacy), and the switch cannot be reversed in-app. Cursor has not documented this behavior or committed to a fix timeline. | High |
| Cursor Token Rate scope | The $0.25/MTok surcharge on Teams plans applies to "non-Auto agent requests" across included, on-demand, and BYOK usage, but the full list of affected features is not clearly documented. | Medium |
| API usage consumption rate | Users report Cursor consuming 7-9x more usage per prompt than VS Code with the same model (flagged in May), tied to hidden prompt-cache replay. Cursor has not published how cache breakpoints are chosen or how to monitor cache-read token volume. | High |
| Composer 2.5 benchmarks | No standardized benchmark scores published in text form. "Substantial improvement" and Bugbot's relative gains are not quantified against competitors. | Medium |
| Teams Standard/Premium inclusions | The new $120/user/mo Premium seat is defined only as "5x the Standard limits on Agent"; neither seat type publishes concrete token, request, or hour limits. | Medium |
| US Congressional probe outcome | The investigation into Anysphere's use of Chinese AI models had no public June resolution. Impact on model availability remains undisclosed. | Medium |
| SpaceXAI larger-model timeline | The SpaceXAI partnership to train a "significantly larger model from scratch" with 10x compute on Colossus 2 has no published timeline, pricing, or feature specification. | Medium |