# Seats & Tokens: full data (snapshot 2026-10-21)

Source: https://seatsandtokens.com. License CC BY 4.0. Prices are USD list prices excluding tax. "Not published" means the vendor does not state it.

## Cursor (Cursor (sold via X Subscription Corp.))
Page: https://seatsandtokens.com/tools/cursor/

| Plan | Price/month | Key limits | Source | Checked |
|---|---|---|---|---|
| Free | $0 | Limited agent & chat requests | https://cursor.com/pricing | 2026-10-21 |
| Plus | $8 | Basic shared usage pool, refreshes weekly | https://cursor.com/pricing | 2026-10-21 |
| Premium | $30 | 4x Plus usage; third-party models billed at API list price on top | https://cursor.com/pricing | 2026-10-21 |
| Super | $100 | 15x Plus usage; priority at peak | https://cursor.com/pricing | 2026-10-21 |
| Ultra | $200 | 50x Plus usage | https://cursor.com/pricing | 2026-10-21 |
| Teams | $30 | per user/month, from; 4-50x Plus | https://cursor.com/pricing | 2026-10-21 |

Verdict: Since the October 2026 repricing, Cursor is cheapest if you are happy with Grok models: Premium ($30) covers them from a shared pool. If you mainly want Claude or GPT, every token is billed on top at API list price, so heavy Claude users should compare against Claude Max or Copilot's credit plans.

## Claude Code (Anthropic)
Page: https://seatsandtokens.com/tools/claude-code/

| Plan | Price/month | Key limits | Source | Checked |
|---|---|---|---|---|
| Pro | $20 | $17/mo billed annually ($200/yr); ≥5x Free per 5-hour session + weekly limits; Claude Code included | https://claude.com/pricing | 2026-10-21 |
| Max 5x | $100 | 5x Pro per 5-hour session; $100/mo API credits; Fable 50% of weekly limits | https://claude.com/pricing | 2026-10-21 |
| Max 20x | Not published | 20x Pro per session; $200/mo API credits | https://claude.com/pricing | 2026-10-21 |
| Enterprise | $20 | per seat/month + usage at API rates, annual | https://claude.com/pricing | 2026-10-21 |

Verdict: Claude Code comes with any paid Claude plan and shares one usage pool with chat. For steady daily agent work on Claude models, a flat Max plan is usually easier to budget than paying per token, but the limits are published only as multiples of Pro, not as token counts.

## GitHub Copilot (GitHub / Microsoft)
Page: https://seatsandtokens.com/tools/github-copilot/

| Plan | Price/month | Key limits | Source | Checked |
|---|---|---|---|---|
| Free | $0 | 2,000 completions + 50 chat requests; limited agent | https://github.com/features/copilot/plans | 2026-10-21 |
| Pro | $10 | $15/mo total AI credits ($10 base + $5 flex); 1 credit=$0.01; completions unlimited | https://github.com/features/copilot/plans | 2026-10-21 |
| Pro+ | $39 | $70/mo total AI credits ($39 base + $31 flex) | https://github.com/features/copilot/plans | 2026-10-21 |
| Max | $100 | $200/mo total AI credits ($100 base) | https://github.com/features/copilot/plans | 2026-10-21 |

Verdict: Copilot is now the most transparent mainstream option: every plan includes a dollar-denominated AI Credit allowance (1 credit = $0.01) and completions stay unlimited on paid plans. Pro at $10 includes $15 of credits, which makes it the cheapest way to sample many frontier models.

## OpenAI Codex (OpenAI)
Page: https://seatsandtokens.com/tools/openai-codex/

| Plan | Price/month | Key limits | Source | Checked |
|---|---|---|---|---|
| Free | $0 | Limited | https://developers.openai.com/codex/pricing | 2026-10-21 |
| Go | $8 | Lightweight | https://developers.openai.com/codex/pricing | 2026-10-21 |
| Plus | $20 | Per 5h local msgs: GPT-6.1 Sol 15-160, GPT-6 Astra 5-45, Luna 350-3,000; weekly limits | https://developers.openai.com/codex/pricing | 2026-10-21 |
| Pro | $100 | Tiers $100/$200/$500; no 5-hour limit; Ultrafast on $500 | https://developers.openai.com/codex/pricing | 2026-10-21 |
| Business | $25 | per user/month billed monthly (annual cheaper, 2+ users) | https://developers.openai.com/codex/pricing | 2026-10-21 |

Verdict: Codex comes with ChatGPT plans, and OpenAI publishes unusually concrete limits (local messages per 5 hours by model) plus a credit rate card. Plus ($20) suits a few focused sessions a week; Pro tiers ($100/$200/$500) remove the 5-hour limit.

## Windsurf (Devin Desktop) (Cognition)
Page: https://seatsandtokens.com/tools/windsurf/

| Plan | Price/month | Key limits | Source | Checked |
|---|---|---|---|---|
| Free | $0 | Light daily/weekly quota | https://devin.ai/pricing | 2026-10-21 |
| Pro | $20 | Daily + weekly quotas; overage at API pricing | https://devin.ai/pricing | 2026-10-21 |
| Max | $200 | Much larger quota | https://devin.ai/pricing | 2026-10-21 |
| Teams | $80 | $80/mo minimum; $40 per full seat; up to 200 users | https://devin.ai/pricing | 2026-10-21 |

Verdict: Windsurf, now sold as Devin Desktop, uses daily plus weekly quotas instead of credits. Pro at $20 is competitive, but the quotas are described only loosely, so budgeting heavy use is guesswork. Overage is billed at API pricing.

## Gemini Code Assist (Google)
Page: https://seatsandtokens.com/tools/gemini-code-assist/

| Plan | Price/month | Key limits | Source | Checked |
|---|---|---|---|---|
| Standard | $22.80 | per user/month monthly; $19 annual | https://codeassist.google/ | 2026-10-21 |
| Enterprise | $54 | per user/month monthly; $45 annual | https://codeassist.google/ | 2026-10-21 |

Verdict: Gemini Code Assist is now a business product priced per seat (Standard from $19/user on annual terms). Individuals were moved to Antigravity in June 2026, so solo developers should look there instead.

## JetBrains AI (Junie) (JetBrains)
Page: https://seatsandtokens.com/tools/jetbrains-ai/

| Plan | Price/month | Key limits | Source | Checked |
|---|---|---|---|---|
| AI Free | $0 | 3 AI credits / 30 days (1 credit=$1) | https://www.jetbrains.com/help/ai-assistant/licensing-and-subscriptions.html | 2026-10-21 |
| AI Pro | $10 | 10 AI credits / 30 days; top-ups valid 12 months | https://www.jetbrains.com/help/ai-assistant/licensing-and-subscriptions.html | 2026-10-21 |
| AI Ultimate | $30 | 35 AI credits / 30 days | https://www.jetbrains.com/help/ai-assistant/licensing-and-subscriptions.html | 2026-10-21 |

Verdict: JetBrains AI prices quota in dollars (1 credit = $1), and completion doesn't consume credits. AI Pro ($10) is a light add-on for JetBrains users. AI Ultimate ($30 with 35 credits) is the tier JetBrains recommends for regular Junie use.

## Kiro (AWS)
Page: https://seatsandtokens.com/tools/kiro/

| Plan | Price/month | Key limits | Source | Checked |
|---|---|---|---|---|
| Free | $0 | 50 credits | https://kiro.dev/pricing/ | 2026-10-21 |
| Pro | $20 | per user; add-on credits $0.04 each | https://kiro.dev/pricing/ | 2026-10-21 |
| Pro+ | $40 | Not published | https://kiro.dev/pricing/ | 2026-10-21 |
| Pro Max | $100 | Not published | https://kiro.dev/pricing/ | 2026-10-21 |
| Power | $200 | Not published | https://kiro.dev/pricing/ | 2026-10-21 |

Verdict: Kiro has a wide price ladder ($20 to $200) and clear overage pricing ($0.04 per credit), but a credit is a unit of work, not tokens, so you can't convert it into model prices. It's best for teams that want AWS billing and spec-driven workflows.

## Zed (Zed Industries)
Page: https://seatsandtokens.com/tools/zed/

| Plan | Price/month | Key limits | Source | Checked |
|---|---|---|---|---|
| Personal | $0 | BYO keys / external agents unlimited | https://zed.dev/pricing | 2026-10-21 |
| Pro | $10 | $5 tokens included; overage at API list +10%; unlimited edit predictions | https://zed.dev/pricing | 2026-10-21 |
| Business | $30 | per seat/month; no bundled credits | https://zed.dev/pricing | 2026-10-21 |

Verdict: Zed is one of the few tools with a fully transparent formula: $10/month includes $5 of tokens, then API list price +10%. Or pay nothing and bring your own key. It's a good fit if you want predictable pass-through pricing.

## Augment Code (Augment)
Page: https://seatsandtokens.com/tools/augment-code/

| Plan | Price/month | Key limits | Source | Checked |
|---|---|---|---|---|
| Standard | $20 | flat for up to 50 seats; $20 usage incl.; LLM at list +40% fee | https://www.augmentcode.com/pricing | 2026-10-21 |
| Business | $100 | flat up to 50 seats; $100 usage incl. | https://www.augmentcode.com/pricing | 2026-10-21 |

Verdict: Augment's flat team pricing (one price for up to 50 seats) is unusual and good for small teams with uneven usage. Note the 40% service fee on LLM usage, which makes heavy individual use pricier than going direct.

## Replit (Replit)
Page: https://seatsandtokens.com/tools/replit/

| Plan | Price/month | Key limits | Source | Checked |
|---|---|---|---|---|
| Core | $20 | $18/mo annual; $20 toward models; up to 30h Free Mode chat | https://replit.com/pricing | 2026-10-21 |
| Pro | $100 | $90/mo annual; $100 toward models; 15 collaborators | https://replit.com/pricing | 2026-10-21 |

Verdict: Replit bundles hosting, a workspace and agent usage. The $20 tier includes $20 toward more powerful models, so it competes with app builders more than with IDE assistants.

## Lovable (Lovable)
Page: https://seatsandtokens.com/tools/lovable/

| Plan | Price/month | Key limits | Source | Checked |
|---|---|---|---|---|
| Pro (100 credits) | $25 | 100 credits/mo + 5 daily build credits; tiers to 10,000 credits ($2,250) | https://docs.lovable.dev/introduction/subscription-plans | 2026-10-21 |
| Business (100 credits) | $50 | 100 credits/mo; SSO etc. | https://docs.lovable.dev/introduction/subscription-plans | 2026-10-21 |

Verdict: Lovable prices by credits with a long ladder (100 to 10,000 credits per month). Pro credits cost $0.25 each at the base tier. Example prompts on its pricing page cost 0.5 to 1.7 credits, which is the best guide to real cost.

## Bolt (StackBlitz)
Page: https://seatsandtokens.com/tools/bolt/

| Plan | Price/month | Key limits | Source | Checked |
|---|---|---|---|---|
| Free | $0 | 1M tokens/mo, 300K/day | https://bolt.new/pricing | 2026-10-21 |
| Pro | $18 | 10M tokens/mo with rollover ($216/yr shown; asterisked price) | https://bolt.new/pricing | 2026-10-21 |
| Teams | $27 | per member; 10M tokens/member | https://bolt.new/pricing | 2026-10-21 |

Verdict: Bolt is one of the few app builders that meters in tokens (10M per month on Pro, with rollover). Token use grows with project size because Bolt syncs your file system to the model.

## v0 (Vercel)
Page: https://seatsandtokens.com/tools/v0/

| Plan | Price/month | Key limits | Source | Checked |
|---|---|---|---|---|
| Free | $0 | 7 messages/day | https://v0.app/pricing | 2026-10-21 |
| Team | $30 | per user; $30 included credits + $2 daily login credits | https://v0.app/pricing | 2026-10-21 |
| Business | $100 | per user; training opt-out | https://v0.app/pricing | 2026-10-21 |

Verdict: v0 publishes per-token rates for its own models (v0 Mini to v0 Max Fast), which makes costs auditable. It's best if you deploy on Vercel anyway.

## Cline, Roo Code & Aider (Open source)
Page: https://seatsandtokens.com/tools/open-source-byok/

| Plan | Price/month | Key limits | Source | Checked |
|---|---|---|---|---|
| Open source | $0 | BYO API key; pay provider token rates | n/a | 2026-10-21 |

Verdict: Free tools: you pay your model provider directly at token prices, which our calculator shows exactly. Costs scale linearly with usage, and there is no subscription buffer.

## Tabnine (Tricentis)
Page: https://seatsandtokens.com/tools/tabnine/

| Plan | Price/month | Key limits | Source | Checked |
|---|---|---|---|---|
| — | Not published | No public self-serve pricing; acquired by Tricentis | https://www.tabnine.com/pricing/ | 2026-10-21 |

Verdict: Tabnine is now part of Tricentis, and tabnine.com/pricing redirects to a Tricentis contact page. Self-serve pricing is not published.

## Amazon Q Developer (AWS)
Page: https://seatsandtokens.com/tools/amazon-q-developer/

| Plan | Price/month | Key limits | Source | Checked |
|---|---|---|---|---|
| — | Not published | Reported closed to new signups since 2026-05-15 (end of support 2027-04-30) | https://devtoolsreview.com/reports/ai-coding-pricing-october-2026/ | 2026-10-21 |

Verdict: A third-party report says Amazon Q Developer has been closed to new signups since 15 May 2026, with end of support on 30 April 2027. Kiro is AWS's current AI IDE.

## Model token prices (USD per 1M tokens)

| Model | Input | Output | Cached input | Source | Checked |
|---|---|---|---|---|---|
| Anthropic: Claude Opus 5.5 | $4 | $20 | $0.2 | https://claude.com/pricing | 2026-10-21 |
| Anthropic: Claude Sonnet 5.5 | $2 | $10 | $0.1 | https://claude.com/pricing | 2026-10-21 |
| Anthropic: Claude Haiku 5.5 | $0.1 | $0.5 | $0.01 | https://claude.com/pricing | 2026-10-21 |
| Anthropic: Claude Fable 5.1 | $10 | $50 | $0.25 | https://claude.com/pricing | 2026-10-21 |
| Anthropic: Claude Opus 5 | $5 | $25 | $0.5 | https://claude.com/pricing | 2026-10-21 |
| Anthropic: Claude Sonnet 5 | $2 | $10 | $0.2 | https://claude.com/pricing | 2026-10-21 |
| OpenAI: GPT-6.1 Sol | $2 | $10 | $0.1 | https://openrouter.ai/api/v1/models | 2026-10-21 |
| OpenAI: GPT-6 Sol | $2 | $10 | $0.2 | https://openrouter.ai/api/v1/models | 2026-10-21 |
| OpenAI: GPT-6 Astra | $10 | $50 | $1 | https://openrouter.ai/api/v1/models | 2026-10-21 |
| OpenAI: GPT-6 Luna | $0.1 | $0.5 | $0.01 | https://openrouter.ai/api/v1/models | 2026-10-21 |
| OpenAI: GPT-5.6 Terra | $2 | $12 | $0.2 | https://openrouter.ai/api/v1/models | 2026-10-21 |
| OpenAI: GPT-5.5 | $5 | $30 | $0.5 | https://openrouter.ai/api/v1/models | 2026-10-21 |
| Google: Gemini 4 Argon | $2 | $10 | $0.1 | https://openrouter.ai/api/v1/models | 2026-10-21 |
| Google: Gemini 3.8 Flash | $0.825 | $4.13 | $0.0825 | https://openrouter.ai/api/v1/models | 2026-10-21 |
| Google: Gemini 3.5 Flash Lite | $0.3 | $2.5 | $0.03 | https://openrouter.ai/api/v1/models | 2026-10-21 |
| Google: Gemini 3.1 Pro Preview | $2 | $12 | $0.2 | https://openrouter.ai/api/v1/models | 2026-10-21 |
| SpaceXAI: Grok 4.7 | $2 | $6 | $0.5 | https://openrouter.ai/api/v1/models | 2026-10-21 |
| SpaceXAI: Grok Build 0.1 | $1 | $2 | $0.2 | https://openrouter.ai/api/v1/models | 2026-10-21 |
| DeepSeek: DeepSeek V4 Pro | $1.32 | $3.96 | $0.044 | https://api-docs.deepseek.com/quick_start/pricing | 2026-10-21 |
| DeepSeek: DeepSeek V4.1 Flash | $0.3 | $1.2 | $0.006 | https://api-docs.deepseek.com/quick_start/pricing | 2026-10-21 |
| Qwen: Qwen3.8 Max Prime | $4 | $12 | $0.5 | https://openrouter.ai/api/v1/models | 2026-10-21 |
| Qwen: Qwen3 Coder Next | $0.3 | $1.5 | Not published | https://www.alibabacloud.com/help/en/model-studio/qwen3-coder-next | 2026-10-21 |
| MoonshotAI: Kimi K3 | $0.239 | $11.94 | $0.21 | https://openrouter.ai/api/v1/models | 2026-10-21 |
| MoonshotAI: Kimi K2.7 Code | $0.671 | $3.35 | $0.18 | https://openrouter.ai/api/v1/models | 2026-10-21 |
| Z.ai: GLM 5.3 | $1.4 | $4.4 | $0.26 | https://docs.z.ai/guides/overview/pricing | 2026-10-21 |
| Mistral: Devstral 2 2512 | $0.4 | $2 | $0.04 | https://openrouter.ai/api/v1/models | 2026-10-21 |
| MiniMax: MiniMax M3 | $0.3 | $1.2 | $0.06 | https://openrouter.ai/api/v1/models | 2026-10-21 |
| Meta: Muse Spark 1.3 | $1.25 | $4.25 | $0.15 | https://openrouter.ai/api/v1/models | 2026-10-21 |

## Change log

- 2026-10-21: First Seats & Tokens snapshot. Baseline capture of plan prices for 17 tools and token prices for 28 models.
- 2026-10-21: Model prices: switched four models to first-party list prices. DeepSeek V4 Pro and V4.1 Flash (peak rates; off-peak is half), Qwen3 Coder Next (Alibaba Model Studio, international) and GLM-5.3 (Z.ai) now show the vendor's own price instead of an OpenRouter routed price. Kimi K3 and K2.7 Code still show OpenRouter prices because Moonshot's price table could not be read. Source: https://api-docs.deepseek.com/quick_start/pricing
- 2026-10-14: GPT-5.5 retired from Codex (ChatGPT sign-in). GPT-5.5 retires from Codex with ChatGPT sign-in on 14 Oct 2026 at 19:00 UTC. API-key sessions are not affected. Source: https://developers.openai.com/codex/pricing
- 2026-10: Cursor reprices: $8 Grok-only Plus, $30 Premium. Cursor's pricing page now lists Free, Plus $8 (Grok models only), Premium $30, Super $100, Ultra $200, Teams from $30/user. Third-party models are billed separately at API list price. A third-party report says the previous $20 Pro tier is gone. Source: https://cursor.com/pricing
- 2026-07-30: Tabnine acquired by Tricentis. tabnine.com/pricing now redirects to Tricentis; there's no self-serve pricing. Closing date per a third-party report. Source: https://devtoolsreview.com/reports/ai-coding-pricing-october-2026/
- 2026-06-18: Gemini Code Assist for individuals replaced by Antigravity. The unpaid individual tier and Gemini CLI for individuals were replaced by Antigravity CLI and Antigravity. Source: https://codeassist.google/
- 2026-06-13: Lovable simplifies billing into one credit balance. Plan prices unchanged. Free, Pro and Business workspaces get 20 Cloud credits per month; dollar-based Cloud/AI balances were converted to credits. Source: https://lovable.dev/blog/simplifying-billing
- 2026-06-01: Copilot code review starts consuming Actions minutes. Beginning 1 June 2026, Copilot code review workflows also consume GitHub Actions minutes. Copilot bills chat and agents in GitHub AI Credits (1 credit = $0.01). Source: https://github.com/features/copilot/plans
- 2026-05-15: Amazon Q Developer closed to new signups (reported). Reported closed to new signups since 15 May 2026; end of support 30 April 2027. Not yet verified on an AWS page. Source: https://devtoolsreview.com/reports/ai-coding-pricing-october-2026/
- 2026-03-19: Windsurf replaces credits with daily and weekly quotas; adds $200 Max. New plans: Free, Pro $20, Teams $40/seat, Max $200 (new). Overage at API pricing. Existing Pro/Teams prices grandfathered. Teams later changed to an $80/month minimum plus $40 per full seat (Devin docs). Source: https://devin.ai/blog/windsurf-pricing-plans