Google AI US personal monthly plans, DeepSeek API, Claude subscriptions and Sonnet 5.5, and GPT-6.1 Sol API were checked on 2026-10-24. Other provider prices retain their 2026-10-20 source scope. Compare fees, token rates and usage limits separately: an API cost scenario does not establish how many tasks a subscription can complete.
Pick by Budget
- Under $10/month (try it out):
Xiaomi MiMo Token Plan Liteat$6/month(¥39/month), 4.1B Credits, compatible with Claude Code / OpenClaw / OpenCode. Convert Credits using the model you select. Compare the separate $10 MiniMax Starter fee against its supported uses and limits. - Lower fixed fees:
Tencent Cloud Liteat¥40/month(~18,000 requests/month), orMiMo Token Plan Standardat$16/month(¥99/month, 11B Credits). Requests and Credits are different units; MiMo is restricted to supported coding tools.Kimi Andanteat¥49/monthif already in the Kimi ecosystem. - $20/month (main coding subscription):
Claude ProorChatGPT Plus. Claude Code users go with Claude Pro. ChatGPT ecosystem (Codex, Deep Research) go with Plus, but Plus has limited Codex quota.Google AI Procosts$19.99/monthfor US personal monthly billing and suits users who also need Gemini, Jules, Antigravity and 5 TB storage. - $50–100/month (higher subscription budget): Claude Max 5x costs $100/month on the web and provides five times Pro's allowance per five-hour session, with weekly limits. Upgrade after checking the limits your workload actually reaches; 15 coding sessions per day do not by themselves imply an API bill above $300. MiMo Token Plan Pro's ¥329/month and 38B Credits retain the October 20 scope; convert credits for the model you use.
- $100–200/month (higher usage tier): Claude Max 20x costs $200/month on the web and provides twenty times Pro's allowance per five-hour session, with weekly limits. It does not guarantee 40 coding sessions per day or unthrottled use. Check feature, model and plan restrictions for ChatGPT Pro and MiMo Max separately.
- API only, no monthly fee: DeepSeek Flash uncached input/output rates per million tokens are $0.15/$0.60 off-peak and $0.30/$1.20 at peak; V4 Pro costs $0.66/$1.98 off-peak and $1.32/$3.96 at peak. MiMo V2.5 retains the table's $0.14/$0.28 rates, so the two no longer share a price. Cached input and separate search/fetch calls have their own rates.
Monthly Cost Estimate
Assume 3 rounds per session, each with about 50K input and 5K output tokens: 150K input and 15K output per session over a 30-day month. Estimates exclude cache discounts, search/fetch calls and taxes. The DeepSeek range assumes all off-peak or all peak usage; these are budget scenarios, not actual monthly bills.
| Billing option | Light (5/day) | Moderate (15/day) | Heavy (40/day) |
|---|---|---|---|
| Claude Sonnet 5.5 API | $67.50/mo | $202.50/mo | $540.00/mo |
| DeepSeek Flash API (off-peak to peak) | $4.73 – $9.45/mo | $14.18 – $28.35/mo | $37.80 – $75.60/mo |
| MiMo V2.5 API | $3.78/mo | $11.34/mo | $30.24/mo |
| GPT-6.1 Sol API | $67.50/mo | $202.50/mo | $540.00/mo |
Claude Max web subscriptions cost $100/month for 5x or $200/month for 20x. The official usage limits include five-hour and weekly limits shared with Claude Code, so they cannot be converted directly into the session counts above. Eligible Max users may also claim separate Claude Platform API credits; these cannot fund Claude Code overages or be counted again as subscription coding capacity.
Takeaways:
- Under these assumptions, Sonnet 5.5 and GPT-6.1 Sol each cost about $202.50/month at 15 sessions per day. A fixed subscription fee does not promise the same workload; check its actual limits before comparing spending.
- Recalculated using the October 20 rate, the moderate MiMo example is $11.34/month. DeepSeek Flash costs $14.18–28.35/month for the same usage, depending on timing. Token Plan credits require model-specific conversion, so a tier cannot be declared sufficient from session counts alone.
- At low usage, compare the API estimate with the subscription benefits you actually need; task counts alone cannot establish that either billing option is always cheaper.
All of the above counts text tokens only. If the budget also includes image or video generation allowances, estimating from token prices will underestimate it: images are billed per image or per token (see the AI image generation guide), and video is billed in credits where per-second cost varies more than twofold between models (see the AI video generation comparison for the conversion).
When Not to Pick the Cheapest
- For complex debugging or cross-file refactoring, compare models on your own fixed tasks. Check test results, retry tokens and manual repair time before paying for another model. Token prices and vendor claims do not establish first-attempt success on your workload.
- Caching reduces only eligible repeated input. The official Sonnet 5.5 rates are $2 per million base input tokens and $0.10 for cache reads: 95% less for the cached portion. Cache writes and output cost extra. A total bill cannot be reduced to one third without knowing the hit rate and write volume.
- Both Coding Plans and APIs can have quota, balance and rate limits. Check whether reaching a limit stops work, requires waiting or incurs extra charges, and set a budget. Pay-as-you-go billing does not imply unlimited calls.
- MiMo Token Plan is restricted to coding tools (Claude Code / OpenClaw / OpenCode etc.) — it cannot be used as general-purpose API credits. MiniMax Token Plan can be used as general API.
- Choosing a plan and controlling the credential are separate jobs. If a provider key hosted in OpenCode Console may have leaked, the order of steps is in responding to an OpenCode Console key exposure.
Pricing Reference
Personal Subscriptions
| Product | Price | Coding benefits | Key limits |
|---|---|---|---|
Claude Pro |
$20/month or $200/year | Claude Code and other personal benefits | Five-hour and weekly limits, shared with Claude Code |
Claude Max 5x |
$100/month (web subscription) | 5x Pro allowance per five-hour session | Weekly limits remain; no fixed coding-session guarantee |
Claude Max 20x |
$200/month (web subscription) | 20x Pro allowance per five-hour session | Weekly limits remain; check extra-usage charges separately |
ChatGPT Plus |
$20/mo |
Codex (light), Deep Research | Codex 5h estimate table separate; not a heavy Codex plan |
ChatGPT Pro $100 |
$100/mo |
Codex, Deep Research, Pro models | 5x Plus; exclude expired promotions from budgets |
ChatGPT Pro $200 |
$200/mo |
Highest usage | 20x Plus; don't budget on expired promos |
Google AI Plus |
$4.99/mo (US personal monthly, USD) | Gemini, Gemini Notebook and other benefits | About 2x Free access; compute limits refresh every 5 hours, with a weekly limit |
Google AI Pro |
$19.99/mo (US personal monthly, USD) | Gemini, Gemini Notebook and other benefits | About 4x Free access; availability and limits depend on account and region |
Google AI Ultra |
$99.99/month (5x) or $199.99/month (20x), US personal monthly, USD | Gemini, Gemini Notebook and other benefits | The 5x/20x comparison is against Pro, not Free |
Gemini Code Assist Standard |
$19–22.80/user/mo |
Team IDE assist, CLI, agent mode | Enterprise seat product, not personal subscription |
Google AI prices and multipliers come from the US personal subscription page, checked on October 24, 2026. They are separate from API or enterprise per-seat billing. The old fixed daily prompt counts have been removed; the current page describes compute-based limits, five-hour windows and a weekly limit.
Claude prices and limits were checked on October 24, 2026 against the official Max plan and Claude Code usage documentation. API estimates use standard rates. The GPT-6.1 Sol model page specifies higher rates above 272K input tokens per request; Fast/Ultrafast modes, cache writes, tool calls and regional premiums are excluded here.
Coding Plan / Token Plan
| Product | Price | Quota | Fit |
|---|---|---|---|
MiMo Token Plan Lite |
$6/mo (¥39/mo) |
4.1B Credits; off-peak ×0.8 | Light coding; cheapest entry |
MiMo Token Plan Standard |
$16/mo (¥99/mo) |
11B Credits; off-peak ×0.8 | Convert Credits using the selected model; no fixed session guarantee |
MiMo Token Plan Pro |
$50/mo (¥329/mo) |
38B Credits; off-peak ×0.8 | Heavy coding |
MiMo Token Plan Max |
$100/mo (¥659/mo) |
82B Credits; off-peak ×0.8 | Credits depend on the selected model |
MiniMax Starter (global) |
$10/mo |
1,500 M2.7 requests/5h | Multimodal general, not coding-specific |
MiniMax Starter (China) |
¥29/mo |
600 M2.7 requests/5h | Lowest RMB fixed fee |
MiniMax Plus (global/China) |
$20/mo; ¥49/mo |
4,500 / 1,500 requests/5h | Moderate multimodal |
MiniMax Max (global/China) |
$50/mo; ¥119/mo |
15,000 / 4,500 requests/5h | Heavy multimodal |
Kimi Andante |
¥49/mo |
Unified pool; Kimi Code 1x | Kimi ecosystem users |
Kimi Moderato |
¥99/mo |
Unified pool; Kimi Code 4x | Moderate Kimi use |
Kimi Allegretto |
¥199/mo |
Unified pool; Kimi Code 20x | Includes Kimi Claw benefits |
Kimi Allegro |
¥699/mo |
Unified pool; Kimi Code 60x | Kimi top tier |
Tencent Cloud Lite |
¥40/mo |
~18,000 requests/mo | Request-limited allowance |
Tencent Cloud Pro |
¥200/mo |
~90,000 requests/mo | High volume, clear table |
Alibaba Bailian Pro |
¥200/mo |
90,000 requests/mo | High volume; Lite discontinued |
Z.AI GLM Coding Plan |
from $18/mo |
Lite/Pro/Max: 80–1,600 prompts/5h | GLM-5.1 ecosystem |
MiMo Token Plan is coding-tool only. MiniMax Token Plan works as general API. Bailian/Tencent Cloud/Z.AI are restricted to supported tools.
Pay-as-you-go API
| Platform | Model | Input / Cached input / Output (per 1M tokens) | Notes |
|---|---|---|---|
OpenAI |
GPT-6 Astra | $10 / — / $50 |
1.05M context; higher token rates |
OpenAI |
GPT-6.1 Sol | $2 / $0.10 / $10 |
Standard rates; long context and accelerated modes cost more |
OpenAI |
GPT-6 Luna | $0.10 / — / $0.50 |
High-volume light tasks |
Anthropic |
Claude Fable 5.1 | $10 / $0.25 / $50 |
Hard reasoning and long-horizon agents |
Anthropic |
Claude Opus 5.5 | $4 / $0.20 / $20 |
Coding and knowledge work workhorse |
Anthropic |
Claude Sonnet 5.5 | $2 / $0.10 / $10 |
Speed and intelligence balance |
Anthropic |
Claude Haiku 5.5 | $0.10 / $0.01 / $0.50 |
$0.50/$2.50 above 100K tokens |
DeepSeek |
V4.1 Flash (off-peak) | $0.15 / $0.003 / $0.60 |
deepseek-flash |
DeepSeek |
V4.1 Flash (peak) | $0.30 / $0.006 / $1.20 |
deepseek-flash |
DeepSeek |
V4 Pro (off-peak) | $0.66 / $0.022 / $1.98 |
deepseek-v4-pro |
DeepSeek |
V4 Pro (peak) | $1.32 / $0.044 / $3.96 |
deepseek-v4-pro |
Xiaomi MiMo |
V2.5 | $0.14 / $0.0028 / $0.28 |
MiMo rate retained from October 20, 2026 |
Xiaomi MiMo |
V2.5-Pro | $0.435 / $0.0036 / $0.87 |
MiMo rate retained from October 20, 2026 |
Kimi |
K2.6 | $0.95 / $0.16 / $4.00 |
256K context |
Z.AI |
GLM-5.1 | $1.40 / $0.26 / $4.40 |
Cache storage free for limited time |
Qwen |
3.6 Plus | ¥2 / — / ¥12 (≤256K) |
Long context enters higher tier |
The official DeepSeek pricing page, checked on October 24, 2026, defines peak hours as Monday–Friday 01:00–04:00 and 06:00–10:00 UTC, excluding Chinese public holidays; all other hours are off-peak. Search and Fetch each cost $0.75 per 1,000 successful requests from November 1, 2026 at 00:00 Beijing time; the page says they are free before then. The token estimates above exclude these calls.
MiMo China pricing: V2.5 input ¥1.00 / output ¥2.00; V2.5-Pro input ¥3.00 / output ¥6.00 (per 1M tokens).
Batch and caching need separate estimates. Standard OpenAI and Anthropic Batch input/output rates can be 50% lower; other providers retain the October 20 scope. Caching discounts eligible input, while writes, output and tool calls may still be charged. These discounts do not guarantee a lower total for every coding task.
Sources
- Claude: pricing, Pro usage, Max usage
- ChatGPT: Pro tiers, Codex limits
- Google: AI subscriptions, Gemini limits, Code Assist pricing
- OpenAI API: pricing
- Anthropic API: pricing
- DeepSeek: pricing
- Xiaomi MiMo: API pricing, Token Plan
- MiniMax: global, China
- Kimi: membership, API
- Alibaba Bailian: Coding Plan, model pricing
- Tencent Cloud: Coding Plan
- Z.AI: Coding Plan, pricing
- Volcano Ark: Coding Plan