OpenCode vs Claude Code: pricing and usage limits
Compare OpenCode Go and Go Plus limits, Zen pay-as-you-go prices, and weekly token estimates from our Claude Code and Codex coding runs.
Can I use my Claude plan in OpenCode?
OpenCode is a free, MIT-licensed coding agent made by Anomaly. Model access can cost extra, depending on the provider you connect. OpenCode, Anomaly
OpenCode has no supported Claude Pro or Max sign-in. It removed its built-in Claude authentication plugin in v1.3.0, released on March 22, 2026, following PR #18186, merged on March 19. Anthropic’s Claude Code legal page says third-party developers may not offer Claude.ai login or route requests through users’ Free, Pro or Max credentials. OpenCode providers, OpenCode v1.3.0 release, OpenCode PR #18186, Claude Code legal and compliance
You can use Claude Code with a Claude plan, or use OpenCode with Go at $10 a month, Go Plus at $40 a month, prepaid Console/Zen credits, or your own provider API keys. Go and Go Plus are optional and are for single-member workspaces. OpenCode Go plans, OpenCode Go limits and token prices, OpenCode Console model prices, OpenCode providers
Claude Pro costs $20 a month with monthly billing, or $200 up front for a year, advertised as $17 a month. Max 5x costs $100 a month and Max 20x costs $200 a month, with monthly billing. Claude Code shares the plan’s usage limits with Claude. Claude pricing, What is the Max plan?, Claude Code with Pro or Max
How OpenCode Go limits work
OpenCode publishes a monthly dollar limit for each model. The five-hour limit is 20% of that amount and the weekly limit is 50%. Go Plus uses the same percentages and token prices, with higher monthly limits. These dollar limits describe included usage, not the subscription fee. OpenCode Go limits and token prices
At a limit, you can continue with free models. If you have prepaid Console credit, enabling Extra Usage lets requests use that balance after the Go allowance runs out. OpenCode says usage limits may change. OpenCode Go limits and token prices
The first group contains OpenCode Data’s top three models by tokens over the past seven days, updated 2026-10-25 04:02:43 UTC. Its scope is not stated: the page does not say whether it covers Go, Console or all OpenCode traffic. The second group adds Kimi K3.1 Pro, GLM-5.3 and MiMo-V2.6-Pro, whose $15 Go monthly limits are below the first group’s $20 to $80. OpenCode Data
| Model | Go monthly limit | Go weekly limit | Go 5-hour limit | Go Plus monthly limit | Input / 1M | Output / 1M | Cached read / 1M | Go tokens / week (calculated) | Go Plus tokens / week (calculated) |
|---|---|---|---|---|---|---|---|---|---|
| Most used on OpenCode | |||||||||
| DeepSeek V4.1 Flash | $80 | $40 | $16 | $160 | $0.30 | $1.20 | $0.006 | 360M | 720M |
| MiniMax M3.1 Flash | $20 | $10 | $4 | $80 | $0.15 | $0.60 | $0.015 | 150M | 610M |
| GLM-5.3-Flash | $60 | $30 | $12 | $180 | $0.15 | $0.50 | $0.03 | 420M | 1.2B |
| Kimi, GLM and MiMo models | |||||||||
| Kimi K3.1 Pro | $15 | $7.50 | $3 | $60 | $2.00 | $10.00 | $0.20 | 7.7M | 31M |
| GLM-5.3 | $15 | $7.50 | $3 | $120 | $1.40 | $4.40 | $0.26 | 12M | 93M |
| MiMo-V2.6-Pro | $15 | $7.50 | $3 | $60 | $0.435 | $0.87 | $0.003625 | 65M | 260M |
| Also on Claude plans | |||||||||
| Claude Haiku 5.5 | $15 | $7.50 | $3 | $60 | $0.10 | $0.50 | $0.01 | 140M | 560M |
Prices are USD per one million tokens. We use the lowest listed context tier, including prompts up to 100K tokens for Haiku. DeepSeek uses its higher peak prices; peak hours are 01:00–04:00 and 06:00–10:00 UTC, Monday through Friday. OpenCode Go limits and token prices
The calculation also uses the listed cached-write prices: $2.00 per million tokens for Kimi K3.1 Pro and $0.125 for Haiku 5.5. A dash in OpenCode’s write-price column is treated as no additional write charge, not free input. OpenCode Go limits and token prices
Go’s monthly limit still applies. Two full weekly allowances equal one monthly allowance. Treat each row as usage on that model; do not add the rows together.
Token estimates are calculated, not measured on OpenCode. Published prices retain their listed precision; token estimates use two significant figures. Prices and limits checked 2026-10-25.
Claude Code vs OpenCode Go: weekly tokens
Claude Pro figures extrapolate measured coding runs to a full weekly limit using the same calculation as the home page. Max figures are projections from Pro at 5x and 20x; those per-session multiples are assumed to apply weekly. OpenCode figures are calculated from published Go limits and a reference token mix. What is the Max plan?, Claude Code with Pro or Max
Haiku 5.5 is the shared model in this comparison. Its Go estimate uses its own Claude Pro coding run. The other six Go models use the GPT-6.1 Sol coding run on ChatGPT Plus.
| Model | Claude Pro ($20/month, measured) | Claude Max 5x ($100/month, projected) | Claude Max 20x ($200/month, projected) | OpenCode Go ($10/month, calculated) | OpenCode Go Plus ($40/month, calculated) |
|---|---|---|---|---|---|
| Claude Haiku 5.5 | > 1.0B | Not shown | Not shown | 140M | 560M |
| Claude Opus 5.5 | 74M | 370M | 1.5B | Not on Go | Not on Go |
“Measured” identifies the source run; the weekly total is an extrapolation. A > prefix means more than the displayed figure. Haiku appears only in the API’s Pro benchmark rows, so no Haiku Max projection is shown.
Each Claude column represents using that model for the full weekly allowance, not a separate allowance for every model. Five-hour limits can stop work before the weekly limit.
Claude Pro’s weekly Opus 5.5 usage at API prices (calculated): $230.
At the 2026-10-25 check, Console’s listed Opus input, output and cached-read rates match the home page’s benchmark prices. This weekly value uses the home page’s cache-write convention. Console’s table lists one cache-write price, so what it charges for one-hour cache writes isn’t shown. OpenCode Console model prices, Claude pricing
How the token estimates are calculated
Weekly tokens equal the weekly dollar limit multiplied by the reference run’s total tokens, divided by the cost of that run at the selected Go prices.
weekly tokens = weekly dollar limit × reference-run tokens ÷ reference-run cost
We price uncached input, cached reads, output and any listed cache-write premium separately. Recorded input already includes cache creation, so only the write price above the input price is added. Reasoning tokens are already included in output and are not added again.
For Go, we apply the single listed cached-write rate to both recorded write durations. OpenCode’s table lists only one cache-write price and doesn’t say whether one-hour writes cost more. This is an assumption in the calculation, not an observation of OpenCode billing. OpenCode Go limits and token prices
The home page’s Claude calculation instead retains its existing cache-write convention: five-minute writes add 25% of the input price and one-hour writes add 100%. Its monthly API value is divided by 365.25 / 12 / 7 to obtain the weekly value shown here.
OpenCode Zen (Console): pay as you go
The Zen product page and the current Console documentation describe OpenCode’s pay-as-you-go model access. OpenCode says Zen charges per request at provider prices without a markup. Console uses prepaid credits; paid requests fail with HTTP 402 when the balance reaches zero. OpenCode Zen, OpenCode Console model prices, OpenCode Console billing
A processing fee of 4.6% plus $0.31 is added to each credit purchase. A $20 balance top-up therefore costs $21.23: $20 in credits and a $1.23 fee. OpenCode Console billing, OpenCode Zen
| Model | Input | Output | Cached read | Cached write |
|---|---|---|---|---|
| Claude Opus 5.5 | $4.00 | $20.00 | $0.20 | $5.00 |
| Claude Sonnet 5.5 | $2.00 | $10.00 | $0.20 | $2.50 |
These are Console’s published prices, checked 2026-10-25. For Sonnet 5.5, the captured Claude pricing page lists a $0.10 cached-read price, while Console lists $0.20. Use the Console figure when reading this table. OpenCode Console model prices, Claude pricing
You can also use Go inside Claude Code through OpenCode’s Anthropic-compatible endpoint, with a key from a Go workspace. Only models that accept Anthropic Messages work in that setup; it uses OpenCode billing rather than your Claude subscription allowance. Follow OpenCode’s Claude Code setup guide.
What this comparison leaves out
tokenusage.dev has not measured OpenCode. The calculated estimates borrow token mixes from Claude Code and Codex runs; another agent, task or cache hit rate can produce different results. Tokens from different models do not represent equal amounts of work or equal answer quality.
Long-context requests can cost more than the lowest tiers used here. Go’s five-hour and monthly limits also apply, and prices, model availability and usage limits may change. OpenCode Data’s traffic scope is not stated. OpenCode Go limits and token prices, OpenCode Data
This page uses the coding benchmark records in our subscription API, not figures from shared reports. Read how the numbers work, compare Claude Pro and Max, or compare Cursor, Claude Code and Codex.
Sources
Sources checked 2026-10-25. OpenCode Data was updated 2026-10-25 04:02:43 UTC.
- OpenCode
- Anomaly
- OpenCode providers
- OpenCode v1.3.0 release
- OpenCode PR #18186
- Claude Code legal and compliance
- OpenCode Go plans
- OpenCode Go limits and token prices
- OpenCode Data
- OpenCode Zen
- OpenCode Console model prices
- OpenCode Console billing
- OpenCode’s Claude Code setup guide
- Claude pricing
- What is the Max plan?
- Claude Code with Pro or Max
tokenusage.dev is not affiliated with, endorsed by or sponsored by Anthropic or Anomaly.