OpenCode is selling $500 of GLM 5.3 usage for $10 until the end of the week. The open-source coding agent posted the offer on X at 08:25 UTC on November 18, 2026, the third and largest step in about a day: “$10 for $500 GLM 5.3 till end of the week”.
The first step came on November 17 at 08:00 UTC: “$100 of GLM 5.3 usage for $10”, “available to all OpenCode Go subscribers”. Eleven hours later OpenCode wrote “we’re bored we’re just gonna do phase 2 now” and doubled it to “$200 of GLM 5.3 usage for $10”.
OpenCode, the open-source coding agent, is selling $500 of GLM 5.3 usage for $10 to OpenCode Go subscribers until the end of the week.
It is the third offer in about a day: $100 for $10 at 08:00 UTC on November 17, $200 for $10 at 18:48 UTC, and $500 for $10 at 08:25 UTC on November 18.
The three posts had about 773,000, 1.16 million and 622,000 views on X when we read them on November 18.
At Z.ai’s own list price on OpenRouter ($1.40 per million input tokens, $4.40 per million output), $500 buys about 114 million output tokens or 357 million input tokens.
OpenCode has not said in these posts who pays the difference or how many subscribers can take the offer.
§ 01The three offers
| Phase | Posted (UTC) | You pay | GLM 5.3 usage | Views on X |
|---|---|---|---|---|
| 1 | Nov 17, 08:00 | $10 | $100 | About 773,000 |
| 2 | Nov 17, 18:48 | $10 | $200 | About 1.16 million |
| 3 | Nov 18, 08:25 | $10 | $500 | About 622,000 |
Each step is a bigger multiple of the same $10: ten times, then twenty, then fifty. The second post, the “phase 2” one, drew the most attention of the three.
§ 02What $500 of GLM 5.3 buys
OpenCode prices the offer in dollars, not tokens. To turn it into model use we read Z.ai’s own listing for GLM 5.3 on OpenRouter on November 18: $1.40 per million input tokens and $4.40 per million output tokens, with a context of about one million tokens. Other hosts on OpenRouter charge less, some under $0.10 per million input tokens, so the same $500 would stretch further there.
| If the $500 goes to | Tokens |
|---|---|
| Input only | About 357 million |
| Output only | About 114 million |
| Three input tokens for each output token | About 233 million |
A coding agent reads far more than it writes, which is why the three-to-one row is the useful one: about 233 million tokens for $10.
§ 03What OpenCode has not said
The posts leave three questions open. They do not say who covers the gap between $10 and $500, whether Z.ai or OpenCode. They do not say how many subscribers can take the offer, or whether it stacks with the earlier phases. And “till end of the week” has no exact cutoff.
OpenCode is running a second promotion beside it. Tencent’s Hy4 Flash has been free on OpenCode Go since November 16, and on November 18 OpenCode wrote: “Still exclusively free on OpenCode for the next 10 days”.
§ 04Where CellCog sits
CellCog does not route any work to GLM 5.3. Our agents run Claude Opus 5.5 on Flash and Core, and Claude Fable 5.2 on Agent Max and Creative Max. For more on GLM 5.3 itself, see our post on GLM 5.3 for AI agents.
§ 05Sources
- OpenCode on X: Phase 1, Phase 2, the $500 offer, Hy4 Flash, read November 18, 2026.
- OpenRouter, GLM 5.3 endpoints, read November 18, 2026.
Q1What is GLM 5.3?
Z.ai’s open-weight model, listed on OpenRouter with a context of about one million tokens. Z.ai lists it at $1.40 per million input tokens and $4.40 per million output tokens there.
Q2What is OpenCode Go?
OpenCode’s paid subscription for model access inside its open-source coding agent. The GLM offers were made to Go subscribers.
Q3How long does the $500 offer last?
Until the end of the week, per OpenCode’s November 18 post. It does not give an exact cutoff time.
Q4Is anything else free on OpenCode Go right now?
Tencent’s Hy4 Flash. OpenCode said on November 16 that usage would be near unlimited and free for 14 days, and on November 18 that it is still free for the next 10 days.
Q5Does CellCog use GLM 5.3?
No. CellCog runs Claude Opus 5.5 on Flash and Core and Claude Fable 5.2 on Agent Max and Creative Max.
