Ask
74

glm-5.2:cloud burned ~15% of my Ollama 5-hour session limit on a single question

Ollama Pro, $20/month, running glm-5.2:cloud from the VS Code extension.

One agent turn. About three minutes wall clock. One question: compare two versions of the same Markdown spec (24k words and 18k words) and list the differences. Nothing else was in the workspace, the extension had access to nothing else.

That single turn took roughly 15% of my 5-hour session allowance. At that rate the whole window is gone in about twenty minutes. I have run much heavier things than this on the same plan for weeks and never come near the limit.

I understand the meter in principle. Ollama's pricing page says each plan has session limits that reset every 5 hours and weekly limits that reset every 7 days, and that usage is based on the model plus the number of input, cached input and output tokens processed rather than a fixed token cap. Fine. What I cannot work out is which of these is true:

  • glm-5.2 is simply far more expensive per turn than what I was running before
  • something about the metering changed recently
  • the VS Code extension is quietly resending far more context than I think it is

The extension shows me no token count anywhere I can find, so I am guessing. Has anyone actually measured a single turn on glm-5.2:cloud against the same prompt on a different cloud model?

11 answers Share
Report

Answering anonymously — a moderator will review it first.

  • @p99_hana · 2h ago · 3 replies

    There are two things stacked here and they multiply, so you will not be able to tell them apart by feel.

    First, the meter. Ollama's own wording is that limits are based on the model and the number of input, cached input and output tokens processed, and that they deliberately do not cap you at a fixed token count because different models use different amounts of compute. So the real unit is closer to GPU time than to tokens, and a large thinking model generates a lot of it per turn.

    Second, your prompt. 24k plus 18k words is somewhere in the region of 55-60k tokens of input before any system prompt, tool schemas or file listings. Then it is a thinking model, so the reasoning tokens you never see count as output. "Find every difference between these two documents" is precisely the task shape that makes a reasoning model produce an enormous amount of it.

    Before you conclude the metering changed, get a number. On my machine ollama run --verbose glm-5.2:cloud prints prompt and eval counts after the response — confirm that on yours. Run your exact prompt once through the CLI, write the counts down, then run the same prompt through the extension and compare how far the session bar moves each time. If the CLI turn is cheap and the extension turn is expensive, it is your harness and not the model.

    41
    Share
    Reply

    Answering anonymously — a moderator will review it first.

    Report
    • @first_repo_finn · 2h ago

      That is a fair split and I can run it tonight. --verbose does print counts here for the cloud tag too, I had just never looked at them because I assumed cloud meant no local metrics.

      12
      Share
      Reply

      Answering anonymously — a moderator will review it first.

      Report
    • @p99_hana · 2h ago

      Do both from a fresh session. If you were iterating in an existing chat, the extension may be replaying the entire thread every turn, in which case "one question" is really one question plus everything that came before it, re-billed.

      9
      Share
      Reply

      Answering anonymously — a moderator will review it first.

      Report
  • @pinned_versions · 2h ago · 2 replies

    Look at the model page before you make anything your default.

    The library page for glm-5.2 lists it at 756B parameters with a 976K context window and marks it as high usage. deepseek-v4-flash's cloud tag, by comparison, is marked medium usage. Those labels are Ollama telling you in advance roughly what a turn will do to your allowance, and people scroll straight past them.

    Also relevant to your options: glm-5.2 is cloud-only there. The only tag listed is glm-5.2:cloud. There is no local variant to fall back to when you run out, so every single turn is metered.

    For a document diff I would sit in the medium band by default and only escalate when the cheaper model visibly fails.

    33
    Share
    Reply

    Answering anonymously — a moderator will review it first.

    Report
    • @nadia_brill · 2h ago

      Worth adding that the label describes the model, not your prompt. A medium-usage model that thinks for four thousand tokens can cost you more than a high-usage one that answers in two hundred. The label narrows the search. It does not finish it.

      8
      Share
      Reply

      Answering anonymously — a moderator will review it first.

      Report
  • @idempotent_ian · 5h ago

    My money is on the harness, not the model.

    Editor extensions attach context generously: open tabs, recently touched files, a workspace tree, tool schemas, and then a rolling transcript on top. You believe you sent two files. You may have sent two files per turn plus every previous turn.

    The test is cheap. Same prompt, same model, from a plain terminal, brand new session, only the two files in the prompt. If the cost falls off a cliff, stop blaming GLM.

    Separately: for a diff task specifically, do not hand a language model 60k tokens of prose and ask it to spot changes. Run an actual diff first and give the model only the changed hunks plus a little surrounding context. Same answer, a fraction of the input, and it is more accurate because the model is not doing string comparison in its head.

    27
    Share
    Reply

    Answering anonymously — a moderator will review it first.

    Report
  • @thermal_theo · 2h ago

    Setting the metering question aside: 42,000 words of Markdown is not a language model problem. git diff, or diff -u, or any three-way merge tool gives you a perfect answer for free in about forty milliseconds.

    Use the model on the output of that. "Here are the changed hunks, tell me what changed semantically and what it means for the implementation" costs you maybe two percent of what today cost, and the result is better because the model is reasoning about meaning instead of playing spot-the-difference.

    15
    Share
    Reply

    Answering anonymously — a moderator will review it first.

    Report
  • @tabula_rasa_dev · 2h ago

    One practical note if your plan is "I will just upgrade my way out of this": when I looked, Max was showing new signups as paused, so upgrading past Pro may not be on the table today.

    What is on the table on Pro is adding extra usage balance — the plan's included limits get consumed first and then it draws from that balance. Worth knowing before you sit down for a long session and discover at hour two that you are finished for the next three.

    9
    Share
    Reply

    Answering anonymously — a moderator will review it first.

    Report
  • @csv_apologist · 2h ago · 3 replies

    Pro is not a token bucket, it is a concurrency limit. You cannot "use up 15%" of anything — you just get queued if you run too many models at once. Something else is going on here.

    6
    Share
    Reply

    Answering anonymously — a moderator will review it first.

    Report
    • @row_level_sam · 2h ago

      That is not right, and this misunderstanding is exactly why people get surprised.

      Concurrency is a separate line on the plan — Free one concurrent model, Pro three, Max ten. On top of that, the pricing page explicitly describes session limits that reset every 5 hours and weekly limits that reset every 7 days, and it describes Pro and Max users being able to add extra usage balance which is drawn on after the included limits are spent. You cannot add extra balance to a thing that does not run out.

      22
      Share
      Reply

      Answering anonymously — a moderator will review it first.

      Report
    • @csv_apologist · 2h ago

      Fair enough, I was reasoning from how it behaved when cloud was still in preview. Corrected.

      4
      Share
      Reply

      Answering anonymously — a moderator will review it first.

      Report