Skip to content

Model vs model

Kimi K3 vs GLM 5.3

Kimi K3 and GLM 5.3 are both in llmwise. What each gets on every plan, what it reads, how it's served and what happens when its provider fails, from the catalog and the code that runs them.

Model prices and specs checked against OpenRouter's Kimi K3 page, OpenRouter's GLM 5.3 page. Updated .

Short answer

GLM 5.3 gets twice as many messages: up to 250 messages a month on Pro, against up to 125 messages a month for Kimi K3. Otherwise, only Kimi K3 reads images. In our test runs, Kimi K3 passed 46 of the 50 prompts both answered and GLM 5.3 48; 4 prompts split them, most on summarization (3 to 4).

Kimi K3 vs GLM 5.3, prompt by prompt

Every prompt Kimi K3 and GLM 5.3 both answered, compared directly, their biggest differences first. One run each, through OpenRouter: a wait depends on the provider and the load that day, so a lead under 10% counts as close.

Of the 50 prompts both answered, both passed 45, only Kimi K3 passed 1, only GLM 5.3 passed 3, and neither passed 1. Kimi K3 answered sooner on 2 of the 50 and GLM 5.3 on 44; the rest were within 10% of each other. The 50 replies cost $0.1938 on Kimi K3 and $0.0364 on GLM 5.3: 5.3× less on GLM 5.3.

The 4 prompts only one of Kimi K3 and GLM 5.3 passed

  • A product announcement with five rules (writing): GLM 5.3 passed and Kimi K3 didn't. Kimi K3: Graded 4.3 of 5 on average (lowest 4); but doesn't end with a question. GLM 5.3: Graded 4.3 of 5 on average (lowest 4).

  • Argue both sides of free buses (writing): Kimi K3 passed and GLM 5.3 didn't. Kimi K3: Graded 4.7 of 5 on average (lowest 4). GLM 5.3: Graded 3.7 of 5 on average (lowest 3); but a paragraph of 92 words, over the 90 allowed.

  • An email thread in one sentence (summarization): GLM 5.3 passed and Kimi K3 didn't. Kimi K3: Graded 5.0 of 5 on average (lowest 5); but 32 words, over the 30 allowed. GLM 5.3: Graded 4.3 of 5 on average (lowest 3).

  • A refund request outside the window (customer support): GLM 5.3 passed and Kimi K3 didn't. Kimi K3: Graded 4.0 of 5 on average (lowest 2). GLM 5.3: Graded 4.7 of 5 on average (lowest 4).

Job by job, the widest gaps first

  • Customer support: Kimi K3 passed 4 of 5 and GLM 5.3 5 of 5. GLM 5.3 answered 1.9× sooner at the median, 4.3 s against 2.3 s. GLM 5.3 cost 9.8× less, $0.0228 against $0.0023 for the 5 replies. Kimi K3's replies ran 16% longer, in tokens of reply, thinking not counted.

  • Summarization: Kimi K3 passed 3 of 5 and GLM 5.3 4 of 5. GLM 5.3 answered 3.0× sooner at the median, 3.4 s against 1.1 s. GLM 5.3 cost 6.3× less, $0.0172 against $0.0027 for the 5 replies. Kimi K3's replies ran 14% longer, in tokens of reply, thinking not counted.

  • Math: Kimi K3 passed 5 of 5 and GLM 5.3 5 of 5. GLM 5.3 answered 5.7× sooner at the median, 5.4 s against 0.9 s. GLM 5.3 cost 10.6× less, $0.0187 against $0.0018 for the 5 replies. Their replies ran to about the same length.

  • SQL: Kimi K3 passed 5 of 5 and GLM 5.3 5 of 5. GLM 5.3 answered 1.4× sooner at the median, 1.6 s against 1.2 s. GLM 5.3 cost 7.5× less, $0.0118 against $0.0016 for the 5 replies. Kimi K3's replies ran 19% longer, in tokens of reply, thinking not counted.

  • Writing: Kimi K3 passed 4 of 5 and GLM 5.3 4 of 5. GLM 5.3 answered 1.7× sooner at the median, 3.0 s against 1.8 s. GLM 5.3 cost 6.7× less, $0.0192 against $0.0029 for the 5 replies. Their replies ran to about the same length.

  • Agents and tool use: Kimi K3 passed 5 of 5 and GLM 5.3 5 of 5. GLM 5.3 answered 2.2× sooner at the median, 2.0 s against 0.9 s. GLM 5.3 cost 6.1× less, $0.0135 against $0.0022 for the 5 replies. Kimi K3's replies ran 25% longer, in tokens of reply, thinking not counted.

  • Translation: Kimi K3 passed 5 of 5 and GLM 5.3 5 of 5. GLM 5.3 answered 3.3× sooner at the median, 5.2 s against 1.6 s. GLM 5.3 cost 5.1× less, $0.0179 against $0.0035 for the 5 replies. Kimi K3's replies ran 34% longer, in tokens of reply, thinking not counted.

  • Coding: Kimi K3 passed 5 of 5 and GLM 5.3 5 of 5. GLM 5.3 answered 3.9× sooner at the median, 4.1 s against 1.1 s. GLM 5.3 cost 4.3× less, $0.0316 against $0.0073 for the 5 replies. Kimi K3's replies ran 17% longer, in tokens of reply, thinking not counted.

  • Data analysis: Kimi K3 passed 5 of 5 and GLM 5.3 5 of 5. GLM 5.3 answered 1.8× sooner at the median, 3.5 s against 1.9 s. GLM 5.3 cost 3.5× less, $0.0339 against $0.0096 for the 5 replies. Kimi K3's replies ran 15% longer, in tokens of reply, thinking not counted.

  • RAG and answering from documents: Kimi K3 passed 5 of 5 and GLM 5.3 5 of 5. GLM 5.3 answered 4.1× sooner at the median, 2.6 s against 0.6 s. GLM 5.3 cost 2.9× less, $0.0073 against $0.0025 for the 5 replies. Kimi K3's replies ran 47% longer, in tokens of reply, thinking not counted.

All 50 prompts: who passed, who answered sooner, who cost less
Kimi K3 and GLM 5.3 on each prompt of our test runs
PromptResultSoonerCheaper
Turn a title into a URL slugBoth passedGLM 5.3, 5.9×took 2.6 s and 0.4 sGLM 5.3, 3.4×cost $0.0035 and $0.0010
Parse a duration like “1h 30m”Both passedGLM 5.3, 6.7×took 7.1 s and 1.1 sGLM 5.3, 4.4×cost $0.0063 and $0.0014
Merge overlapping intervalsBoth passedGLM 5.3, 2.4×took 2.0 s and 0.8 sGLM 5.3, 4.4×cost $0.0036 and $0.0008
Evaluate an arithmetic expression, no evalBoth passedKimi K3, 1.3×took 9.9 s and 12.5 sGLM 5.3, 6.6×cost $0.0113 and $0.0017
Parse CSV with quoted fieldsBoth passedGLM 5.3, 1.6×took 4.1 s and 2.6 sGLM 5.3, 2.9×cost $0.0069 and $0.0024
Announce a second bakery shop on LinkedInBoth passedGLM 5.3, 1.4×took 3.9 s and 2.9 sGLM 5.3, 12.2×cost $0.0046 and $0.0004
Rewrite corporate jargon in plain wordsBoth passedGLM 5.3, 1.1×took 2.0 s and 1.8 sGLM 5.3, 15.0×cost $0.0030 and $0.0002
Decline a meeting and offer two timesBoth passedGLM 5.3, 2.1×took 3.0 s and 1.4 sGLM 5.3, 12.6×cost $0.0034 and $0.0003
A product announcement with five rulesOnly GLM 5.3GLM 5.3, 2.7×took 2.3 s and 0.8 sGLM 5.3, 4.3×cost $0.0033 and $0.0008
Argue both sides of free busesOnly Kimi K3GLM 5.3, 2.0×took 3.6 s and 1.8 sGLM 5.3, 3.9×cost $0.0049 and $0.0013
A discount, then sales taxBoth passedGLM 5.3, 5.3×took 5.0 s and 0.9 sGLM 5.3, 18.3×cost $0.0037 and $0.0002
Pens at 3 for $4Both passedGLM 5.3, 3.6×took 5.4 s and 1.5 sGLM 5.3, 26.0×cost $0.0046 and $0.0002
Compound interest over three yearsBoth passedGLM 5.3, 10.2×took 5.9 s and 0.6 sGLM 5.3, 8.1×cost $0.0044 and $0.0005
Four-digit numbers whose digits sum to 9Both passedGLM 5.3, 3.2×took 6.0 s and 1.9 sGLM 5.3, 12.8×cost $0.0032 and $0.0003
The highest of three dice is a 5Both passedGLM 5.3, 2.8×took 2.4 s and 0.9 sGLM 5.3, 4.7×cost $0.0027 and $0.0006
An article in three bulletsNeither passedClosetook 2.8 s and 2.8 sGLM 5.3, 6.3×cost $0.0028 and $0.0004
An email thread in one sentenceOnly GLM 5.3GLM 5.3, 6.2×took 6.4 s and 1.0 sGLM 5.3, 12.7×cost $0.0030 and $0.0002
Decisions and action items from a meetingBoth passedGLM 5.3, 2.2×took 3.4 s and 1.5 sGLM 5.3, 18.5×cost $0.0044 and $0.0002
A quarterly memo for the CEOBoth passedGLM 5.3, 14.2×took 15.7 s and 1.1 sGLM 5.3, 5.0×cost $0.0049 and $0.0010
A study with a negative resultBoth passedGLM 5.3, 3.9×took 3.2 s and 0.8 sGLM 5.3, 2.6×cost $0.0021 and $0.0008
The region with the most revenueBoth passedClosetook 2.2 s and 2.1 sGLM 5.3, 12.4×cost $0.0049 and $0.0004
Average order value in AugustBoth passedGLM 5.3, 1.4×took 2.7 s and 1.9 sGLM 5.3, 14.7×cost $0.0054 and $0.0004
Revenue change from July to AugustBoth passedGLM 5.3, 4.6×took 6.4 s and 1.4 sGLM 5.3, 4.1×cost $0.0064 and $0.0016
A median, filtered two waysBoth passedGLM 5.3, 3.7×took 3.5 s and 1.0 sGLM 5.3, 3.9×cost $0.0042 and $0.0011
Correlation between ad spend and sign-upsBoth passedGLM 5.3, 3.6×took 20.2 s and 5.6 sGLM 5.3, 2.1×cost $0.0131 and $0.0062
A late orderBoth passedClosetook 2.5 s and 2.6 sGLM 5.3, 11.5×cost $0.0045 and $0.0004
A return inside the windowBoth passedGLM 5.3, 2.7×took 4.3 s and 1.6 sGLM 5.3, 14.3×cost $0.0033 and $0.0002
A frustrated customerBoth passedGLM 5.3, 2.5×took 5.6 s and 2.3 sGLM 5.3, 20.2×cost $0.0063 and $0.0003
A refund request outside the windowOnly GLM 5.3GLM 5.3, 2.3×took 2.6 s and 1.1 sGLM 5.3, 4.9×cost $0.0045 and $0.0009
A message with a planted instructionBoth passedGLM 5.3, 2.2×took 11.5 s and 5.3 sGLM 5.3, 8.7×cost $0.0041 and $0.0005
A delivery message into SpanishBoth passedGLM 5.3, 2.0×took 2.9 s and 1.5 sGLM 5.3, 13.6×cost $0.0031 and $0.0002
A product description into FrenchBoth passedGLM 5.3, 3.7×took 5.4 s and 1.5 sGLM 5.3, 12.4×cost $0.0037 and $0.0003
A meeting note into GermanBoth passedGLM 5.3, 1.9×took 3.0 s and 1.6 sGLM 5.3, 7.5×cost $0.0014 and $0.0002
Idioms into natural JapaneseBoth passedGLM 5.3, 3.0×took 5.2 s and 1.7 sGLM 5.3, 4.5×cost $0.0053 and $0.0012
A lease clause into Brazilian PortugueseBoth passedGLM 5.3, 5.9×took 10.1 s and 1.7 sGLM 5.3, 2.7×cost $0.0044 and $0.0016
Customers in one countryBoth passedGLM 5.3, 1.4×took 1.1 s and 0.8 sGLM 5.3, 13.3×cost $0.0018 and $0.0001
Count orders by statusBoth passedGLM 5.3, 1.3×took 1.5 s and 1.2 sGLM 5.3, 6.6×cost $0.0009 and $0.0001
Revenue by categoryBoth passedGLM 5.3, 1.3×took 1.6 s and 1.2 sGLM 5.3, 9.2×cost $0.0019 and $0.0002
Every customer, even those without ordersBoth passedGLM 5.3, 2.6×took 2.5 s and 1.0 sGLM 5.3, 4.3×cost $0.0031 and $0.0007
Monthly revenue with a running totalBoth passedKimi K3, 1.3×took 2.5 s and 3.2 sGLM 5.3, 11.0×cost $0.0042 and $0.0004
A fact from one sectionBoth passedGLM 5.3, 7.3×took 6.5 s and 0.9 sGLM 5.3, 3.8×cost $0.0010 and $0.0003
Core hours and start timesBoth passedGLM 5.3, 2.4×took 1.2 s and 0.5 sGLM 5.3, 1.5×cost $0.0011 and $0.0007
Two sections in one answerBoth passedGLM 5.3, 5.3×took 2.6 s and 0.5 sGLM 5.3, 4.3×cost $0.0020 and $0.0005
A later amendment changes the answerBoth passedGLM 5.3, 3.8×took 4.8 s and 1.3 sGLM 5.3, 3.7×cost $0.0022 and $0.0006
A question the handbook doesn't answerBoth passedGLM 5.3, 1.5×took 0.9 s and 0.6 sGLM 5.3, 2.2×cost $0.0010 and $0.0005
Pick the tool and work out the dateBoth passedGLM 5.3, 4.2×took 3.7 s and 0.9 sGLM 5.3, 14.6×cost $0.0032 and $0.0002
Convert a currencyBoth passedGLM 5.3, 2.6×took 2.9 s and 1.1 sGLM 5.3, 18.2×cost $0.0025 and $0.0001
Book a meeting from a sentenceBoth passedClosetook 1.6 s and 1.7 sGLM 5.3, 5.6×cost $0.0031 and $0.0006
Search, but don't bookBoth passedGLM 5.3, 3.7×took 2.0 s and 0.5 sGLM 5.3, 6.3×cost $0.0034 and $0.0005
Two calls with a unit conversionBoth passedGLM 5.3, 2.4×took 1.7 s and 0.7 sGLM 5.3, 1.6×cost $0.0012 and $0.0008

Kimi K3 vs GLM 5.3 in our test runs

Kimi K3 and GLM 5.3 on the same prompts, job by job: how many replies passed their check.

Based on 100 of our test runs on , through OpenRouter with the app's own prompt and settings. Updated .

Kimi K3 and GLM 5.3 in our test runs, job by job
JobKimi K3GLM 5.3
Coding5 of 55 of 5
Writing4 of 54 of 5
Math5 of 55 of 5
Summarization3 of 54 of 5
Data analysis5 of 55 of 5
Customer support4 of 55 of 5
Translation5 of 55 of 5
SQL5 of 55 of 5
RAG and answering from documents5 of 55 of 5
Agents and tool use5 of 55 of 5
All jobs46 of 50 · 1 message each · $0.0039 a reply48 of 50 · 1 message each · $0.0007 a reply

The same prompts, sent to both through llmwise's own pipeline and checked the same way: five for each job. A bold count is the one that passed more. Messages: what each reply counted as on Pro. Cost: what OpenRouter charged us per reply, on average. Every prompt, and how it's scored.

Where they differed most, summarization: both replies to a prompt only one of them passed.

An everyday prompt: “An email thread in one sentence”, in full.

  • Kimi K3

    Failed: Graded 5.0 of 5 on average (lowest 5); but 32 words, over the 30 allowed.

    Due to a paper shortage delaying brochure shipping to the 18th, Rosa and Idris arranged to pick them up themselves, booking a van for the 18th ahead of the 20th trade fair.

    574 tokens in, 65 out (9 of them reasoning) · 6.4 s · $0.0030 · 1 message on Pro · answered by moonshotai/kimi-k3 via Fireworks ·

  • GLM 5.3

    Passed: Graded 4.3 of 5 on average (lowest 3).

    The brochure shipment was delayed to the 18th due to a paper shortage, so Rosa will pick them up herself, with Idris booking a van for that morning.

    503 tokens in, 37 out · 1.0 s · $0.0002 · 1 message on Pro · answered by z-ai/glm-5.3 via Baidu ·

Kimi K3 and GLM 5.3 on every plan

Whether the one-time free trial reaches each model, then each paid plan's messages on it.

Messages on Kimi K3 and GLM 5.3, plan by plan
PlanPriceKimi K3GLM 5.3
Free$0In the one-time trial of 5 messagesIn the one-time trial of 5 messages
Pro$20 a monthUp to 125 a monthUp to 250 a month
Max$50 a monthUp to 400 a monthUp to 800 a month
Ultra$100 a monthUp to 900 a monthUp to 1,800 a month
Studio$200 a monthUp to 2,000 a monthUp to 4,000 a month

Prices don't include tax, which is added where it applies and shown before you pay. A paid plan's month is one allowance shared by every model, so each monthly count is the most you get if all of it goes to that model. It renews each billing period; everyday models refill daily at 00:00 UTC. Long chats count more per reply. How pricing works.

Every limit is published. Paid plans also have a monthly fair-use limit on AI cost: Pro $7.50, Max $20, Ultra $42, Studio $85. Using every message on your plan at typical sizes stays under it; very large messages and heavy research use it faster. Every limit, explained.

What differs

  • Messages on Pro

    GLM 5.3 gets twice as many messages: up to 250 messages a month on Pro, against up to 125 messages a month for Kimi K3.

  • Context window

    Both take up to 1.05M tokens of context. A chat in llmwise holds up to 200k tokens, which fits in either, so the difference shows only through each maker's own API.

  • Images and PDFs

    GLM 5.3 doesn't read images. Kimi K3 and GLM 5.3 get a PDF's text rather than the file itself.

  • On Free

    Both are in the free trial.

  • Where messages go

    Kimi K3: Served through OpenRouter, only by hosts that don't store or train on prompts. The maker's own endpoint is never asked. GLM 5.3: Served through OpenRouter, only by hosts that don't store or train on prompts. The maker's own endpoint is never asked.

Fact by fact

Kimi K3 and GLM 5.3, fact by fact
FactKimi K3GLM 5.3
Context window1.05M tokens1.05M tokens
Reads imagesYesNo
PDFsText onlyText only
ReasoningYesYes
API price (September 2026)$3.00 in / $15.00 out per million tokens$1.40 in / $4.40 out per million tokens
A typical message at API prices (4,000 tokens in, 700 out)$0.0225$0.0087
A $10 top-up adds100 messages200 messages
Where a message goesServed through OpenRouter, only by hosts that don't store or train on prompts. The maker's own endpoint is never asked.Served through OpenRouter, only by hosts that don't store or train on prompts. The maker's own endpoint is never asked.
If the provider failsWhen one host is down, OpenRouter moves the request to another host that meets the same rules.When one host is down, OpenRouter moves the request to another host that meets the same rules.
Anthropic's safety fallbackDoesn't applyDoesn't apply
API prices are what our model catalog lists (Kimi: Moonshot's list price; GLM: Z.ai's list price). In llmwise you pay per message, not per token: the counts above are what you get.

Kimi K3 or GLM 5.3?

From the facts above and our test runs: the rest is how their answers suit your work, which one chat can show you.

  • Pick Kimi K3: it reads images.

  • Pick GLM 5.3: more messages for the money, up to 250 messages a month on Pro.

Each model's page, the families, and other pairs

Kimi K3 vs GLM 5.3 is one pair of models. The page below covers the whole families.

Questions

Is Kimi K3 or GLM 5.3 cheaper in llmwise?

GLM 5.3 gets twice as many messages: up to 250 messages a month on Pro, against up to 125 messages a month for Kimi K3. Every paid plan's monthly allowance is shared by all models, so each count is the most you get if it all goes to that model.

Can I try Kimi K3 and GLM 5.3 for free?

Yes: both are in the free trial of 5 messages.

Which has the bigger context window, Kimi K3 or GLM 5.3?

Neither: both take 1.05M tokens. A chat in llmwise holds up to 200k tokens, which fits in either, so the difference shows only through each maker's own API.

Can I use Kimi K3 and GLM 5.3 in the same chat?

Yes. Pick Kimi K3 for one message and GLM 5.3 for the next; the second sees the whole chat, including the first one's answer.

Which did better in your test runs, Kimi K3 or GLM 5.3?

On the same 50 prompts, run on September 27, 2026, Kimi K3 passed 46 and GLM 5.3 passed 48. The table on this page has each job, and every reply is published.

Claude, GPT, Gemini, DeepSeek, Grok, Kimi, and GLM, in one chat.

See what a message costs before you send it. Free is 5 messages to try; sign in with an email link, no password or card.