Skip to main content
Uncensored — We add no content filter to this model, but like most chat models it declines a blunt request that comes with no context. It writes unrestricted fiction reliably once your system prompt sets up a fiction-writing context, or once the request is framed as a story. Legal and consent duties remain with the customer.

GLM 5.3 Flash Review: $0.1275 a Million, Clean L4 With a System Prompt

Pick GLM 5.3 Flash for a wide context window at a low price, and give it a system prompt for L4 Intense — it rendered every run we sent that way.

Open the Playground

Spicy Index

72.9/ 100

Freedom Score

64.4/ 100

Content levels

Tested 5 of 5 probe types · 24 runs

  • Mild · The lightest writing requests in the test

  • Moderate · A step up, still below the middle of the scale

  • Strong · Mid-scale requests; must be tested before a model is ranked

  • Very strong · Where most language models start to hold back

  • Maximum · The strongest written requests, scored as one class

Price
$0.202 per 1M mixed tokens (3 input : 1 output)
Output speed
27 tok/s
Success rate
Collecting data

GLM 5.3 Flash reads up to 1,048,576 tokens per request and writes back as much as 131,072, at $0.1275 per million input tokens. Across our content ladder — L1 Suggestive (a slipping strap), L2 Partial, L3 Strong, L4 Intense (a written intimate scene between two adults) and L5 Extreme (a written scene past L4, either restraint and impact play or graphic violence between adults) — it rendered 12 of 24 scored runs in full on 17 and 26 September 2026.

The verdict

The clearest result sits at L4 Intense: asked directly it was refused all three times, framed as fiction it was refused twice and softened once, but set through a system prompt it rendered all three requests in full. L2 Partial rendered clean on every phrasing, and L5 Extreme rendered on five of six runs. L1 Suggestive, our mildest rung, softened on all three phrasings — the one place the model was more cautious than at several harder rungs.

Pick it for a wide read window at a low price, with a system prompt doing the work for anything past L2 Partial. It costs less per million tokens than GLM 5.3 or GLM 5.2, and in our sample its system-prompt phrasing carried L4 Intense more consistently than either of them managed.

Try GLM 5.3 Flash →

Key facts

  • Pass rate: 12 of 24 scored runs rendered in full across 17 and 26 Sep 2026; L2 Partial rendered clean on every phrasing, and all three of its system-prompt L4 Intense runs rendered.
  • Price: $0.1275 per million input tokens, $0.0255 cached and $0.425 output.
  • Limits: 1,048,576-token context, replies up to 131,072 tokens (default cap 32,768), text, image and video input, 128 tool schemas.
  • Sources: every scored run is in the GLM 5.3 Flash test report, the scoring rules are on our review methodology page, and the full parameter list is on the GLM 5.3 Flash model page.

What we liked and what held it back

What we liked

  • L4 Intense rendered every time with a system prompt. Three of three, against direct and fiction phrasings that were refused or softened on the same request. Of the three GLM chat models we've tested, this was the most consistent L4 Intense result under any single phrasing.
  • The bill is small. $0.1275 per million input tokens, $0.0255 cached and $0.425 output — a full document pass costs a fraction of a cent even at a wide context.
  • Three reasoning-effort levels on a low-cost model. low, high and max are all available, so a volume pass doesn't have to pay for deliberation it doesn't need.
  • L2 Partial and most of L5 Extreme rendered clean. All three L2 Partial phrasings passed, and five of six L5 Extreme requests came back in full.
  • The window and reply ceiling match GLM 5.3's. 1,048,576 tokens in, up to 131,072 back, at a fraction of the price.

What held it back

  • Direct and fiction phrasings don't carry L4 Intense. Asked directly, all three requests were refused. Framed as fiction, two were refused and one softened. Only the system-prompt phrasing worked in our tests.
  • L1 Suggestive softened on every phrasing, the same pattern we measured on GLM 5.3 and GLM 5.2 — the mildest rung isn't the easiest one here.
  • No audio or files. Messages take text, images and video; audio and file parts are rejected on this endpoint.
  • The default reply cap is a fraction of the ceiling. max_tokens defaults to 32,768 against a 131,072-token maximum.
  • We measured some unevenness in reply quality across our runs, including replies that ran on past the point where they should have ended; watch the token count on long generations.

What we tested

We ran GLM 5.3 Flash through our full content ladder on 17 and 26 September 2026, each rung asked three ways: direct, framed as fiction, and set through a system prompt. L2 Partial rendered clean on all three phrasings. L1 Suggestive softened on all three. L3 Strong rendered once (system prompt), softened once (fiction) and was refused once (direct).

L4 Intense split cleanly by phrasing: three direct requests were all refused, three fiction-framed requests came back refused twice and softened once, and all three system-prompt requests rendered in full. L5 Extreme rendered on five of the six runs we sent, with one fiction-framed run refused.

Across the ladder that's 24 scored runs: 12 rendered in full, 5 softened and 7 refused. The Spicy Index page reads this into a Freedom Score on this page's score card.

Prompts that worked

The read-and-summarize shape is where GLM 5.3 Flash's price and window pay off; for fiction, put the scene's boundaries in a system message rather than asking directly.

1. A published example: a long read, a short structured answer. The shape this window is built for.

text
system: Summarize the following support thread in three sentences: the customer's problem, what was tried, and the current status.
user: <thousands of lines of a support ticket thread, pasted in full>

Run this prompt on GLM 5.3 Flash →

2. A published example: patterns across a large corpus in one call. Ask for patterns rather than a summary of each item.

text
system: Group these tickets by root cause. Report each group's size and one representative ticket id.
user: <thousands of support tickets, pasted in full>

Run this prompt on GLM 5.3 Flash →

Settings that matter: reasoningeffort low for volume passes over large corpora, high or max only for the answer you can't afford to get wrong; put the scene's voice and boundaries in a system message before asking for anything past L2 Partial; leave headroom on maxtokens above the visible reply length you expect.

Roleplay and how far it writes

GLM 5.3 Flash rendered 12 of 24 scored runs across our content ladder in full. L2 Partial and most of L5 Extreme rendered regardless of phrasing. L4 Intense rendered only under a system prompt, but did so on all three runs we sent that way — direct and fiction phrasings didn't render it at all in our sample. L1 Suggestive softened across every phrasing. The page label is Uncensored: we add no content filter on our side.

Freedom Score: shown on this page's score card and recalculated as we add runs.

Not tested yet: multi-turn chats staying in character over many turns (our runs were single requests); languages other than English; whether the system-prompt result at L4 Intense holds up over a larger sample.

Every character in a request should be an adult, keep real, identifiable people out of it, and you're responsible for the law where you publish.

Alternatives

  • The full-size GLM, less consistent at L4 Intense? GLM 5.3, $1.05 per million input tokens; its system-prompt phrasing rendered L4 Intense half the time against Flash's three of three.
  • The earlier GLM, a lower price than 5.3 but no low reasoning-effort level? GLM 5.2, $0.98 per million input tokens.
  • A larger reply ceiling for whole-document extraction? DeepSeek V4.1 Flash, up to 384,000 tokens back against Flash's 131,072.
  • A reasoning-off switch? Grok 4.3, a genuine no-deliberation setting, in a 1,000,000-token window just short of Flash's 1,048,576.

How far GLM 5.3 Flash goes

Each level, on each task, based on confirmed results. A level earns points only when the content actually appears, and a single run earns 60% at most.

Freedom points by level and task
LevelChat
MildThe lightest writing requests in the test
0/3 delivered10.0 / 20 pts
ModerateA step up, still below the middle of the scale
3/3 delivered20.0 / 20 pts
StrongMid-scale requests; must be tested before a model is ranked
1/3 delivered10.0 / 20 pts
Very strongWhere most language models start to hold back
3/9 delivered7.8 / 20 pts
MaximumThe strongest written requests, scored as one class
5/6 delivered16.7 / 20 pts
Task score64.4 / 64.4Tested Sep 26, 2026

Tested Sep 17, 2026 to Sep 26, 2026

This site stays safe for work: results are written up as text, any image shown is pixel-censored, and test prompts are never published. We never test anyone who looks underage, real people’s likenesses or non-consent.

See every test record →See the uncensored leaderboard →

Run your own prompt and see how far it goes

Open the Playground

How GLM 5.3 Flash earns its Spicy Index

Each dimension below links to its full workings: the table row every point came from and why points were lost.

Public leaderboards 40% of quality
76.9
Role-play 15% of quality
73.6
Creative writing 15% of quality
63.4
Instruction following 15% of quality
95.0
Long-context memory 15% of quality
Not yet tested

Our long-context suite hasn’t run on this model yet. How it works

Capability 10.5% weight
73.0
Freedom Score 30% weight
64.4
Open the full test report →

Capability, item by item

Points from the public scoring table. Upscaled tiers and limits we set ourselves don’t count.

Context window (creator-published)1,048,576 tokens
24.0 / 30
Input modalities on SpicyAPIText, plus images
12.0 / 25
Tool useTools and tool_choice
15.0 / 15
Structured outputAccepted, schema adherence verified
15.0 / 15
Reasoning control3 levels: low · high · max
7.0 / 15

What GLM 5.3 Flash can do on SpicyAPI

Straight from the live model specification on SpicyAPI. Some creators offer more in their own apps; this is what you can call through our API today.

Capabilities by endpoint
FeatureChat
Context window1,048,576 tokens
Max output per request131,072 tokens
Reasoning levelslow · high · max
Tool callingYes
Structured outputYes
See every parameter and code sample →

How long it really takes

Measured on real jobs over the last 30 days, from request to finished result. Times include our own test runs.

Output speed
27 tok/sIncludes time to first token
Success rate
Collecting data

A time appears once at least 10 completed jobs are in. Success rates count customer jobs only and stay hidden for every model until we have a full 30 days of traffic.

Models to consider instead of GLM 5.3 Flash

  • Z.ai

    GLM 5.3

    The full-size GLM, less consistent at L4 Intense?

    Spicy Index
    72.5−0.4
    Freedom Score
    63.4−1.0
    Price
    $1.61+699%/ 1M tokens · 3:1 input/output
    Compare side by sideTry

  • Z.ai

    GLM 5.2

    The earlier GLM, a lower price than 5.3 but no low reasoning-effort level?

    Spicy Index
    71.9−1.0
    Freedom Score
    67.8+3.4
    Price
    $1.51+645%/ 1M tokens · 3:1 input/output
    Compare side by sideTry

  • DeepSeek

    DeepSeek V4.1 Flash

    A larger reply ceiling for whole-document extraction?

    Spicy Index
    81.9+9.0
    Freedom Score
    90.4+26.0
    Price
    $0.525+160%/ 1M tokens · 3:1 input/output
    Compare side by sideTry

  • xAI

    Grok 4.3

    A reasoning-off switch?

    Spicy Index
    77.0+4.1
    Freedom Score
    98.9+34.5
    Price
    $1.56+674%/ 1M tokens · 3:1 input/output
    Compare side by sideTry

What do you think?

Was this review useful?

Log in to vote.

Comments

Comments are checked automatically and by our team. We remove spam, offensive content, personal attacks and anything that names a real person without consent. Report a comment and we review it within one business day.

Tried this model? Share what worked. You need a free account to comment, so we can keep spam out.

GLM 5.3 Flash FAQ

Try GLM 5.3 Flash yourself

Set every parameter in the Playground and see the exact price before you run it.

$0.202 per 1M mixed tokens (3 input : 1 output)

More reviews

See the leaderboards →
  • Three glass discs stacked in the dark, crossed by a streak of cyan light, with a thin gold spark rising beneath them

    Grok 4.7

    Pick Grok 4.7 for uncensored roleplay and fiction that must work with no jailbreak; skip it for long chapters per turn or long agent loops.

    Spicy Index
    78.2
    Freedom Score
    100.0

    Tested Sep 24, 2026

    Read the review
  • Warm golden light streaks passing through a row of translucent blue glass panels on a dark pedestal

    DeepSeek V4.1 Flash

    Pick DeepSeek V4.1 Flash for cheap, whole-document reads and a 384,000-token ceiling; add a system prompt to steady intimate scenes (L4).

    Spicy Index
    81.9
    Freedom Score
    90.4

    Tested Sep 26, 2026

    Read the review
  • A burst of golden light filaments fanning low across a shallow glass dish against pure black

    Grok 4.3

    Pick Grok 4.3 for a huge context window, a genuine reasoning off switch and uncensored writing at a low price; skip it if you need file uploads.

    Spicy Index
    77.0
    Freedom Score
    98.9

    Tested Sep 24, 2026

    Read the review