BitdepthIndependent grades for inference endpoints
z-ai/glm-5.3-flash · 44 endpoints

GLM 5.3 Flash

First look at GLM 5.3 Flash: 44 endpoints on OpenRouter, 2 nights measured. Grades start at 7 nights.

Last run Nov 11, 2026
2 nights measured
0
match the reference
44
first look, not yet ranked
0
outside the reference
The board

Every endpoint, named.

#EndpointDeclaredString matchSelf-agreeNightsDrift$ in / out per MQuality-adjustedState
First look: fewer than 7 nights, not ranked
relace/fp4fp485.0%n/a2n/a$0.040 / $0.078$0.058First look, 2 nights
wafer/usnvfp479.2%n/a2n/a$0.042 / $0.18$0.098First look, 2 nights
morph/fp8fp895.7%n/a2n/a$0.093 / $0.12$0.10First look, 2 nights
darkbloom/fp4fp487.5%n/a2n/a$0.065 / $0.18$0.11First look, 2 nights
primeintellectunknown87.5%n/a2n/a$0.052 / $0.25$0.12First look, 2 nights
wafer/nvfp4nvfp475.0%n/a2n/a$0.065 / $0.24$0.14First look, 2 nights
gmicloud/fp8fp879.2%n/a2n/a$0.075 / $0.25$0.15First look, 2 nights
open-inference/fp4fp479.2%n/a2n/a$0.060 / $0.30$0.15First look, 2 nights
novita/fp8fp887.5%n/a2n/a$0.084 / $0.28$0.15First look, 2 nights
streamlake/fp8fp887.5%n/a2n/a$0.088 / $0.29$0.16First look, 2 nights
tenstorrent/fp8fp868.2%n/a2n/a$0.075 / $0.25$0.17First look, 2 nights
inference-net/usfp487.5%n/a2n/a$0.11 / $0.35$0.19First look, 2 nights
rekaunknown79.2%n/a2n/a$0.12 / $0.30$0.21First look, 2 nights
inference-net/fp4fp479.2%n/a2n/a$0.11 / $0.35$0.21First look, 2 nights
digitaloceanunknown83.3%n/a2n/a$0.11 / $0.38$0.21First look, 2 nights
makora/fp4fp475.0%n/a2n/a$0.14 / $0.24$0.22First look, 2 nights
near-ai/fp8fp883.3%n/a2n/a$0.080 / $0.50$0.22First look, 2 nights
coreweave/nvfp4nvfp483.3%n/a2n/a$0.12 / $0.38$0.22First look, 2 nights
nextbit/fp4fp483.3%n/a2n/a$0.12 / $0.45$0.24First look, 2 nights
particle-ai/nvfp4nvfp487.5%n/a2n/a$0.12 / $0.50$0.25First look, 2 nights
deepinfra/fp4fp491.7%n/a2n/a$0.15 / $0.50$0.26First look, 2 nights
veniceunknown91.7%n/a2n/a$0.15 / $0.50$0.26First look, 2 nights
z-ai/fp8fp891.7%n/a2n/a$0.15 / $0.50$0.26First look, 2 nights
cohere/nvfp4nvfp487.5%n/a2n/a$0.15 / $0.50$0.27First look, 2 nights
crusoe/fp4fp483.3%n/a2n/a$0.15 / $0.50$0.29First look, 2 nights
modal/fp4fp483.3%n/a2n/a$0.15 / $0.50$0.29First look, 2 nights
siliconflow/fp8fp883.3%n/a2n/a$0.15 / $0.50$0.29First look, 2 nights
nebius/fp4fp479.2%n/a2n/a$0.15 / $0.50$0.30First look, 2 nights
friendliunknown79.2%n/a2n/a$0.15 / $0.50$0.30First look, 2 nights
mistralunknown79.2%n/a2n/a$0.15 / $0.50$0.30First look, 2 nights
togetherunknown79.2%n/a2n/a$0.15 / $0.50$0.30First look, 2 nights
baseten/fp8fp879.2%n/a2n/a$0.15 / $0.50$0.30First look, 2 nights
inceptron/fp8fp879.2%n/a2n/a$0.17 / $0.50$0.32First look, 2 nights
parasail/fastfp490.9%n/a2n/a$0.19 / $0.62$0.33First look, 2 nights
atlas-cloud/fp8fp870.8%n/a2n/a$0.15 / $0.50$0.34First look, 2 nights
sail-research/fp4fp483.3%n/a2n/a$0.17 / $0.62$0.35First look, 2 nights
sail-research/usfp483.3%n/a2n/a$0.17 / $0.62$0.35First look, 2 nights
fireworksunknown66.7%n/a2n/a$0.15 / $0.50$0.36First look, 2 nights
riverunknown91.7%n/a1n/an/a / n/an/aFirst look, 1 night
phala/nvfp4nvfp487.5%n/a1n/an/a / n/an/aFirst look, 1 night
baseten/fp8/usfp8n/an/a2n/a$0.15 / $0.50n/aFirst look, 2 nights
dekallmunknownn/an/a1n/a$0.10 / $0.50n/aFirst look, 1 night
ktunknownn/an/a1n/a$0.15 / $0.50n/aFirst look, 1 night
parasail/fp4fp4n/an/a2n/a$0.15 / $0.50n/aFirst look, 2 nights

Cite: Bitdepth · bitdepth.dev/m/glm-5.3-flash · as of Nov 11, 2026

Reference: plurality across all endpoints (fewer than two declare bf16 or fp16). State rule: outside when self-agreement exceeds agreement with the reference by more than 15 points, or when the endpoint's mean on the model's first primary column is more than 8 points below the median of endpoints with 7+ nights; no_data under 7 nights; stale when the data is more than 3 days old.

Price against agreement

Cheap and wrong sits low on the left.

Each dot is one endpoint. Across: blended price per million tokens. Up: agreement with the reference.

Matches referenceFirst lookOutside reference
Blended price against agreement with the reference for 38 endpoints serving GLM 5.3 Flash.55%65%75%85%95%$0$0.05$0.1$0.15$0.2$0.25$0.3Blended price per million tokensdarkbloom/fp4: $0.094 blended, 87.5%deepinfra/fp4: $0.24 blended, 91.7%inceptron/fp8: $0.26 blended, 79.2%near-ai/fp8: $0.18 blended, 83.3%nextbit/fp4: $0.20 blended, 83.3%venice: $0.24 blended, 91.7%relace/fp4: $0.050 blended, 85.0%cohere/nvfp4: $0.24 blended, 87.5%coreweave/nvfp4: $0.18 blended, 83.3%crusoe/fp4: $0.24 blended, 83.3%gmicloud/fp8: $0.12 blended, 79.2%nebius/fp4: $0.24 blended, 79.2%streamlake/fp8: $0.14 blended, 87.5%parasail/fast: $0.30 blended, 90.9%friendli: $0.24 blended, 79.2%open-inference/fp4: $0.12 blended, 79.2%primeintellect: $0.10 blended, 87.5%mistral: $0.24 blended, 79.2%morph/fp8: $0.10 blended, 95.7%novita/fp8: $0.13 blended, 87.5%particle-ai/nvfp4: $0.21 blended, 87.5%sail-research/fp4: $0.29 blended, 83.3%inference-net/fp4: $0.17 blended, 79.2%sail-research/us: $0.29 blended, 83.3%together: $0.24 blended, 79.2%z-ai/fp8: $0.24 blended, 91.7%baseten/fp8: $0.24 blended, 79.2%fireworks: $0.24 blended, 66.7%makora/fp4: $0.17 blended, 75.0%wafer/us: $0.077 blended, 79.2%modal/fp4: $0.24 blended, 83.3%siliconflow/fp8: $0.24 blended, 83.3%tenstorrent/fp8: $0.12 blended, 68.2%atlas-cloud/fp8: $0.24 blended, 70.8%digitalocean: $0.18 blended, 83.3%inference-net/us: $0.17 blended, 87.5%wafer/nvfp4: $0.11 blended, 75.0%reka: $0.16 blended, 79.2%

Cite: Bitdepth · bitdepth.dev/m/glm-5.3-flash#price · as of Nov 11, 2026

Night by night

Does an endpoint change?

Exact match with the reference each night, one line per endpoint. × marks a night with no row.

Matches referenceFirst lookOutside reference
Exact match with the reference each night for 44 endpoints serving GLM 5.3 Flash.708090100Nov 10Nov 11darkbloom/fp4deepinfra/fp4inceptron/fp8near-ai/fp8nextbit/fp4venicerelace/fp4cohere/nvfp4coreweave/nvfp4crusoe/fp4gmicloud/fp8nebius/fp4riverstreamlake/fp8parasail/fastfriendliopen-inference/fp4primeintellectmistralmorph/fp8novita/fp8particle-ai/nvfp4sail-research/fp4inference-net/fp4sail-research/ustogetherz-ai/fp8baseten/fp8fireworksmakora/fp4wafer/usmodal/fp4phala/nvfp4siliconflow/fp8tenstorrent/fp8atlas-cloud/fp8digitaloceaninference-net/uswafer/nvfp4reka

Cite: Bitdepth · bitdepth.dev/m/glm-5.3-flash#nightly · as of Nov 11, 2026

For providers

Graded by Bitdepth.

An endpoint that matches the reference can show this badge on its docs or pricing page. It links back to its row here, and it updates every night.

Graded by Bitdepth badge sample

Badge code for a matching endpoint, after the first ranked read:

<a href="https://bitdepth.dev/m/glm-5.3-flash"><img src="https://bitdepth.dev/badge/glm-5.3-flash/PROVIDER.svg" alt="Graded by Bitdepth"></a>
Method

How this page is measured.

The reference

For each model, the answer most endpoints declaring full precision (bf16 or fp16) give at temperature zero. Where fewer than two declare it, the plurality answer across all endpoints. Method.

Agreement and self-agreement

Agreement is how often an endpoint gives the reference answer. Self-agreement is how often it gives its own answer twice. An endpoint that agrees with itself far more than with the reference is serving something different.

How rank works

Endpoints that match the reference are ranked by quality-adjusted price: the blended price, (3 × input + output) ÷ 4 per million tokens, divided by agreement. A cheap endpoint that misses the reference is never shown as the cheapest.

First look

An endpoint needs 7 nights before it is ranked. Until then it shows its night count and its numbers, with no rank.

Drift

The mean of the last three nights against the first three. A move of more than 10 points reads as improving or worsening; anything smaller is stable.

How to cite

Every chart and table carries a cite line with the date. Copy it with the figure. Wrong model name or missing endpoint? Tell us.

Numbers against a stated reference. Not a claim about intent.Listed prices from OpenRouter, read nightly; grades are Bitdepth's.