Dataset Viewer
Auto-converted to Parquet Duplicate
route_id
stringlengths
23
54
model_id
stringclasses
4 values
model
stringclasses
4 values
provider_id
stringlengths
4
20
provider
stringlengths
6
27
provider_model
stringlengths
12
45
upstream
stringlengths
5
40
⌀
modification
stringclasses
4 values
region
stringclasses
1 value
generated_at
stringclasses
1 value
methodology_version
stringclasses
1 value
window_hours
int64
72
72
demo
bool
1 class
status
stringclasses
3 values
requests
int64
72
72
successes
int64
0
72
success_rate
float64
0
1
ttft_ms_p50
float64
427
2.16k
⌀
ttft_ms_p90
float64
827
13.9k
⌀
output_tokens_per_second_p50
float64
43.7
248
⌀
output_tokens_per_second_p90
float64
73.8
543
⌀
end_to_end_500_ms_p50
float64
2.77k
13.7k
⌀
end_to_end_500_ms_p90
float64
4.66k
38.1k
⌀
blended_usd_per_mtok
float64
0.12
3.68
abliteration/deepseek-v4.1-flash/abliterated
deepseek-v4.1-flash
DeepSeek V4.1 Flash (undated derivative)
abliteration
abliteration.ai
abliterated-model-large-flash
Abliterated Large Flash
abliterated
all
2026-11-21T02:40:40.412105Z
1.0.0
72
false
ok
72
72
1
1,175.672286
1,703.228094
210.739307
542.818862
3,603.814987
6,085.804413
1.25
abliteration/glm-5.3/abliterated
glm-5.3
GLM 5.3
abliteration
abliteration.ai
abliterated-model-large-v2
Abliterated Large v2
abliterated
all
2026-11-21T02:40:40.412105Z
1.0.0
72
false
ok
72
72
1
1,251.814742
1,835.567767
183.969077
346.235561
3,903.21085
6,148.695301
3.5
explabs/deepseek-v4.1-flash-1017/uncensored
deepseek-v4.1-flash-1017
DeepSeek V4.1 Flash 1017
explabs
Experiential Labs
deepseek-v4.1-flash-uncensored
dealignai
uncensored
all
2026-11-21T02:40:40.412105Z
1.0.0
72
false
ok
72
72
1
787.271169
1,361.931928
247.928805
356.027318
2,766.414812
4,663.51172
0.39375
explabs/glm-5.3-flash/uncensored
glm-5.3-flash
GLM 5.3 Flash
explabs
Experiential Labs
glm-5.3-flash-uncensored
dealignai
uncensored
all
2026-11-21T02:40:40.412105Z
1.0.0
72
false
ok
72
72
1
1,151.589699
2,002.499046
161.376152
219.556969
4,296.99283
7,083.59743
0.2375
nanogpt-abliteration/deepseek-v4.1-flash/abliterated
deepseek-v4.1-flash
DeepSeek V4.1 Flash (undated derivative)
nanogpt-abliteration
abliteration.ai via NanoGPT
abliteration-ai/abliterated-model-large-flash
abliteration
abliterated
all
2026-11-21T02:40:40.412105Z
1.0.0
72
false
ok
72
72
1
1,917.328557
2,945.224739
221.78962
405.551688
4,441.956582
7,517.055869
1.3125
nanogpt-abliteration/glm-5.3/abliterated
glm-5.3
GLM 5.3
nanogpt-abliteration
abliteration.ai via NanoGPT
abliteration-ai/abliterated-model-large-v2
abliteration
abliterated
all
2026-11-21T02:40:40.412105Z
1.0.0
72
false
ok
72
72
1
1,997.958253
3,043.736558
185.919639
354.10535
4,728.055947
7,196.401955
3.675
nanogpt-arnict/glm-5.3-flash/uncensored
glm-5.3-flash
GLM 5.3 Flash
nanogpt-arnict
Arnict via NanoGPT
z-ai/glm-5.3-flash-uncensored
arnict
uncensored
all
2026-11-21T02:40:40.412105Z
1.0.0
72
false
degraded
72
62
0.861111
1,362.899672
2,091.512455
160.37705
235.520675
4,597.392361
8,966.95376
0.2295
nanogpt-arnict/glm-5.3/uncensored
glm-5.3
GLM 5.3
nanogpt-arnict
Arnict via NanoGPT
z-ai/glm-5.3-uncensored
arnict
uncensored
all
2026-11-21T02:40:40.412105Z
1.0.0
72
false
degraded
72
60
0.833333
2,164.844837
4,539.175086
64.285296
190.322219
9,916.302726
33,000.345753
1.3025
nanogpt-heabsy/deepseek-v4.1-flash-1017/uncensored
deepseek-v4.1-flash-1017
DeepSeek V4.1 Flash 1017
nanogpt-heabsy
Heabsy via NanoGPT
deepseek/deepseek-v4.1-flash-1017-uncensored
heabsy
uncensored
all
2026-11-21T02:40:40.412105Z
1.0.0
72
false
ok
72
72
1
1,234.907759
2,475.614953
241.377388
434.429257
3,382.910984
5,687.880421
0.61
nanogpt-heabsy/glm-5.3-flash/uncensored
glm-5.3-flash
GLM 5.3 Flash
nanogpt-heabsy
Heabsy via NanoGPT
z-ai/glm-5.3-flash-uncensored
heabsy
uncensored
all
2026-11-21T02:40:40.412105Z
1.0.0
72
false
ok
72
72
1
1,100.602427
1,885.614377
172.812899
264.768933
4,151.569141
7,090.791759
0.35
nanogpt-orcarouter/deepseek-v4.1-flash-1017/uncensored
deepseek-v4.1-flash-1017
DeepSeek V4.1 Flash 1017
nanogpt-orcarouter
OrcaRouter via NanoGPT
deepseek/deepseek-v4.1-flash-1017-uncensored
orcarouter
uncensored
all
2026-11-21T02:40:40.412105Z
1.0.0
72
false
failing
72
0
0
null
null
null
null
null
null
1.05
nanogpt-orcarouter/glm-5.3-flash/uncensored
glm-5.3-flash
GLM 5.3 Flash
nanogpt-orcarouter
OrcaRouter via NanoGPT
z-ai/glm-5.3-flash-uncensored
orcarouter
uncensored
all
2026-11-21T02:40:40.412105Z
1.0.0
72
false
failing
72
0
0
null
null
null
null
null
null
1.05
nanogpt-sable/deepseek-v4.1-flash-1017/heretic
deepseek-v4.1-flash-1017
DeepSeek V4.1 Flash 1017
nanogpt-sable
Sable via NanoGPT
pqhaz/DeepSeek-V4.1-Flash-1017-heretic
sable
heretic
all
2026-11-21T02:40:40.412105Z
1.0.0
72
false
degraded
72
66
0.916667
1,936.410061
5,874.467732
113.667736
195.664177
6,585.804791
14,893.163529
0.303
nanogpt-venice/deepseek-v4.1-flash/uncensored
deepseek-v4.1-flash
DeepSeek V4.1 Flash (undated derivative)
nanogpt-venice
Venice via NanoGPT
deepseek/deepseek-v4.1-flash-uncensored
venice
uncensored
all
2026-11-21T02:40:40.412105Z
1.0.0
72
false
ok
72
72
1
1,834.541527
2,825.693522
191.879077
357.386169
4,615.573459
6,917.675577
0.60775
openrouter/deepseek-v4.1-flash-1017/none
deepseek-v4.1-flash-1017
DeepSeek V4.1 Flash 1017
openrouter
OpenRouter
deepseek/deepseek-v4.1-flash-1017
morph/fp8
none
all
2026-11-21T02:40:40.412105Z
1.0.0
72
false
ok
72
72
1
427.314936
827.1545
104.831236
173.281319
5,578.891394
7,475.299466
0.1165
openrouter/glm-5.3-flash/none
glm-5.3-flash
GLM 5.3 Flash
openrouter
OpenRouter
z-ai/glm-5.3-flash
z-ai/fp8
none
all
2026-11-21T02:40:40.412105Z
1.0.0
72
false
ok
72
72
1
1,374.459516
3,050.650703
50.668957
73.787381
11,913.503832
16,167.275915
0.2375
openrouter/glm-5.3/none
glm-5.3
GLM 5.3
openrouter
OpenRouter
z-ai/glm-5.3
z-ai/fp8
none
all
2026-11-21T02:40:40.412105Z
1.0.0
72
false
ok
72
72
1
1,479.846628
3,465.942263
65.708861
110.561765
9,498.470998
14,296.644747
2.15
refuseless/deepseek-v4.1-flash/abliterated
deepseek-v4.1-flash
DeepSeek V4.1 Flash (undated derivative)
refuseless
Refuseless
deepseek-v4.1-flash-abliterated
null
abliterated
all
2026-11-21T02:40:40.412105Z
1.0.0
72
false
degraded
72
68
0.944444
1,067.357509
2,964.81485
52.138515
139.876648
11,275.197984
19,151.452402
0.2025
refuseless/glm-5.3-flash/abliterated
glm-5.3-flash
GLM 5.3 Flash
refuseless
Refuseless
glm-5.3-flash-abliterated
null
abliterated
all
2026-11-21T02:40:40.412105Z
1.0.0
72
false
ok
72
72
1
1,686.366029
11,222.043212
43.710068
84.790717
13,689.10273
38,142.398121
0.215
refuseless/glm-5.3/abliterated
glm-5.3
GLM 5.3
refuseless
Refuseless
glm-5.3-abliterated
null
abliterated
all
2026-11-21T02:40:40.412105Z
1.0.0
72
false
degraded
72
68
0.944444
1,587.138162
13,898.826522
44.657711
90.044289
12,808.200582
37,499.556231
1.25
venice/deepseek-v4.1-flash-1017/uncensored
deepseek-v4.1-flash-1017
DeepSeek V4.1 Flash 1017
venice
Venice
deepseek-v4-1-flash-1017-uncensored
null
uncensored
all
2026-11-21T02:40:40.412105Z
1.0.0
72
false
ok
72
72
1
547.663174
1,037.262448
171.6941
355.755885
3,422.58934
8,627.815707
0.5775
venice/glm-5.3/abliterated
glm-5.3
GLM 5.3
venice
Venice
abliteration-abliterated-model-large-v2
Abliterated Large v2 via abliteration.ai
abliterated
all
2026-11-21T02:40:40.412105Z
1.0.0
72
false
ok
72
72
1
1,420.265638
2,267.136046
213.480171
293.585414
3,748.289497
5,079.044235
3.5
yeta/deepseek-v4.1-flash/uncensored
deepseek-v4.1-flash
DeepSeek V4.1 Flash (undated derivative)
yeta
Yeta Labs
deepseek-v4.1-flash-uncensored
null
uncensored
all
2026-11-21T02:40:40.412105Z
1.0.0
72
false
ok
72
70
0.972222
1,998.220453
3,826.839189
134.093939
191.86632
6,055.174143
8,720.708481
0.1475

Abliterated Model API Benchmarks

Customer-observed API latency, output speed, reliability and catalogue prices for abliterated and uncensored open-weight models, with standard-model controls. Data and charts: benchmarks.redrouter.org. These measurements do not assess model quality, refusal rates or worldwide uptime.

Generated: 2026-11-21T02:40:40.412105Z. Methodology: 1.0.0.

Current routes by model

Fastest means highest pooled 72-hour p50 output speed; cheapest means lowest catalogue blended price among routes with at least eight attempted requests in that window. Missing speed is excluded; unavailable results are marked Unavailable. Ties use route ID order. Modifications are distinct products.

Model Fastest route p50 tok/s Cheapest route USD / million tokens
DeepSeek V4.1 Flash (undated derivative) abliteration.ai via NanoGPT 222 Yeta Labs 0.147
DeepSeek V4.1 Flash 1017 dealignai via Experiential Labs 248 morph/fp8 via OpenRouter 0.117
GLM 5.3 Abliterated Large v2 via abliteration.ai via Venice 213 Refuseless 1.25
GLM 5.3 Flash Heabsy via NanoGPT 173 Refuseless 0.215

Columns

summary.csv contains one row per route. history.csv contains one row per route, region and time bucket. Empty CSV fields are missing values, not zeroes.

Columns Definition
route_id, model_id, provider_id Stable route, model and provider identifiers
model, provider Display names
provider_model, upstream, modification Requested model identifier, declared supplier if known, and modification label; none denotes a standard-model control
region Client measurement region; all in summary pools observed requests from both regions
generated_at, methodology_version, window_hours, demo Source generation time in UTC, protocol version, rolling window length, and synthetic-data flag
status Summary availability: ok, degraded, failing, stale or insufficient_data
requests, successes, success_rate Attempted and successful request counts and their ratio (0–1); summary counts cover the rolling window, history counts cover the bucket
ttft_ms_p50, ttft_ms_p90 Time from sending to the first content delta, milliseconds
output_tokens_per_second_p50, output_tokens_per_second_p90 Decode speed after the first chunk, tokens per second
end_to_end_500_ms_p50, end_to_end_500_ms_p90 Per-request TTFT plus extrapolated time for 500 output tokens, milliseconds
blended_usd_per_mtok Summary catalogue price: (3 × input + output) / 4, USD per million tokens, without cache discounts
t, bucket_hours History bucket start in UTC and bucket duration in hours

History contains the chart's rolling-window p50 values only, ending at the earlier of bucket end and generation time; p90 and price are summary-only. History success rate describes the bucket, not the rolling window.

Methodology and updates

Measurements run every 8 hours from two US client regions: us-east-1 (N. Virginia) and us-west-2 (Oregon). Data is refreshed after each site update. Each run attempts four streaming requests per route across prose/code and approximately 1,000/8,000 input tokens, with a 512-token output cap, no retries and no warm-up calls. Percentiles pool successful requests over the trailing 72 hours and require at least eight successful and eight usable samples per metric. Failures remain in the success-rate denominator. Prices are dated catalogue declarations rather than billing receipts. See the methodology for formulas, limitations and reproduction details.

Licence and attribution

Results are licensed under CC BY 4.0. Attribute as: RedRouter Abliterated Benchmarks, CC BY 4.0. Include the methodology version and observation period when reusing results.

@misc{redrouter_abliterated_benchmarks,
  author = {RedRouter},
  title = {Abliterated Model API Benchmarks},
  year = {2026},
  url = {https://benchmarks.redrouter.org},
  note = {Methodology 1.0.0; generated 2026-11-21T02:40:40.412105Z}
}
Downloads last month
446