Stage 2 / independent measurements

Inference weather, not a universal leaderboard.

Compare exact service routes only inside the same scenario, region, concurrency, methodology, and time window. Small samples stay visibly insufficient; percentiles are never extrapolated.

Matching aggregates44

Fresh24

Complete windows2

Public incidents / alerts0 / 0

Latest window per exact slice

Observed service behavior

Ranking: none

Ling 2 6 FlashOpenRouter

OpenAI compatibility 1K / 250singapore · c1

3/33 completed

100.0%eligible runs

1,009.2ms / Withheldthreshold-gated

123.2 tok/sgeneration

collectingFresh · Aug 03, 2026, 10:51 PM UTC

Qwen3 5 9bVenice

Interactive 1K / 250singapore · c1

3/33 completed

100.0%eligible runs

820.2ms / Withheldthreshold-gated

0 tok/sgeneration

collectingFresh · Aug 03, 2026, 10:42 PM UTC

Ling 2 6 FlashOpenRouter

Interactive 1K / 250western-europe · c1

2/32 completed

100.0%eligible runs

Withheld / Withheldthreshold-gated

Withheldgeneration

insufficient dataFresh · Aug 03, 2026, 11:46 PM UTC

Mistral Nemo Instruct 2407DeepInfra

Interactive 1K / 250western-europe · c1

3/33 completed

100.0%eligible runs

14,262.7ms / Withheldthreshold-gated

35.6 tok/sgeneration

collectingFresh · Aug 03, 2026, 11:06 PM UTC

Ling 2 6 FlashOpenRouter

OpenAI compatibility 1K / 250western-europe · c1

2/33 completed

66.7%eligible runs

Withheld / Withheldthreshold-gated

Withheldgeneration

insufficient dataFresh · Aug 03, 2026, 11:31 PM UTC

GPT OSS 120bCerebras

OpenAI compatibility 1K / 250singapore · c1

3/33 completed

100.0%eligible runs

452.8ms / Withheldthreshold-gated

87.4 tok/sgeneration

collectingFresh · Aug 03, 2026, 10:37 PM UTC

Ling 2 6 FlashOpenRouter

OpenAI compatibility 1K / 250us-east · c1

1/33 completed

33.3%eligible runs

Withheld / Withheldthreshold-gated

Withheldgeneration

insufficient dataFresh · Aug 03, 2026, 11:36 PM UTC

Qwen3 5 9bVenice

Interactive 1K / 250western-europe · c1

3/33 completed

100.0%eligible runs

673.2ms / Withheldthreshold-gated

0 tok/sgeneration

collectingFresh · Aug 03, 2026, 07:19 PM UTC

GPT OSS 120bCerebras

Interactive 1K / 250western-europe · c1

3/33 completed

100.0%eligible runs

212.8ms / Withheldthreshold-gated

42.3 tok/sgeneration

collectingFresh · Aug 03, 2026, 11:11 PM UTC

Qwen3 5 9bVenice

OpenAI compatibility 1K / 250singapore · c1

3/33 completed

100.0%eligible runs

793.9ms / Withheldthreshold-gated

0 tok/sgeneration

collectingFresh · Aug 03, 2026, 11:26 PM UTC

Qwen3 5 9bVenice

Interactive 1K / 250us-east · c1

3/33 completed

100.0%eligible runs

553.9ms / Withheldthreshold-gated

0 tok/sgeneration

collectingFresh · Aug 03, 2026, 11:16 PM UTC

Mistral Nemo Instruct 2407DeepInfra

OpenAI compatibility 1K / 250western-europe · c1

3/33 completed

100.0%eligible runs

2,480ms / Withheldthreshold-gated

94.3 tok/sgeneration

collectingFresh · Aug 03, 2026, 09:13 PM UTC

GPT OSS 120bCerebras

Interactive 1K / 250us-east · c1

3/33 completed

100.0%eligible runs

188ms / Withheldthreshold-gated

131.7 tok/sgeneration

collectingFresh · Aug 03, 2026, 09:28 PM UTC

Qwen3 5 9bVenice

OpenAI compatibility 1K / 250western-europe · c1

3/33 completed

100.0%eligible runs

678.5ms / Withheldthreshold-gated

0 tok/sgeneration

collectingFresh · Aug 03, 2026, 09:03 PM UTC

Mistral Nemo Instruct 2407DeepInfra

OpenAI compatibility 1K / 250us-east · c1

2/32 completed

100.0%eligible runs

Withheld / Withheldthreshold-gated

Withheldgeneration

insufficient dataFresh · Aug 03, 2026, 08:24 PM UTC

GPT OSS 120bCerebras

OpenAI compatibility 1K / 250us-east · c1

3/33 completed

100.0%eligible runs

103.8ms / Withheldthreshold-gated

85.5 tok/sgeneration

collectingFresh · Aug 03, 2026, 10:56 PM UTC

Qwen3 5 9bVenice

OpenAI compatibility 1K / 250us-east · c1

3/33 completed

100.0%eligible runs

589.5ms / Withheldthreshold-gated

0 tok/sgeneration

collectingFresh · Aug 03, 2026, 11:41 PM UTC

Mistral Nemo Instruct 2407DeepInfra

OpenAI compatibility 1K / 250singapore · c1

3/33 completed

100.0%eligible runs

5,124.6ms / Withheldthreshold-gated

71.2 tok/sgeneration

collectingFresh · Aug 03, 2026, 10:47 PM UTC

Mistral Nemo Instruct 2407DeepInfra

Interactive 1K / 250us-east · c1

3/33 completed

100.0%eligible runs

419ms / Withheldthreshold-gated

87.7 tok/sgeneration

collectingFresh · Aug 03, 2026, 08:38 PM UTC

GPT OSS 120bCerebras

Interactive 1K / 250singapore · c1

3/33 completed

100.0%eligible runs

335.3ms / Withheldthreshold-gated

113.8 tok/sgeneration

collectingFresh · Aug 03, 2026, 10:12 PM UTC

Mistral Nemo Instruct 2407DeepInfra

Interactive 1K / 250singapore · c1

3/33 completed

100.0%eligible runs

3,802.8ms / Withheldthreshold-gated

82.9 tok/sgeneration

collectingFresh · Aug 03, 2026, 09:33 PM UTC

GPT OSS 120bCerebras

OpenAI compatibility 1K / 250western-europe · c1

3/33 completed

100.0%eligible runs

197.3ms / Withheldthreshold-gated

5 tok/sgeneration

collectingFresh · Aug 03, 2026, 11:01 PM UTC

Ling 2 6 FlashOpenRouter

Interactive 1K / 250singapore · c1

3/33 completed

100.0%eligible runs

1,163.6ms / Withheldthreshold-gated

149.8 tok/sgeneration

collectingFresh · Aug 03, 2026, 08:58 PM UTC

Ling 2 6 FlashOpenRouter

Interactive 1K / 250us-east · c1

1/33 completed

33.3%eligible runs

Withheld / Withheldthreshold-gated

Withheldgeneration

insufficient dataFresh · Aug 03, 2026, 11:21 PM UTC

GPT OSS 120bCerebras

Interactive 1K / 250singapore · c1

2/32 completed

100.0%eligible runs

Withheld / Withheldthreshold-gated

Withheldgeneration

insufficient dataStale · Aug 03, 2026, 01:49 AM UTC

Qwen3 5 9bVenice

OpenAI compatibility 1K / 250western-europe · c1

1/31 completed

100.0%eligible runs

Withheld / Withheldthreshold-gated

Withheldgeneration

insufficient dataStale · Aug 03, 2026, 01:00 AM UTC

Mistral Nemo Instruct 2407DeepInfra

Interactive 1K / 250singapore · c1

2/32 completed

100.0%eligible runs

Withheld / Withheldthreshold-gated

Withheldgeneration

insufficient dataStale · Aug 03, 2026, 02:04 AM UTC

GPT OSS 120bCerebras

OpenAI compatibility 1K / 250western-europe · c1

2/32 completed

100.0%eligible runs

Withheld / Withheldthreshold-gated

Withheldgeneration

insufficient dataStale · Aug 03, 2026, 01:44 AM UTC

GPT OSS 120bCerebras

Interactive 1K / 250us-east · c1

1/31 completed

100.0%eligible runs

Withheld / Withheldthreshold-gated

Withheldgeneration

insufficient dataStale · Aug 03, 2026, 12:55 AM UTC

Ling 2 6 FlashOpenRouter

Interactive 1K / 250us-east · c1

2/32 completed

100.0%eligible runs

Withheld / Withheldthreshold-gated

Withheldgeneration

insufficient dataStale · Aug 03, 2026, 01:39 AM UTC

Qwen3 5 9bVenice

Interactive 1K / 250singapore · c1

1/31 completed

100.0%eligible runs

Withheld / Withheldthreshold-gated

Withheldgeneration

insufficient dataStale · Aug 03, 2026, 12:59 AM UTC

GPT OSS 120bCerebras

OpenAI compatibility 1K / 250singapore · c1

3/33 completed

100.0%eligible runs

330.2ms / Withheldthreshold-gated

114.2 tok/sgeneration

completeStale · Aug 03, 2026, 02:24 AM UTC

Mistral Nemo Instruct 2407DeepInfra

OpenAI compatibility 1K / 250western-europe · c1

1/31 completed

100.0%eligible runs

Withheld / Withheldthreshold-gated

Withheldgeneration

insufficient dataStale · Aug 03, 2026, 12:53 AM UTC

Qwen3 5 9bVenice

OpenAI compatibility 1K / 250us-east · c1

3/33 completed

100.0%eligible runs

699.2ms / Withheldthreshold-gated

0 tok/sgeneration

completeStale · Aug 03, 2026, 02:04 AM UTC

Qwen3 5 9bVenice

OpenAI compatibility 1K / 250singapore · c1

1/31 completed

100.0%eligible runs

Withheld / Withheldthreshold-gated

Withheldgeneration

insufficient dataStale · Aug 03, 2026, 01:15 AM UTC

Mistral Nemo Instruct 2407DeepInfra

OpenAI compatibility 1K / 250us-east · c1

2/32 completed

100.0%eligible runs

Withheld / Withheldthreshold-gated

Withheldgeneration

insufficient dataStale · Aug 03, 2026, 01:05 AM UTC

Qwen3 5 9bVenice

Interactive 1K / 250western-europe · c1

1/31 completed

100.0%eligible runs

Withheld / Withheldthreshold-gated

Withheldgeneration

insufficient dataStale · Aug 03, 2026, 12:53 AM UTC

Ling 2 6 FlashOpenRouter

Interactive 1K / 250singapore · c1

1/31 completed

100.0%eligible runs

Withheld / Withheldthreshold-gated

Withheldgeneration

insufficient dataStale · Aug 03, 2026, 01:29 AM UTC

Ling 2 6 FlashOpenRouter

OpenAI compatibility 1K / 250us-east · c1

2/32 completed

100.0%eligible runs

Withheld / Withheldthreshold-gated

Withheldgeneration

insufficient dataStale · Aug 03, 2026, 02:33 AM UTC

Ling 2 6 FlashOpenRouter

OpenAI compatibility 1K / 250singapore · c1

1/32 completed

50.0%eligible runs

Withheld / Withheldthreshold-gated

Withheldgeneration

insufficient dataStale · Aug 03, 2026, 02:19 AM UTC

GPT OSS 120bCerebras

Interactive 1K / 250western-europe · c1

1/31 completed

100.0%eligible runs

Withheld / Withheldthreshold-gated

Withheldgeneration

insufficient dataStale · Aug 03, 2026, 01:34 AM UTC

Ling 2 6 FlashOpenRouter

Interactive 1K / 250western-europe · c1

1/31 completed

100.0%eligible runs

Withheld / Withheldthreshold-gated

Withheldgeneration

insufficient dataStale · Aug 03, 2026, 02:28 AM UTC

Mistral Nemo Instruct 2407DeepInfra

Interactive 1K / 250us-east · c1

2/32 completed

100.0%eligible runs

Withheld / Withheldthreshold-gated

Withheldgeneration

insufficient dataStale · Aug 03, 2026, 02:14 AM UTC

Ling 2 6 FlashOpenRouter

OpenAI compatibility 1K / 250western-europe · c1

1/31 completed

100.0%eligible runs

Withheld / Withheldthreshold-gated

Withheldgeneration

insufficient dataStale · Aug 02, 2026, 11:20 PM UTC