# modelstatus.dev — current status Snapshot: 2026-08-21 15:25 UTC (2m ago) Endpoints: 1200 across 73 providers State: 720 healthy, 59 degraded, 54 down, 367 no data Notes on interpretation: - "no data" means the endpoint carries no recent traffic, so OpenRouter reports null. It is not an outage. - There is no site-wide uptime figure. Averaging every endpoint hides both long-dead endpoints nobody uses and total outages at a single large provider. Use the counts above, or the per-endpoint numbers below. - Latency figures are OpenRouter 30-minute rolling windows. ## Incidents in progress (56) - [major] Kimi K3 @ DeepInfra [deepinfra/bf16]: deranked by OpenRouter; 4h 57m so far, since 2026-08-21 10:30 UTC - [major] Gemini 2.5 Flash @ Google [google-vertex]: deranked by OpenRouter; 4h 32m so far, since 2026-08-21 10:55 UTC - [major] DeepSeek V3.2 Exp @ SiliconFlow [siliconflow/fp8]: uptime fell to 53.18%; 4h 22m so far, since 2026-08-21 11:05 UTC - [major] Gemma 4 31B @ Phala [phala]: uptime fell to 0.57%; 3h 32m so far, since 2026-08-21 11:55 UTC - [major] GLM 4.6V @ Z.AI [z-ai/fp8]: deranked by OpenRouter; 3h 27m so far, since 2026-08-21 12:00 UTC - [major] Kimi K3 @ Wafer [wafer]: deranked by OpenRouter; 3h 22m so far, since 2026-08-21 12:05 UTC - [major] MiMo-V2.5-Pro @ DeepInfra [deepinfra/fp8]: deranked by OpenRouter; 3h 17m so far, since 2026-08-21 12:10 UTC - [major] MiniMax M2.7 @ DeepInfra [deepinfra/turbo]: deranked by OpenRouter; 3h 17m so far, since 2026-08-21 12:10 UTC - [major] Qwen3.5-27B @ Phala [phala]: deranked by OpenRouter; 2h 32m so far, since 2026-08-21 12:55 UTC - [major] gpt-oss-120b @ SambaNova [sambanova]: deranked by OpenRouter; 2h 32m so far, since 2026-08-21 12:55 UTC - [major] Nemotron 3.5 Lightning @ CoreWeave [coreweave/bf16]: uptime fell to 12.50%; 2h 32m so far, since 2026-08-21 12:55 UTC - [major] Kimi K2.6 @ Crusoe [crusoe/bf16]: deranked by OpenRouter; 2h 32m so far, since 2026-08-21 12:55 UTC - [major] Kimi K2 Thinking @ Google [google-vertex]: deranked by OpenRouter; 2h 27m so far, since 2026-08-21 13:00 UTC - [major] gpt-oss-120b @ DeepInfra [deepinfra/bf16]: uptime fell to 66.36%; 2h 17m so far, since 2026-08-21 13:10 UTC - [major] Qwen3.6 27B @ Chutes [chutes/fp8]: deranked by OpenRouter; 1h 32m so far, since 2026-08-21 13:55 UTC - [major] Grok 4.6 @ xAI [xai/priority]: uptime fell to 65.94%; 1h 27m so far, since 2026-08-21 14:00 UTC - [major] Grok 4.6 @ xAI [xai]: uptime fell to 65.94%; 1h 27m so far, since 2026-08-21 14:00 UTC - [major] Kimi K3 @ Fireworks [fireworks/us]: uptime fell to 0.00%; 1h 22m so far, since 2026-08-21 14:05 UTC - [major] GLM 4.6 @ AtlasCloud [atlas-cloud/fp8]: deranked by OpenRouter; 1h 17m so far, since 2026-08-21 14:10 UTC - [major] MiMo-V2.5 @ GMICloud [gmicloud/fp8]: deranked by OpenRouter; 1h 7m so far, since 2026-08-21 14:20 UTC - [major] Qwen3 235B A22B Instruct 2507 @ Venice [venice/fp8]: deranked by OpenRouter; 1h 7m so far, since 2026-08-21 14:20 UTC - [major] GLM 5 @ Amazon Bedrock [amazon-bedrock]: deranked by OpenRouter; 57m so far, since 2026-08-21 14:30 UTC - [major] Qwen3 235B A22B Instruct 2507 @ DeepInfra [deepinfra/fp8]: deranked by OpenRouter; 52m so far, since 2026-08-21 14:35 UTC - [major] GLM 5.1 @ Venice [venice/fp8]: deranked by OpenRouter; 42m so far, since 2026-08-21 14:45 UTC - [major] Kimi K2.6 @ DigitalOcean [digitalocean]: uptime fell to 25.00%; 32m so far, since 2026-08-21 14:55 UTC - [major] Inkling @ DeepInfra [deepinfra/fp8]: deranked by OpenRouter; 27m so far, since 2026-08-21 15:00 UTC - [major] Qwen3.8 2.4T A95B @ DeepInfra [deepinfra/fp4]: uptime fell to 17.65%; 27m so far, since 2026-08-21 15:00 UTC - [major] Qwen3.6 27B @ Venice [venice/fp8]: uptime fell to 78.18%; 27m so far, since 2026-08-21 15:00 UTC - [major] GLM 4.7 @ Mancer 2 [mancer/fp4]: uptime fell to 47.44%; 27m so far, since 2026-08-21 15:00 UTC - [major] Nemotron 3 Ultra @ Venice [venice/fp8]: deranked by OpenRouter; 27m so far, since 2026-08-21 15:00 UTC - [major] Llama 4 Scout @ Novita [novita/bf16]: deranked by OpenRouter; 17m so far, since 2026-08-21 15:10 UTC - [major] Kimi K3 @ Phala [phala]: deranked by OpenRouter; 12m so far, since 2026-08-21 15:15 UTC - [major] DeepSeek V4 Pro 0813 @ Cloudflare [cloudflare]: deranked by OpenRouter; 12m so far, since 2026-08-21 15:15 UTC - [major] GLM 5.2 @ Ambient [ambient/fp8]: uptime fell to 66.67%; 7m so far, since 2026-08-21 15:20 UTC - [major] Qwen2.5 VL 72B Instruct @ Nebius [nebius/fp8]: uptime fell to 0.00%; 7m so far, since 2026-08-21 15:20 UTC - [degraded] Kimi K2.6 @ AtlasCloud [atlas-cloud/int4]: uptime fell to 83.33%; 5h 17m so far, since 2026-08-21 10:10 UTC - [degraded] DeepSeek V3.2 @ SiliconFlow [siliconflow/fp8]: uptime fell to 60.16%; 3h 57m so far, since 2026-08-21 11:30 UTC - [degraded] Mistral Nemo @ Io Net [io-net/fp16]: uptime fell to 68.09%; 1h 52m so far, since 2026-08-21 13:35 UTC - [degraded] DeepSeek V4 Flash 0423 @ Mancer 2 [mancer/fp8]: uptime fell to 94.63%; 1h 17m so far, since 2026-08-21 14:10 UTC - [degraded] Qwen3.5-35B-A3B @ Alibaba [alibaba]: uptime fell to 94.59%; 47m so far, since 2026-08-21 14:40 UTC - [degraded] Claude Sonnet 4.5 @ Anthropic [anthropic]: uptime fell to 90.54%; 47m so far, since 2026-08-21 14:40 UTC - [degraded] DeepSeek V4 Pro 0813 @ StreamLake [streamlake]: uptime fell to 90.63%; 32m so far, since 2026-08-21 14:55 UTC - [degraded] Inkling @ Together [together]: uptime fell to 80.67%; 27m so far, since 2026-08-21 15:00 UTC - [degraded] Qwen3.6 35B A3B @ Phala [phala]: uptime fell to 84.29%; 27m so far, since 2026-08-21 15:00 UTC - [degraded] DeepSeek V3.2 @ StreamLake [streamlake/fp8]: uptime fell to 64.50%; 27m so far, since 2026-08-21 15:00 UTC - [degraded] DeepSeek V4 Flash 0423 @ SiliconFlow [siliconflow/fp8]: uptime fell to 89.29%; 22m so far, since 2026-08-21 15:05 UTC - [degraded] Qwen3 235B A22B Instruct 2507 @ Novita [novita/fp8]: uptime fell to 84.17%; 17m so far, since 2026-08-21 15:10 UTC - [degraded] GLM 5.1 @ Parasail [parasail/fp8]: uptime fell to 97.44%; 12m so far, since 2026-08-21 15:15 UTC - [degraded] GLM 5.1 @ GMICloud [gmicloud/fp8]: uptime fell to 97.14%; 12m so far, since 2026-08-21 15:15 UTC - [degraded] MiniMax M3 @ Morph [morph/fp4]: uptime fell to 91.27%; 12m so far, since 2026-08-21 15:15 UTC - [degraded] MiniMax M2.5 @ Venice [venice]: uptime fell to 97.70%; 12m so far, since 2026-08-21 15:15 UTC - [degraded] Gemma 4 31B @ CoreWeave [coreweave/bf16]: uptime fell to 95.03%; 12m so far, since 2026-08-21 15:15 UTC - [degraded] GLM 5.2 @ Together [together]: uptime fell to 85.97%; 7m so far, since 2026-08-21 15:20 UTC - [degraded] GLM 5.2 @ Alibaba [alibaba/fp8]: uptime fell to 97.92%; 7m so far, since 2026-08-21 15:20 UTC - [degraded] GLM 5.2 @ Decart [decart/fp4]: uptime fell to 91.76%; 7m so far, since 2026-08-21 15:20 UTC - [degraded] gpt-oss-20b @ SiliconFlow [siliconflow/fp8]: uptime fell to 88.28%; 7m so far, since 2026-08-21 15:20 UTC ## Down for more than 6 hours (28) These are standing states rather than news. OpenRouter still lists them, and its own 24-hour uptime figure will not make clear that a provider is simply dead. - [degraded] GLM 4.7 Flash @ Novita [novita/bf16]: uptime fell to 40.63%; 2d 12h so far, since 2026-08-19 03:10 UTC - [major] Nemotron Nano 12B 2 VL (free) @ Nvidia [nvidia]: deranked by OpenRouter; 2d 12h so far, since 2026-08-19 03:10 UTC - [major] Llama 3.3 70B Instruct @ Nebius [nebius/fp8]: deranked by OpenRouter; 2d 12h so far, since 2026-08-19 03:10 UTC - [major] Qwen3.5 397B A17B @ GMICloud [gmicloud/fp8]: deranked by OpenRouter; 2d 12h so far, since 2026-08-19 03:25 UTC - [major] Kimi K2.7 Code @ Together [together]: deranked by OpenRouter; 2d 5h so far, since 2026-08-19 09:35 UTC - [degraded] Kimi K2.7 Code @ ModelRun [modelrun/fp4]: uptime fell to 56.76%; 2d 5h so far, since 2026-08-19 10:25 UTC - [degraded] Nemotron 3 Super @ Nebius [nebius/fp4]: uptime fell to 0.00%; 2d 4h so far, since 2026-08-19 10:45 UTC - [major] Qwen2.5 7B Instruct @ Together [together]: uptime fell to 0.00%; 1d 19h so far, since 2026-08-19 20:00 UTC - [major] Mistral Small 4 @ Mistral [mistral/zdr]: deranked by OpenRouter; 1d 14h so far, since 2026-08-20 01:20 UTC - [degraded] GLM 5.1 @ Phala [phala]: uptime fell to 17.50%; 1d 1h so far, since 2026-08-20 13:30 UTC - [major] Grok 4.20 Multi-Agent @ xAI [xai/priority]: deranked by OpenRouter; 23h 12m so far, since 2026-08-20 16:15 UTC - [major] Grok 4.20 Multi-Agent @ xAI [xai]: deranked by OpenRouter; 23h 12m so far, since 2026-08-20 16:15 UTC - [major] GLM 5.1 @ Chutes [chutes/fp8]: deranked by OpenRouter; 22h 42m so far, since 2026-08-20 16:45 UTC - [degraded] Qwen3.6 35B A3B @ Darkbloom [darkbloom]: uptime fell to 88.71%; 21h 32m so far, since 2026-08-20 17:55 UTC - [degraded] Qwen3.6 35B A3B @ Darkbloom [darkbloom/fp4]: uptime fell to 45.28%; 18h 42m so far, since 2026-08-20 20:45 UTC - [major] gpt-oss-20b @ Google [google-vertex/us-central1]: deranked by OpenRouter; 17h 52m so far, since 2026-08-20 21:35 UTC - [major] Qwen2.5 72B Instruct @ Novita [novita/bf16]: deranked by OpenRouter; 14h 52m so far, since 2026-08-21 00:35 UTC - [major] gpt-oss-120b @ Novita [novita/fp4]: deranked by OpenRouter; 13h 32m so far, since 2026-08-21 01:55 UTC - [major] Gemini 2.5 Pro @ Google [google-vertex/eu]: uptime fell to 13.95%; 12h 17m so far, since 2026-08-21 03:10 UTC - [major] Gemini 2.5 Pro Preview 06-05 @ Google [google-vertex/eu]: deranked by OpenRouter; 12h 12m so far, since 2026-08-21 03:15 UTC - [major] Gemini 2.5 Pro Preview 05-06 @ Google [google-vertex/eu]: deranked by OpenRouter; 12h 7m so far, since 2026-08-21 03:20 UTC - [major] Mistral Nemo @ Novita [novita/fp8]: deranked by OpenRouter; 11h 7m so far, since 2026-08-21 04:20 UTC - [major] Gemma 4 31B @ Chutes [chutes/fp4]: deranked by OpenRouter; 10h 47m so far, since 2026-08-21 04:40 UTC - [major] GPT-5.4 Nano @ OpenAI [openai/flex]: deranked by OpenRouter; 10h 37m so far, since 2026-08-21 04:50 UTC - [major] GPT-5.4 Nano @ OpenAI [openai]: deranked by OpenRouter; 10h 37m so far, since 2026-08-21 04:50 UTC - [degraded] Qwen3.5-122B-A10B @ Alibaba [alibaba]: uptime fell to 80.41%; 9h 27m so far, since 2026-08-21 06:00 UTC - [major] Qwen3 Coder 480B A35B @ Novita [novita/fp8]: deranked by OpenRouter; 8h 22m so far, since 2026-08-21 07:05 UTC - [major] DeepSeek V3.2 @ Venice [venice]: deranked by OpenRouter; 6h 12m so far, since 2026-08-21 09:15 UTC ## Resolved in the last 24 hours (100) - [degraded] GLM 5.2 @ Venice [venice/fp8]: uptime fell to 92.54%; lasted 20m, ended 2026-08-21 15:20 UTC - [degraded] Gemma 3 27B @ Parasail [parasail/fp8]: uptime fell to 94.52%; lasted 20m, ended 2026-08-21 15:20 UTC - [major] DeepSeek V4 Flash 0731 @ Decart [decart/fp4]: deranked by OpenRouter; lasted 30m, ended 2026-08-21 15:25 UTC - [degraded] R1 0528 @ StreamLake [streamlake]: uptime fell to 97.96%; lasted 20m, ended 2026-08-21 15:15 UTC - [degraded] Qwen3 235B A22B Instruct 2507 @ GMICloud [gmicloud/fp8]: uptime fell to 95.65%; lasted 30m, ended 2026-08-21 15:10 UTC - [degraded] Qwen3 VL 30B A3B Instruct @ SiliconFlow [siliconflow/fp8]: uptime fell to 91.19%; lasted 25m, ended 2026-08-21 15:05 UTC - [degraded] Qwen3 235B A22B Instruct 2507 @ StreamLake [streamlake]: uptime fell to 96.75%; lasted 25m, ended 2026-08-21 15:05 UTC - [degraded] Qwen3 235B A22B Instruct 2507 @ Parasail [parasail/fp8]: uptime fell to 85.36%; lasted 25m, ended 2026-08-21 15:05 UTC - [degraded] GPT-5.6 Luna Pro @ Azure [azure]: uptime fell to 95.83%; lasted 25m, ended 2026-08-21 15:05 UTC - [degraded] Nemotron 3.5 Content Safety (free) @ Nvidia [nvidia]: uptime fell to 87.00%; lasted 25m, ended 2026-08-21 15:05 UTC - [degraded] Kimi K3 @ Sail Research [sail-research/fp4]: uptime fell to 95.07%; lasted 25m, ended 2026-08-21 15:05 UTC - [degraded] MiniMax M3 @ DeepInfra [deepinfra/fp8]: uptime fell to 81.20%; lasted 40m, ended 2026-08-21 15:20 UTC - [degraded] Claude Opus 4.6 @ Anthropic [anthropic]: uptime fell to 97.44%; lasted 25m, ended 2026-08-21 15:05 UTC - [degraded] Qwen3.6 35B A3B @ Venice [venice/fp8]: uptime fell to 97.30%; lasted 35m, ended 2026-08-21 15:10 UTC - [degraded] GLM 5.1 @ GMICloud [gmicloud/fp8]: uptime fell to 96.77%; lasted 20m, ended 2026-08-21 14:55 UTC - [degraded] MiMo-V2.5 @ Novita [novita/fp8]: uptime fell to 93.74%; lasted 20m, ended 2026-08-21 14:55 UTC - [degraded] gpt-oss-120b @ Together [together]: uptime fell to 87.21%; lasted 35m, ended 2026-08-21 15:10 UTC - [degraded] GPT-5.6 Luna @ Azure [azure]: uptime fell to 83.02%; lasted 35m, ended 2026-08-21 15:10 UTC - [degraded] GPT-5.1 @ OpenAI [openai/default/priority]: uptime fell to 94.39%; lasted 35m, ended 2026-08-21 15:10 UTC - [degraded] GPT-5.1 @ OpenAI [openai/default/flex]: uptime fell to 94.39%; lasted 35m, ended 2026-08-21 15:10 UTC - [degraded] GPT-5.1 @ OpenAI [openai/default]: uptime fell to 94.39%; lasted 35m, ended 2026-08-21 15:10 UTC - [degraded] MiniMax M2.5 @ AtlasCloud [atlas-cloud/fp8]: uptime fell to 97.96%; lasted 25m, ended 2026-08-21 15:00 UTC - [degraded] Qwen3.6 35B A3B @ AtlasCloud [atlas-cloud/fp8]: uptime fell to 85.19%; lasted 45m, ended 2026-08-21 15:15 UTC - [degraded] GLM 5.2 @ Crusoe [crusoe/fp8]: uptime fell to 97.81%; lasted 20m, ended 2026-08-21 14:50 UTC - [major] gpt-oss-120b @ SiliconFlow [siliconflow/fp8]: deranked by OpenRouter; lasted 50m, ended 2026-08-21 15:20 UTC - [degraded] Kimi K2.5 @ Venice [venice]: uptime fell to 88.70%; lasted 30m, ended 2026-08-21 15:00 UTC - [degraded] GLM 5.1 @ Parasail [parasail/fp8]: uptime fell to 94.44%; lasted 20m, ended 2026-08-21 14:45 UTC - [degraded] Qwen3.6 35B A3B @ CoreWeave [coreweave/fp8]: uptime fell to 90.48%; lasted 40m, ended 2026-08-21 15:00 UTC - [degraded] GLM 5.2 @ BaseTen [baseten/fp8]: uptime fell to 95.95%; lasted 20m, ended 2026-08-21 14:30 UTC - [degraded] Qwen3.5-27B @ Alibaba [alibaba]: uptime fell to 34.40%; lasted 55m, ended 2026-08-21 15:05 UTC - [major] GPT-5.6 Sol @ Azure [azure/us]: uptime fell to 78.26%; lasted 30m, ended 2026-08-21 14:40 UTC - [degraded] Grok 4.6 @ xAI [xai/zdr/priority]: uptime fell to 84.40%; lasted 1h 5m, ended 2026-08-21 15:10 UTC - [degraded] Grok 4.6 @ xAI [xai/zdr]: uptime fell to 84.40%; lasted 1h 5m, ended 2026-08-21 15:10 UTC - [degraded] Kimi K3 @ Moonshot AI [moonshotai/mxfp4]: uptime fell to 96.95%; lasted 25m, ended 2026-08-21 14:30 UTC - [major] DeepSeek V4 Flash 0423 @ Baidu [baidu/fp8]: uptime fell to 73.06%; lasted 40m, ended 2026-08-21 14:45 UTC - [degraded] gpt-oss-20b @ Novita [novita/fp4]: uptime fell to 97.69%; lasted 20m, ended 2026-08-21 14:20 UTC - [degraded] DeepSeek V4 Flash 0731 @ Io Net [io-net/fp8]: uptime fell to 85.39%; lasted 35m, ended 2026-08-21 14:35 UTC - [degraded] DeepSeek V4 Flash 0423 @ Alibaba [alibaba/fp8]: uptime fell to 95.90%; lasted 20m, ended 2026-08-21 14:20 UTC - [degraded] GLM 5.1 @ AtlasCloud [atlas-cloud/fp8]: uptime fell to 85.47%; lasted 30m, ended 2026-08-21 14:25 UTC - [major] DeepSeek V4 Flash 0731 @ Decart [decart/fp4]: deranked by OpenRouter; lasted 35m, ended 2026-08-21 14:30 UTC ## Currently unhealthy endpoints (113) - Qwen2.5 7B Instruct @ Together [together]: red, uptime 0.00%, p50 —, p99 — - Qwen2.5 VL 72B Instruct @ Nebius [nebius/fp8]: red, uptime 0.00%, p50 6.25s, p99 14.4s - Mistral Small 4 @ Mistral [mistral/zdr]: red, uptime 0.00%, p50 —, p99 — - gpt-oss-120b @ SiliconFlow [siliconflow/fp8]: red, uptime 2.01%, p50 2.14s, p99 23.7s - Gemma 4 31B @ Phala [phala]: red, uptime 7.39%, p50 1.38s, p99 14.3s - DeepSeek V3.2 @ Venice [venice]: red, uptime 16.33%, p50 3.90s, p99 75.1s - Kimi K3 @ Morph [morph/fp4]: red, uptime 44.13%, p50 7.69s, p99 71.4s - GLM 4.7 @ Mancer 2 [mancer/fp4]: red, uptime 52.56%, p50 1.45s, p99 22.4s - Mistral Nemo @ Novita [novita/fp8]: red, uptime 53.97%, p50 12.3s, p99 61.0s - GLM 5 @ Amazon Bedrock [amazon-bedrock]: red, uptime 57.04%, p50 1.41s, p99 17.5s - Kimi K3 @ DeepInfra [deepinfra/bf16]: red, uptime 57.81%, p50 1.52s, p99 133s - Nova Micro 1.0 @ Amazon Bedrock [amazon-bedrock/eu-west-1]: red, uptime 59.36%, p50 411ms, p99 5.94s - gpt-oss-120b @ Novita [novita/fp4]: red, uptime 61.13%, p50 585ms, p99 13.8s - Kimi K3 @ Morph [morph/fast]: red, uptime 63.23%, p50 8.77s, p99 76.4s - GLM 5.2 @ Ambient [ambient/fp8]: red, uptime 66.67%, p50 959ms, p99 27.1s - Qwen3 32B @ SiliconFlow [siliconflow/fp8]: red, uptime 67.32%, p50 3.54s, p99 285s - GLM 4.6V @ Z.AI [z-ai/fp8]: red, uptime 73.33%, p50 2.17s, p99 187s - Gemma 4 31B @ SiliconFlow [siliconflow/fp8]: red, uptime 75.12%, p50 2.36s, p99 41.8s - GLM 5.1 @ Chutes [chutes/fp8]: red, uptime 75.44%, p50 3.34s, p99 74.5s - Gemini 2.5 Flash @ Google [google-vertex]: red, uptime 76.34%, p50 2.25s, p99 44.2s - MiMo-V2.5-Pro @ DigitalOcean [digitalocean]: red, uptime 76.68%, p50 975ms, p99 9.53s - Qwen3 235B A22B Instruct 2507 @ DeepInfra [deepinfra/fp8]: red, uptime 76.98%, p50 566ms, p99 21.0s - Qwen3 Coder 480B A35B @ Novita [novita/fp8]: red, uptime 84.05%, p50 803ms, p99 11.5s - Qwen3 235B A22B Instruct 2507 @ Venice [venice/fp8]: red, uptime 84.85%, p50 4.75s, p99 36.7s - gpt-oss-120b @ Google [google-vertex/global]: red, uptime 85.12%, p50 1.40s, p99 10.6s - Qwen3.5-122B-A10B @ Alibaba [alibaba]: red, uptime 85.62%, p50 2.46s, p99 10.5s - GLM 5.2 @ Together [together]: red, uptime 85.97%, p50 756ms, p99 13.7s - Nemotron Nano 12B 2 VL (free) @ Nvidia [nvidia]: red, uptime 86.00%, p50 4.95s, p99 78.9s - DeepSeek V3.2 Exp @ SiliconFlow [siliconflow/fp8]: red, uptime 86.30%, p50 1.72s, p99 6.97s - Gemma 4 31B @ DeepInfra [deepinfra/ultra]: red, uptime 87.14%, p50 5.12s, p99 20.2s - GLM 4.7 @ Venice [venice/fp4]: red, uptime 87.18%, p50 1.45s, p99 23.2s - Qwen3.5-9B @ SiliconFlow [siliconflow/fp8]: red, uptime 87.90%, p50 1.83s, p99 18.3s - Qwen3 235B A22B Instruct 2507 @ Nebius [nebius/fp8]: red, uptime 88.00%, p50 607ms, p99 26.6s - Llama 4 Scout @ Novita [novita/bf16]: red, uptime 88.24%, p50 4.41s, p99 8.02s - gpt-oss-20b @ SiliconFlow [siliconflow/fp8]: red, uptime 88.28%, p50 1.36s, p99 34.1s - DeepSeek V3.1 @ Mara [mara]: red, uptime 88.73%, p50 533ms, p99 11.6s - GLM 4.7 @ DeepInfra [deepinfra/fp4]: red, uptime 88.93%, p50 1.43s, p99 25.4s - MiMo-V2.5 @ GMICloud [gmicloud/fp8]: red, uptime 89.56%, p50 4.62s, p99 11.8s - Grok 4.6 @ xAI [xai]: red, uptime 89.95%, p50 1.43s, p99 83.3s - Grok 4.6 @ xAI [xai/priority]: red, uptime 89.95%, p50 4.82s, p99 60.6s - GLM 4.7 Flash @ Novita [novita/bf16]: red, uptime 93.14%, p50 1.22s, p99 108s - Nemotron 3.5 Lightning @ CoreWeave [coreweave/bf16]: red, uptime 96.97%, p50 524ms, p99 68.7s - Qwen3 Coder 30B A3B Instruct @ SiliconFlow [siliconflow/fp8]: red, uptime 100%, p50 1.33s, p99 14.9s - Qwen3.8 2.4T A95B @ DeepInfra [deepinfra/fp4]: red, uptime 100%, p50 4.16s, p99 60.0s - Llama 3.3 70B Instruct @ Nebius [nebius/fp8]: red, uptime —, p50 1.86s, p99 14.3s - MiniMax M2.7 @ DeepInfra [deepinfra/turbo]: red, uptime —, p50 7.42s, p99 66.9s - Kimi K3 @ Wafer [wafer]: red, uptime —, p50 187s, p99 287s - Kimi K3 @ Phala [phala]: red, uptime —, p50 9.38s, p99 110s - Kimi K2.7 Code @ ModelRun [modelrun/fp4]: red, uptime —, p50 386ms, p99 15.9s - Kimi K2.7 Code @ Together [together]: red, uptime —, p50 708ms, p99 10.7s - Kimi K2.6 @ Crusoe [crusoe/bf16]: red, uptime —, p50 813ms, p99 112s - Kimi K2.6 @ DigitalOcean [digitalocean]: red, uptime —, p50 2.57s, p99 51.5s - GLM 4.6 @ AtlasCloud [atlas-cloud/fp8]: red, uptime —, p50 699ms, p99 12.0s - Qwen3.6 27B @ Venice [venice/fp8]: red, uptime —, p50 2.15s, p99 48.8s - Gemma 4 31B @ Chutes [chutes/fp4]: yellow, uptime 90.09%, p50 2.21s, p99 23.2s - GPT-5.5 @ OpenAI [openai]: yellow, uptime 90.26%, p50 3.52s, p99 33.4s - GPT-5.5 @ OpenAI [openai/flex]: yellow, uptime 90.26%, p50 2.37s, p99 8.29s - GPT-5.5 @ OpenAI [openai/priority]: yellow, uptime 90.26%, p50 1.67s, p99 14.1s - DeepSeek V4 Flash 0731 @ Reka [reka/fp4]: yellow, uptime 90.27%, p50 2.33s, p99 9.00s - gpt-oss-120b @ Together [together]: yellow, uptime 90.77%, p50 371ms, p99 4.94s - Qwen3 Coder 480B A35B @ Venice [venice/fp8]: yellow, uptime 90.85%, p50 1.41s, p99 139s - GPT-5.4 Nano @ OpenAI [openai]: yellow, uptime 90.87%, p50 879ms, p99 6.15s - GPT-5.4 Nano @ OpenAI [openai/flex]: yellow, uptime 90.87%, p50 2.07s, p99 2.65s - Grok 4.6 @ xAI [xai/zdr]: yellow, uptime 91.26%, p50 2.36s, p99 123s - Grok 4.6 @ xAI [xai/zdr/priority]: yellow, uptime 91.26%, p50 3.71s, p99 82.9s - Qwen3.6 35B A3B @ Darkbloom [darkbloom/fp4]: yellow, uptime 91.38%, p50 4.86s, p99 23.3s - GLM 5.2 @ Decart [decart/fp4]: yellow, uptime 91.76%, p50 3.91s, p99 19.3s - gpt-oss-20b @ Phala [phala]: yellow, uptime 91.86%, p50 772ms, p99 9.55s - Qwen3.6 27B @ SiliconFlow [siliconflow/fp8]: yellow, uptime 92.20%, p50 8.51s, p99 38.8s - MiniMax M3 @ Morph [morph/fp4]: yellow, uptime 92.52%, p50 2.31s, p99 21.8s - GLM 4.7 @ Novita [novita/fp8]: yellow, uptime 92.59%, p50 1.85s, p99 35.8s - DeepSeek V4 Flash 0731 @ OpenInference [open-inference/fp4]: yellow, uptime 92.61%, p50 5.11s, p99 30.4s - GLM 5.2 @ Morph [morph/fp4]: yellow, uptime 92.63%, p50 1.76s, p99 23.5s - DeepSeek V3.1 @ Google [google-vertex/us-west2]: yellow, uptime 92.86%, p50 1.25s, p99 28.7s - Qwen3 30B A3B Instruct 2507 @ SiliconFlow [siliconflow/fp8]: yellow, uptime 92.95%, p50 2.15s, p99 8.72s - DeepSeek V4 Pro 0813 @ StreamLake [streamlake]: yellow, uptime 93.33%, p50 3.23s, p99 29.8s - GLM 4.7 @ Z.AI [z-ai/fp4]: yellow, uptime 93.48%, p50 2.20s, p99 25.0s - Gemma 4 31B @ Crusoe [crusoe]: yellow, uptime 93.53%, p50 1.18s, p99 27.3s - GLM 4.7 @ AtlasCloud [atlas-cloud/fp8]: yellow, uptime 93.97%, p50 1.12s, p99 12.2s - MiniMax M3 @ StreamLake [streamlake/fp8]: yellow, uptime 94.00%, p50 1.57s, p99 7.34s ## Providers with unhealthy endpoints - DeepInfra: 11 of 76 unhealthy, worst uptime 57.81% - SiliconFlow: 10 of 37 unhealthy, worst uptime 2.01% - Novita: 9 of 71 unhealthy, worst uptime 53.97% - Together: 7 of 27 unhealthy, worst uptime 0.00% - Venice: 6 of 32 unhealthy, worst uptime 16.33% - OpenAI: 5 of 138 unhealthy, worst uptime 90.26% - StreamLake: 5 of 24 unhealthy, worst uptime 93.33% - Google: 4 of 107 unhealthy, worst uptime 76.34% - Morph: 4 of 7 unhealthy, worst uptime 44.13% - Nebius: 4 of 11 unhealthy, worst uptime 0.00% - Phala: 4 of 19 unhealthy, worst uptime 7.39% - xAI: 4 of 24 unhealthy, worst uptime 89.95% - Chutes: 3 of 6 unhealthy, worst uptime 75.44% - DigitalOcean: 3 of 18 unhealthy, worst uptime 76.68% - SambaNova: 3 of 7 unhealthy, worst uptime 96.71% - Alibaba: 2 of 53 unhealthy, worst uptime 85.62% - Amazon Bedrock: 2 of 48 unhealthy, worst uptime 57.04% - AtlasCloud: 2 of 28 unhealthy, worst uptime 93.97% - CoreWeave: 2 of 20 unhealthy, worst uptime 96.97% - Crusoe: 2 of 7 unhealthy, worst uptime 93.53% - Fireworks: 2 of 12 unhealthy, worst uptime 94.87% - GMICloud: 2 of 18 unhealthy, worst uptime 89.56% - Io Net: 2 of 4 unhealthy, worst uptime 100% - Z.AI: 2 of 12 unhealthy, worst uptime 73.33% - Ambient: 1 of 3 unhealthy, worst uptime 66.67% - Azure: 1 of 70 unhealthy, worst uptime 95.85% - Baidu: 1 of 9 unhealthy, worst uptime 99.43% - Darkbloom: 1 of 3 unhealthy, worst uptime 91.38% - Decart: 1 of 4 unhealthy, worst uptime 91.76% - Mancer 2: 1 of 8 unhealthy, worst uptime 52.56% - Mara: 1 of 3 unhealthy, worst uptime 88.73% - Mistral: 1 of 16 unhealthy, worst uptime 0.00% - ModelRun: 1 of 3 unhealthy, worst uptime 100% - Nvidia: 1 of 8 unhealthy, worst uptime 86.00% - OpenInference: 1 of 2 unhealthy, worst uptime 92.61% - Reka: 1 of 4 unhealthy, worst uptime 90.27% - Wafer: 1 of 3 unhealthy, worst uptime 99.95%