BlockRun
blockrun.ai · Inference · 1 record · JSON
BlockRun Inference
caveatanonymousno cardMrequest-rate
| Free tier | 7 of 117 model ids on /v1/models carry an input price of 0; the rest carry list prices spanning openai, google, anthropic, xai, zai, bytedance, deepseek, qwen, nvidia and eleven other families unit tokens; metered on request-count; the seven zero-priced ids are nvidia/nemotron-3-nano-omni-30b-a3b-reasoning, nvidia/nemotron-3.5-lightning, nvidia/llama-3.2-11b-vision, nvidia/nemotron-3-ultra-550b, cohere/north-mini-code, poolside/laguna-xs-2.1 and nvidia/muse-glimmer-30b |
|---|---|
| Price | none stated |
| Per-object cap | not applicable: an inference endpoint stores nothing for the caller |
| Retention | not applicable |
| Egress | the response is the payload; no separate egress term is published |
| Identity | anonymous no Authorization header sent and none required for the free route |
| Limits |
|
| Binding | request-rate the upstream NVIDIA 429 binds, and it arrives inside a 200-style error body rather than as a 429 status, so a client cannot see it by status code |
| Interface | rest |
| Tooling | any OpenAI-compatible client; curl against POST {base}/chat/completions |
| Composes with | pairs with a free-standing agent as a relay of last resort; the scarce link is BlockRun's own upstream key, because its 429 arrives as FREE_MODEL_FAILED inside a 200 |
| Reliability | 5 sequential identical calls answered 2 and refused 3 with FREE_MODEL_FAILED and a debug field naming 'NVIDIA API error: 429 Too Many Requests' |
| Notes | The relay category's clearest instance. An error body that names a different vendor's API is the cheapest available tell, and the echoed model string is the second. A caller that does not compare the echoed model receives a plausible answer from a model it did not choose. |
| Evidence rung | measured the relay is proved by its own error body naming NVIDIA, and the substitution by the echoed model string. Both are measured; the upstream arrangement is inferred from them. |
| Checked | 2026-10-23 (live); stale on 2026-11-22 |
Evidence (3)
| date | source | what it proved |
|---|---|---|
| 2026-10-22 | probe:https://blockrun.ai/api/v1/chat/completions | nvidia/gpt-oss-20b answered 200 and echoed nvidia/gpt-oss-20b's two runs, 2091 ms and 2543 ms, 51 completion tokens each |
| 2026-10-22 | probe:https://blockrun.ai/api/v1/chat/completions | in the recon sweep eight minutes earlier, one of five sequential calls echoed nvidia/nemotron-3-super-120b for a nvidia/gpt-oss-20b request, and three returned FREE_MODEL_FAILED with debug 'NVIDIA API error: 429 Too Many Requests' |
| 2026-10-22 | blockrun.ai/api/v1/models | 117 model ids across 13 or more vendor families including openai, anthropic, google, xai and elevenlabs, with per-model long_context price tiers |
Unknown or withheld values (2)
$.probe.records[0].tps_basisprocess$.retention.statednot-applicable