{ "model_id": "m6", "model_tag": "nemotron-3.5-lightning:30b-a3b", "box_class": "cpu-mini", "started_utc": "2026-08-25T14:42:47Z", "nproc": 16, "num_predict": 256, "prompt_sha256": "90eedd0c53f9554ae3837674504fcb7090013d9a432a653352183c0f25a7ce5c", "nvidia_before": "no nvidia-smi (no GPU on this box)", "ollama_version": "0.32.15", "runs": [ { "ttft_ms": 170.73, "first_token_was_thinking": false, "wall_s": 17.6689, "eval_count": 256, "eval_duration_ns": 17496760000, "decode_tok_s": 14.631, "prompt_eval_count": 532, "prompt_eval_duration_ns": 166900000, "prefill_tok_s": 3187.537, "load_duration_ns": 792578, "total_duration_ns": 17667059090, "streamed_chunks": 245, "response_chars": 1029, "thinking_chars": 0, "done_reason": "length", "run": 1 }, { "ttft_ms": 183.42, "first_token_was_thinking": false, "wall_s": 17.9467, "eval_count": 256, "eval_duration_ns": 17761648000, "decode_tok_s": 14.413, "prompt_eval_count": 532, "prompt_eval_duration_ns": 179837000, "prefill_tok_s": 2958.234, "load_duration_ns": 859770, "total_duration_ns": 17944807263, "streamed_chunks": 245, "response_chars": 1029, "thinking_chars": 0, "done_reason": "length", "run": 2 }, { "ttft_ms": 183.01, "first_token_was_thinking": false, "wall_s": 17.6825, "eval_count": 256, "eval_duration_ns": 17497770000, "decode_tok_s": 14.63, "prompt_eval_count": 532, "prompt_eval_duration_ns": 179238000, "prefill_tok_s": 2968.121, "load_duration_ns": 912839, "total_duration_ns": 17680412262, "streamed_chunks": 245, "response_chars": 1029, "thinking_chars": 0, "done_reason": "length", "run": 3 }, { "ttft_ms": 184.65, "first_token_was_thinking": false, "wall_s": 18.5615, "eval_count": 256, "eval_duration_ns": 18375634000, "decode_tok_s": 13.931, "prompt_eval_count": 532, "prompt_eval_duration_ns": 180790000, "prefill_tok_s": 2942.641, "load_duration_ns": 816717, "total_duration_ns": 18559637866, "streamed_chunks": 245, "response_chars": 1029, "thinking_chars": 0, "done_reason": "length", "run": 4 }, { "ttft_ms": 185.28, "first_token_was_thinking": false, "wall_s": 17.7165, "eval_count": 256, "eval_duration_ns": 17528753000, "decode_tok_s": 14.605, "prompt_eval_count": 532, "prompt_eval_duration_ns": 181840000, "prefill_tok_s": 2925.649, "load_duration_ns": 800765, "total_duration_ns": 17713689677, "streamed_chunks": 245, "response_chars": 1029, "thinking_chars": 0, "done_reason": "length", "run": 5 } ], "errors": [], "quantization": "Q4_K_M", "param_size": "32.9B", "family": "nemotron_h_moe", "context_length": 1048576, "think_mode": false, "think_note": "think:false accepted", "warmup_discarded": { "ttft_ms": 182.11, "first_token_was_thinking": false, "wall_s": 17.1864, "eval_count": 256, "eval_duration_ns": 17002651999, "decode_tok_s": 15.056, "prompt_eval_count": 532, "prompt_eval_duration_ns": 178354000, "prefill_tok_s": 2982.832, "load_duration_ns": 814486, "total_duration_ns": 17184303273, "streamed_chunks": 245, "response_chars": 1029, "thinking_chars": 0, "done_reason": "length" }, "ps_after_load": [ { "name": "nemotron-3.5-lightning:30b-a3b", "model": "nemotron-3.5-lightning:30b-a3b", "size": 25609340188, "digest": "e7a64ff15fb174c42b4f463e5c888c4f2c7b9cabf9e8d65a1c0874405426c1b2", "details": { "parent_model": "", "format": "gguf", "family": "nemotron_h_moe", "families": [ "nemotron_h_moe" ], "parameter_size": "32.9B", "quantization_level": "Q4_K_M" }, "expires_at": "2026-08-25T14:53:29.214945182Z", "size_vram": 0, "context_length": 4096 } ], "size_vram": 0, "size_total": 25609340188, "cpu_only_receipt": true, "peak_runner_rss_kb": 25340184, "peak_runner_rss_gib": 24.17, "peak_runner_threads": 26, "rss_samples": 176, "derived": { "decode_tok_s": { "min": 13.931, "median": 14.605, "max": 14.631, "mean": 14.442, "n": 5 }, "ttft_ms": { "min": 170.73, "median": 183.42, "max": 185.28, "mean": 181.418, "n": 5 }, "wall_s": { "min": 17.6689, "median": 17.7165, "max": 18.5615, "mean": 17.915, "n": 5 }, "prefill_tok_s": { "min": 2925.649, "median": 2958.234, "max": 3187.537, "mean": 2996.436, "n": 5 }, "eval_count": { "min": 256, "median": 256, "max": 256, "mean": 256.0, "n": 5 } }, "cold_load": [ { "ttft_ms": 26433.68, "first_token_was_thinking": false, "wall_s": 26.4352, "eval_count": 1, "eval_duration_ns": 1000, "decode_tok_s": 1000000.0, "prompt_eval_count": 532, "prompt_eval_duration_ns": 14825830000, "prefill_tok_s": 35.883, "load_duration_ns": 11604416070, "total_duration_ns": 26433371220, "streamed_chunks": 1, "response_chars": 4, "thinking_chars": 0, "done_reason": "length", "wall_to_first_token_s": 26.4352, "ps_confirmed_empty_before": true, "load_duration_s": 11.6044 }, { "ttft_ms": 24422.43, "first_token_was_thinking": false, "wall_s": 24.4236, "eval_count": 1, "eval_duration_ns": 1000, "decode_tok_s": 1000000.0, "prompt_eval_count": 532, "prompt_eval_duration_ns": 13317294000, "prefill_tok_s": 39.948, "load_duration_ns": 11101553664, "total_duration_ns": 24421949987, "streamed_chunks": 1, "response_chars": 4, "thinking_chars": 0, "done_reason": "length", "wall_to_first_token_s": 24.4236, "ps_confirmed_empty_before": true, "load_duration_s": 11.1016 } ], "cold_load_note": "PAGE-CACHE-WARM: no sudo to drop caches, so this is warm-disk reload, not a true cold read from NVMe.", "nvidia_after": "no nvidia-smi (no GPU on this box)", "ps_final": [ { "name": "nemotron-3.5-lightning:30b-a3b", "model": "nemotron-3.5-lightning:30b-a3b", "size": 25609340188, "digest": "e7a64ff15fb174c42b4f463e5c888c4f2c7b9cabf9e8d65a1c0874405426c1b2", "details": { "parent_model": "", "format": "gguf", "family": "nemotron_h_moe", "families": [ "nemotron_h_moe" ], "parameter_size": "32.9B", "quantization_level": "Q4_K_M" }, "expires_at": "2026-08-25T14:55:50.902843972Z", "size_vram": 0, "context_length": 4096 } ], "finished_utc": "2026-08-25T14:45:51Z" }