{ "model_id": "m5", "model_tag": "gemma4:26b", "box_class": "cpu-laptop", "started_utc": "2026-08-25T17:02:26Z", "nproc": 24, "num_predict": 256, "prompt_sha256": "90eedd0c53f9554ae3837674504fcb7090013d9a432a653352183c0f25a7ce5c", "nvidia_before": "0, 15 MiB, 0 %", "ollama_version": "0.32.15", "runs": [ { "ttft_ms": 342.6, "first_token_was_thinking": false, "wall_s": 19.6618, "eval_count": 256, "eval_duration_ns": 19314096000, "decode_tok_s": 13.255, "prompt_eval_count": 536, "prompt_eval_duration_ns": 337604000, "prefill_tok_s": 1587.659, "load_duration_ns": 719342, "total_duration_ns": 19656493291, "streamed_chunks": 256, "response_chars": 1064, "thinking_chars": 0, "done_reason": "length", "run": 1 }, { "ttft_ms": 346.87, "first_token_was_thinking": false, "wall_s": 19.6583, "eval_count": 256, "eval_duration_ns": 19308490000, "decode_tok_s": 13.258, "prompt_eval_count": 536, "prompt_eval_duration_ns": 341468000, "prefill_tok_s": 1569.693, "load_duration_ns": 629064, "total_duration_ns": 19655049545, "streamed_chunks": 256, "response_chars": 1064, "thinking_chars": 0, "done_reason": "length", "run": 2 }, { "ttft_ms": 345.37, "first_token_was_thinking": false, "wall_s": 19.6641, "eval_count": 256, "eval_duration_ns": 19315765000, "decode_tok_s": 13.253, "prompt_eval_count": 536, "prompt_eval_duration_ns": 340644000, "prefill_tok_s": 1573.49, "load_duration_ns": 616282, "total_duration_ns": 19660743673, "streamed_chunks": 256, "response_chars": 1064, "thinking_chars": 0, "done_reason": "length", "run": 3 }, { "ttft_ms": 348.13, "first_token_was_thinking": false, "wall_s": 19.6512, "eval_count": 256, "eval_duration_ns": 19298616000, "decode_tok_s": 13.265, "prompt_eval_count": 536, "prompt_eval_duration_ns": 343331000, "prefill_tok_s": 1561.176, "load_duration_ns": 478771, "total_duration_ns": 19646097886, "streamed_chunks": 256, "response_chars": 1064, "thinking_chars": 0, "done_reason": "length", "run": 4 }, { "ttft_ms": 352.3, "first_token_was_thinking": false, "wall_s": 19.5802, "eval_count": 256, "eval_duration_ns": 19222730000, "decode_tok_s": 13.318, "prompt_eval_count": 536, "prompt_eval_duration_ns": 346565000, "prefill_tok_s": 1546.607, "load_duration_ns": 801182, "total_duration_ns": 19574910747, "streamed_chunks": 256, "response_chars": 1064, "thinking_chars": 0, "done_reason": "length", "run": 5 } ], "errors": [], "quantization": "Q4_K_M", "param_size": "25.2B", "family": "gemma4", "context_length": 262144, "think_mode": false, "think_note": "think:false accepted", "warmup_discarded": { "ttft_ms": 346.06, "first_token_was_thinking": false, "wall_s": 19.6432, "eval_count": 256, "eval_duration_ns": 19294225000, "decode_tok_s": 13.268, "prompt_eval_count": 536, "prompt_eval_duration_ns": 340029000, "prefill_tok_s": 1576.336, "load_duration_ns": 867442, "total_duration_ns": 19639676216, "streamed_chunks": 256, "response_chars": 1064, "thinking_chars": 0, "done_reason": "length" }, "ps_after_load": [ { "name": "gemma4:26b", "model": "gemma4:26b", "size": 9535058083, "digest": "08ae7ec1744bd7f451c4a530afb39d2673ad9d07a8369b8a33a3613b41212a68", "details": { "parent_model": "", "format": "gguf", "family": "gemma4", "families": [ "gemma4" ], "parameter_size": "25.2B", "quantization_level": "Q4_K_M" }, "expires_at": "2026-08-25T17:13:05.03262281Z", "size_vram": 0, "context_length": 4096 } ], "size_vram": 0, "size_total": 9535058083, "cpu_only_receipt": true, "peak_runner_rss_kb": 19900824, "peak_runner_rss_gib": 18.98, "peak_runner_threads": 34, "rss_samples": 192, "loadavg_during_runs": { "min": 1.98, "median": 3.45, "max": 3.84, "mean": 3.28, "samples": 192, "nproc": 24 }, "contention_suspected": false, "contention_note": "loadavg consistent with this bench owning the machine", "foreign_gpu_procs_during": "0, 15 MiB, 0 %", "derived": { "decode_tok_s": { "min": 13.253, "median": 13.258, "max": 13.318, "mean": 13.27, "n": 5 }, "ttft_ms": { "min": 342.6, "median": 346.87, "max": 352.3, "mean": 347.054, "n": 5 }, "wall_s": { "min": 19.5802, "median": 19.6583, "max": 19.6641, "mean": 19.643, "n": 5 }, "prefill_tok_s": { "min": 1546.607, "median": 1569.693, "max": 1587.659, "mean": 1567.725, "n": 5 }, "eval_count": { "min": 256, "median": 256, "max": 256, "mean": 256.0, "n": 5 } }, "cold_load": [ { "ttft_ms": 18360.06, "first_token_was_thinking": false, "wall_s": 18.3623, "eval_count": 1, "eval_duration_ns": 1000, "decode_tok_s": 1000000.0, "prompt_eval_count": 536, "prompt_eval_duration_ns": 8474291999, "prefill_tok_s": 63.25, "load_duration_ns": 9880384064, "total_duration_ns": 18359674914, "streamed_chunks": 1, "response_chars": 2, "thinking_chars": 0, "done_reason": "length", "wall_to_first_token_s": 18.3623, "ps_confirmed_empty_before": true, "load_duration_s": 9.8804 }, { "ttft_ms": 18693.16, "first_token_was_thinking": false, "wall_s": 18.6966, "eval_count": 1, "eval_duration_ns": 1000, "decode_tok_s": 1000000.0, "prompt_eval_count": 536, "prompt_eval_duration_ns": 8540540000, "prefill_tok_s": 62.759, "load_duration_ns": 10147020984, "total_duration_ns": 18692581485, "streamed_chunks": 1, "response_chars": 2, "thinking_chars": 0, "done_reason": "length", "wall_to_first_token_s": 18.6967, "ps_confirmed_empty_before": true, "load_duration_s": 10.147 } ], "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": "0, 15 MiB, 0 %", "ps_final": [ { "name": "gemma4:26b", "model": "gemma4:26b", "size": 9535058083, "digest": "08ae7ec1744bd7f451c4a530afb39d2673ad9d07a8369b8a33a3613b41212a68", "details": { "parent_model": "", "format": "gguf", "family": "gemma4", "families": [ "gemma4" ], "parameter_size": "25.2B", "quantization_level": "Q4_K_M" }, "expires_at": "2026-08-25T17:15:21.223862752Z", "size_vram": 0, "context_length": 4096 } ], "finished_utc": "2026-08-25T17:05:21Z" }